Overcome Failing Document Ingestion & RAG Strategies with Agentic Knowledge Distillation

Stay Ahead, Stay ONMINE

Overcome Failing Document Ingestion & RAG Strategies with Agentic Knowledge Distillation

Introduction Many generative AI use cases still revolve around Retrieval Augmented Generation (RAG), yet consistently fall short of user expectations. Despite the growing body of research on RAG improvements and even adding Agents into the process, many solutions still fail to return exhaustive results, miss information that is critical but infrequently mentioned in the documents, require multiple search iterations, and generally struggle to reconcile key themes across multiple documents. To top it all off, many implementations still rely on cramming as much “relevant” information as possible into the model’s context window alongside detailed system and user prompts. Reconciling all this information often exceeds the model’s cognitive capacity and compromises response quality and consistency. This is where our Agentic Knowledge Distillation + Pyramid Search Approach comes into play. Instead of chasing the best chunking strategy, retrieval algorithm, or inference-time reasoning method, my team, Jim Brown, Mason Sawtell, Sandi Besen, and I, take an agentic approach to document ingestion. We leverage the full capability of the model at ingestion time to focus exclusively on distilling and preserving the most meaningful information from the document dataset. This fundamentally simplifies the RAG process by allowing the model to direct its reasoning abilities toward addressing the user/system instructions rather than struggling to understand formatting and disparate information across document chunks. We specifically target high-value questions that are often difficult to evaluate because they have multiple correct answers or solution paths. These cases are where traditional RAG solutions struggle most and existing RAG evaluation datasets are largely insufficient for testing this problem space. For our research implementation, we downloaded annual and quarterly reports from the last year for the 30 companies in the DOW Jones Industrial Average. These documents can be found through the SEC EDGAR website. The information on EDGAR is accessible and able to be downloaded for free or can be queried through EDGAR public searches. See the SEC privacy policy for additional details, information on the SEC website is “considered public information and may be copied or further distributed by users of the web site without the SEC’s permission”. We selected this dataset for two key reasons: first, it falls outside the knowledge cutoff for the models evaluated, ensuring that the models cannot respond to questions based on their knowledge from pre-training; second, it’s a close approximation for real-world business problems while allowing us to discuss and share our findings using publicly available data. While typical RAG solutions excel at factual retrieval where the answer is easily identified in the document dataset (e.g., “When did Apple’s annual shareholder’s meeting occur?”), they struggle with nuanced questions that require a deeper understanding of concepts across documents (e.g., “Which of the DOW companies has the most promising AI strategy?”). Our Agentic Knowledge Distillation + Pyramid Search Approach addresses these types of questions with much greater success compared to other standard approaches we tested and overcomes limitations associated with using knowledge graphs in RAG systems. In this article, we’ll cover how our knowledge distillation process works, key benefits of this approach, examples, and an open discussion on the best way to evaluate these types of systems where, in many cases, there is no singular “right” answer. Building the pyramid: How Agentic Knowledge Distillation works Image by author and team depicting pyramid structure for document ingestion. Robots meant to represent agents building the pyramid. Overview Our knowledge distillation process creates a multi-tiered pyramid of information from the raw source documents. Our approach is inspired by the pyramids used in deep learning computer vision-based tasks, which allow a model to analyze an image at multiple scales. We take the contents of the raw document, convert it to markdown, and distill the content into a list of atomic insights, related concepts, document abstracts, and general recollections/memories. During retrieval it’s possible to access any or all levels of the pyramid to respond to the user request. How to distill documents and build the pyramid: Convert documents to Markdown: Convert all raw source documents to Markdown. We’ve found models process markdown best for this task compared to other formats like JSON and it is more token efficient. We used Azure Document Intelligence to generate the markdown for each page of the document, but there are many other open-source libraries like MarkItDown which do the same thing. Our dataset included 331 documents and 16,601 pages. Extract atomic insights from each page: We process documents using a two-page sliding window, which allows each page to be analyzed twice. This gives the agent the opportunity to correct any potential mistakes when processing the page initially. We instruct the model to create a numbered list of insights that grows as it processes the pages in the document. The agent can overwrite insights from the previous page if they were incorrect since it sees each page twice. We instruct the model to extract insights in simple sentences following the subject-verb-object (SVO) format and to write sentences as if English is the second language of the user. This significantly improves performance by encouraging clarity and precision. Rolling over each page multiple times and using the SVO format also solves the disambiguation problem, which is a huge challenge for knowledge graphs. The insight generation step is also particularly helpful for extracting information from tables since the model captures the facts from the table in clear, succinct sentences. Our dataset produced 216,931 total insights, about 13 insights per page and 655 insights per document. Distilling concepts from insights: From the detailed list of insights, we identify higher-level concepts that connect related information about the document. This step significantly reduces noise and redundant information in the document while preserving essential information and themes. Our dataset produced 14,824 total concepts, about 1 concept per page and 45 concepts per document. Creating abstracts from concepts: Given the insights and concepts in the document, the LLM writes an abstract that appears both better than any abstract a human would write and more information-dense than any abstract present in the original document. The LLM generated abstract provides incredibly comprehensive knowledge about the document with a small token density that carries a significant amount of information. We produce one abstract per document, 331 total. Storing recollections/memories across documents: At the top of the pyramid we store critical information that is useful across all tasks. This can be information that the user shares about the task or information the agent learns about the dataset over time by researching and responding to tasks. For example, we can store the current 30 companies in the DOW as a recollection since this list is different from the 30 companies in the DOW at the time of the model’s knowledge cutoff. As we conduct more and more research tasks, we can continuously improve our recollections and maintain an audit trail of which documents these recollections originated from. For example, we can keep track of AI strategies across companies, where companies are making major investments, etc. These high-level connections are super important since they reveal relationships and information that are not apparent in a single page or document. Sample subset of insights extracted from IBM 10Q, Q3 2024 (page 4) We store the text and embeddings for each layer of the pyramid (pages and up) in Azure PostgreSQL. We originally used Azure AI Search, but switched to PostgreSQL for cost reasons. This required us to write our own hybrid search function since PostgreSQL doesn’t yet natively support this feature. This implementation would work with any vector database or vector index of your choosing. The key requirement is to store and efficiently retrieve both text and vector embeddings at any level of the pyramid. This approach essentially creates the essence of a knowledge graph, but stores information in natural language, the way an LLM natively wants to interact with it, and is more efficient on token retrieval. We also let the LLM pick the terms used to categorize each level of the pyramid, this seemed to let the model decide for itself the best way to describe and differentiate between the information stored at each level. For example, the LLM preferred “insights” to “facts” as the label for the first level of distilled knowledge. Our goal in doing this was to better understand how an LLM thinks about the process by letting it decide how to store and group related information. Using the pyramid: How it works with RAG & Agents At inference time, both traditional RAG and agentic approaches benefit from the pre-processed, distilled information ingested in our knowledge pyramid. The pyramid structure allows for efficient retrieval in both the traditional RAG case, where only the top X related pieces of information are retrieved or in the Agentic case, where the Agent iteratively plans, retrieves, and evaluates information before returning a final response. The benefit of the pyramid approach is that information at any and all levels of the pyramid can be used during inference. For our implementation, we used PydanticAI to create a search agent that takes in the user request, generates search terms, explores ideas related to the request, and keeps track of information relevant to the request. Once the search agent determines there’s sufficient information to address the user request, the results are re-ranked and sent back to the LLM to generate a final reply. Our implementation allows a search agent to traverse the information in the pyramid as it gathers details about a concept/search term. This is similar to walking a knowledge graph, but in a way that’s more natural for the LLM since all the information in the pyramid is stored in natural language. Depending on the use case, the Agent could access information at all levels of the pyramid or only at specific levels (e.g. only retrieve information from the concepts). For our experiments, we did not retrieve raw page-level data since we wanted to focus on token efficiency and found the LLM-generated information for the insights, concepts, abstracts, and recollections was sufficient for completing our tasks. In theory, the Agent could also have access to the page data; this would provide additional opportunities for the agent to re-examine the original document text; however, it would also significantly increase the total tokens used. Here is a high-level visualization of our Agentic approach to responding to user requests: Image created by author and team providing an overview of the agentic research & response process Results from the pyramid: Real-world examples To evaluate the effectiveness of our approach, we tested it against a variety of question categories, including typical fact-finding questions and complex cross-document research and analysis tasks. Fact-finding (spear fishing): These tasks require identifying specific information or facts that are buried in a document. These are the types of questions typical RAG solutions target but often require many searches and consume lots of tokens to answer correctly. Example task: “What was IBM’s total revenue in the latest financial reporting?” Example response using pyramid approach: “IBM’s total revenue for the third quarter of 2024 was $14.968 billion [ibm-10q-q3-2024.pdf, pg. 4] Total tokens used to research and generate response This result is correct (human-validated) and was generated using only 9,994 total tokens, with 1,240 tokens in the generated final response. Complex research and analysis: These tasks involve researching and understanding multiple concepts to gain a broader understanding of the documents and make inferences and informed assumptions based on the gathered facts. Example task: “Analyze the investments Microsoft and NVIDIA are making in AI and how they are positioning themselves in the market. The report should be clearly formatted.” Example response: Response generated by the agent analyzing AI investments and positioning for Microsoft and NVIDIA. The result is a comprehensive report that executed quickly and contains detailed information about each of the companies. 26,802 total tokens were used to research and respond to the request with a significant percentage of them used for the final response (2,893 tokens or ~11%). These results were also reviewed by a human to verify their validity. Snippet indicating total token usage for the task Example task: “Create a report on analyzing the risks disclosed by the various financial companies in the DOW. Indicate which risks are shared and unique.” Example response: Part 1 of response generated by the agent on disclosed risks. Part 2 of response generated by the agent on disclosed risks. Similarly, this task was completed in 42.7 seconds and used 31,685 total tokens, with 3,116 tokens used to generate the final report. Snippet indicating total token usage for the task These results for both fact-finding and complex analysis tasks demonstrate that the pyramid approach efficiently creates detailed reports with low latency using a minimal amount of tokens. The tokens used for the tasks carry dense meaning with little noise allowing for high-quality, thorough responses across tasks. Benefits of the pyramid: Why use it? Overall, we found that our pyramid approach provided a significant boost in response quality and overall performance for high-value questions. Some of the key benefits we observed include: Reduced model’s cognitive load: When the agent receives the user task, it retrieves pre-processed, distilled information rather than the raw, inconsistently formatted, disparate document chunks. This fundamentally improves the retrieval process since the model doesn’t waste its cognitive capacity on trying to break down the page/chunk text for the first time. Superior table processing: By breaking down table information and storing it in concise but descriptive sentences, the pyramid approach makes it easier to retrieve relevant information at inference time through natural language queries. This was particularly important for our dataset since financial reports contain lots of critical information in tables. Improved response quality to many types of requests: The pyramid enables more comprehensive context-aware responses to both precise, fact-finding questions and broad analysis based tasks that involve many themes across numerous documents. Preservation of critical context: Since the distillation process identifies and keeps track of key facts, important information that might appear only once in the document is easier to maintain. For example, noting that all tables are represented in millions of dollars or in a particular currency. Traditional chunking methods often cause this type of information to slip through the cracks. Optimized token usage, memory, and speed: By distilling information at ingestion time, we significantly reduce the number of tokens required during inference, are able to maximize the value of information put in the context window, and improve memory use. Scalability: Many solutions struggle to perform as the size of the document dataset grows. This approach provides a much more efficient way to manage a large volume of text by only preserving critical information. This also allows for a more efficient use of the LLMs context window by only sending it useful, clear information. Efficient concept exploration: The pyramid enables the agent to explore related information similar to navigating a knowledge graph, but does not require ever generating or maintaining relationships in the graph. The agent can use natural language exclusively and keep track of important facts related to the concepts it’s exploring in a highly token-efficient and fluid way. Emergent dataset understanding: An unexpected benefit of this approach emerged during our testing. When asking questions like “what can you tell me about this dataset?” or “what types of questions can I ask?”, the system is able to respond and suggest productive search topics because it has a more robust understanding of the dataset context by accessing higher levels in the pyramid like the abstracts and recollections. Beyond the pyramid: Evaluation challenges & future directions Challenges While the results we’ve observed when using the pyramid search approach have been nothing short of amazing, finding ways to establish meaningful metrics to evaluate the entire system both at ingestion time and during information retrieval is challenging. Traditional RAG and Agent evaluation frameworks often fail to address nuanced questions and analytical responses where many different responses are valid. Our team plans to write a research paper on this approach in the future, and we are open to any thoughts and feedback from the community, especially when it comes to evaluation metrics. Many of the existing datasets we found were focused on evaluating RAG use cases within one document or precise information retrieval across multiple documents rather than robust concept and theme analysis across documents and domains. The main use cases we are interested in relate to broader questions that are representative of how businesses actually want to interact with GenAI systems. For example, “tell me everything I need to know about customer X” or “how do the behaviors of Customer A and B differ? Which am I more likely to have a successful meeting with?”. These types of questions require a deep understanding of information across many sources. The answers to these questions typically require a person to synthesize data from multiple areas of the business and think critically about it. As a result, the answers to these questions are rarely written or saved anywhere which makes it impossible to simply store and retrieve them through a vector index in a typical RAG process. Another consideration is that many real-world use cases involve dynamic datasets where documents are consistently being added, edited, and deleted. This makes it difficult to evaluate and track what a “correct” response is since the answer will evolve as the available information changes. Future directions In the future, we believe that the pyramid approach can address some of these challenges by enabling more effective processing of dense documents and storing learned information as recollections. However, tracking and evaluating the validity of the recollections over time will be critical to the system’s overall success and remains a key focus area for our ongoing work. When applying this approach to organizational data, the pyramid process could also be used to identify and assess discrepancies across areas of the business. For example, uploading all of a company’s sales pitch decks could surface where certain products or services are being positioned inconsistently. It could also be used to compare insights extracted from various line of business data to help understand if and where teams have developed conflicting understandings of topics or different priorities. This application goes beyond pure information retrieval use cases and would allow the pyramid to serve as an organizational alignment tool that helps identify divergences in messaging, terminology, and overall communication. Conclusion: Key takeaways and why the pyramid approach matters The knowledge distillation pyramid approach is significant because it leverages the full power of the LLM at both ingestion and retrieval time. Our approach allows you to store dense information in fewer tokens which has the added benefit of reducing noise in the dataset at inference. Our approach also runs very quickly and is incredibly token efficient, we are able to generate responses within seconds, explore potentially hundreds of searches, and on average use

Introduction

Many generative AI use cases still revolve around Retrieval Augmented Generation (RAG), yet consistently fall short of user expectations. Despite the growing body of research on RAG improvements and even adding Agents into the process, many solutions still fail to return exhaustive results, miss information that is critical but infrequently mentioned in the documents, require multiple search iterations, and generally struggle to reconcile key themes across multiple documents. To top it all off, many implementations still rely on cramming as much “relevant” information as possible into the model’s context window alongside detailed system and user prompts. Reconciling all this information often exceeds the model’s cognitive capacity and compromises response quality and consistency.

This is where our Agentic Knowledge Distillation + Pyramid Search Approach comes into play. Instead of chasing the best chunking strategy, retrieval algorithm, or inference-time reasoning method, my team, Jim Brown, Mason Sawtell, Sandi Besen, and I, take an agentic approach to document ingestion.

We leverage the full capability of the model at ingestion time to focus exclusively on distilling and preserving the most meaningful information from the document dataset. This fundamentally simplifies the RAG process by allowing the model to direct its reasoning abilities toward addressing the user/system instructions rather than struggling to understand formatting and disparate information across document chunks.

We specifically target high-value questions that are often difficult to evaluate because they have multiple correct answers or solution paths. These cases are where traditional RAG solutions struggle most and existing RAG evaluation datasets are largely insufficient for testing this problem space. For our research implementation, we downloaded annual and quarterly reports from the last year for the 30 companies in the DOW Jones Industrial Average. These documents can be found through the SEC EDGAR website. The information on EDGAR is accessible and able to be downloaded for free or can be queried through EDGAR public searches. See the SEC privacy policy for additional details, information on the SEC website is “considered public information and may be copied or further distributed by users of the web site without the SEC’s permission”. We selected this dataset for two key reasons: first, it falls outside the knowledge cutoff for the models evaluated, ensuring that the models cannot respond to questions based on their knowledge from pre-training; second, it’s a close approximation for real-world business problems while allowing us to discuss and share our findings using publicly available data.

While typical RAG solutions excel at factual retrieval where the answer is easily identified in the document dataset (e.g., “When did Apple’s annual shareholder’s meeting occur?”), they struggle with nuanced questions that require a deeper understanding of concepts across documents (e.g., “Which of the DOW companies has the most promising AI strategy?”). Our Agentic Knowledge Distillation + Pyramid Search Approach addresses these types of questions with much greater success compared to other standard approaches we tested and overcomes limitations associated with using knowledge graphs in RAG systems.

In this article, we’ll cover how our knowledge distillation process works, key benefits of this approach, examples, and an open discussion on the best way to evaluate these types of systems where, in many cases, there is no singular “right” answer.

Building the pyramid: How Agentic Knowledge Distillation works

AI-generated image showing a pyramid structure for document ingestion with labelled sections. — Image by author and team depicting pyramid structure for document ingestion. Robots meant to represent agents building the pyramid.

Overview

Our knowledge distillation process creates a multi-tiered pyramid of information from the raw source documents. Our approach is inspired by the pyramids used in deep learning computer vision-based tasks, which allow a model to analyze an image at multiple scales. We take the contents of the raw document, convert it to markdown, and distill the content into a list of atomic insights, related concepts, document abstracts, and general recollections/memories. During retrieval it’s possible to access any or all levels of the pyramid to respond to the user request.

How to distill documents and build the pyramid:

Convert documents to Markdown: Convert all raw source documents to Markdown. We’ve found models process markdown best for this task compared to other formats like JSON and it is more token efficient. We used Azure Document Intelligence to generate the markdown for each page of the document, but there are many other open-source libraries like MarkItDown which do the same thing. Our dataset included 331 documents and 16,601 pages.
Extract atomic insights from each page: We process documents using a two-page sliding window, which allows each page to be analyzed twice. This gives the agent the opportunity to correct any potential mistakes when processing the page initially. We instruct the model to create a numbered list of insights that grows as it processes the pages in the document. The agent can overwrite insights from the previous page if they were incorrect since it sees each page twice. We instruct the model to extract insights in simple sentences following the subject-verb-object (SVO) format and to write sentences as if English is the second language of the user. This significantly improves performance by encouraging clarity and precision. Rolling over each page multiple times and using the SVO format also solves the disambiguation problem, which is a huge challenge for knowledge graphs. The insight generation step is also particularly helpful for extracting information from tables since the model captures the facts from the table in clear, succinct sentences. Our dataset produced 216,931 total insights, about 13 insights per page and 655 insights per document.
Distilling concepts from insights: From the detailed list of insights, we identify higher-level concepts that connect related information about the document. This step significantly reduces noise and redundant information in the document while preserving essential information and themes. Our dataset produced 14,824 total concepts, about 1 concept per page and 45 concepts per document.
Creating abstracts from concepts: Given the insights and concepts in the document, the LLM writes an abstract that appears both better than any abstract a human would write and more information-dense than any abstract present in the original document. The LLM generated abstract provides incredibly comprehensive knowledge about the document with a small token density that carries a significant amount of information. We produce one abstract per document, 331 total.
Storing recollections/memories across documents: At the top of the pyramid we store critical information that is useful across all tasks. This can be information that the user shares about the task or information the agent learns about the dataset over time by researching and responding to tasks. For example, we can store the current 30 companies in the DOW as a recollection since this list is different from the 30 companies in the DOW at the time of the model’s knowledge cutoff. As we conduct more and more research tasks, we can continuously improve our recollections and maintain an audit trail of which documents these recollections originated from. For example, we can keep track of AI strategies across companies, where companies are making major investments, etc. These high-level connections are super important since they reveal relationships and information that are not apparent in a single page or document.

Sample subset of insights extracted from IBM 10Q, Q3 2024 (page 4)

We store the text and embeddings for each layer of the pyramid (pages and up) in Azure PostgreSQL. We originally used Azure AI Search, but switched to PostgreSQL for cost reasons. This required us to write our own hybrid search function since PostgreSQL doesn’t yet natively support this feature. This implementation would work with any vector database or vector index of your choosing. The key requirement is to store and efficiently retrieve both text and vector embeddings at any level of the pyramid.

This approach essentially creates the essence of a knowledge graph, but stores information in natural language, the way an LLM natively wants to interact with it, and is more efficient on token retrieval. We also let the LLM pick the terms used to categorize each level of the pyramid, this seemed to let the model decide for itself the best way to describe and differentiate between the information stored at each level. For example, the LLM preferred “insights” to “facts” as the label for the first level of distilled knowledge. Our goal in doing this was to better understand how an LLM thinks about the process by letting it decide how to store and group related information.

Using the pyramid: How it works with RAG & Agents

At inference time, both traditional RAG and agentic approaches benefit from the pre-processed, distilled information ingested in our knowledge pyramid. The pyramid structure allows for efficient retrieval in both the traditional RAG case, where only the top X related pieces of information are retrieved or in the Agentic case, where the Agent iteratively plans, retrieves, and evaluates information before returning a final response.

The benefit of the pyramid approach is that information at any and all levels of the pyramid can be used during inference. For our implementation, we used PydanticAI to create a search agent that takes in the user request, generates search terms, explores ideas related to the request, and keeps track of information relevant to the request. Once the search agent determines there’s sufficient information to address the user request, the results are re-ranked and sent back to the LLM to generate a final reply. Our implementation allows a search agent to traverse the information in the pyramid as it gathers details about a concept/search term. This is similar to walking a knowledge graph, but in a way that’s more natural for the LLM since all the information in the pyramid is stored in natural language.

Depending on the use case, the Agent could access information at all levels of the pyramid or only at specific levels (e.g. only retrieve information from the concepts). For our experiments, we did not retrieve raw page-level data since we wanted to focus on token efficiency and found the LLM-generated information for the insights, concepts, abstracts, and recollections was sufficient for completing our tasks. In theory, the Agent could also have access to the page data; this would provide additional opportunities for the agent to re-examine the original document text; however, it would also significantly increase the total tokens used.

Here is a high-level visualization of our Agentic approach to responding to user requests:

Overview of the agentic research & response process — Image created by author and team providing an overview of the agentic research & response process

Results from the pyramid: Real-world examples

To evaluate the effectiveness of our approach, we tested it against a variety of question categories, including typical fact-finding questions and complex cross-document research and analysis tasks.

Fact-finding (spear fishing):

These tasks require identifying specific information or facts that are buried in a document. These are the types of questions typical RAG solutions target but often require many searches and consume lots of tokens to answer correctly.

Example task: “What was IBM’s total revenue in the latest financial reporting?”

Example response using pyramid approach: “IBM’s total revenue for the third quarter of 2024 was $14.968 billion [ibm-10q-q3-2024.pdf, pg. 4]

Screenshot of total tokens used to research and generate response — Total tokens used to research and generate response

This result is correct (human-validated) and was generated using only 9,994 total tokens, with 1,240 tokens in the generated final response.

Complex research and analysis:

These tasks involve researching and understanding multiple concepts to gain a broader understanding of the documents and make inferences and informed assumptions based on the gathered facts.

Example task: “Analyze the investments Microsoft and NVIDIA are making in AI and how they are positioning themselves in the market. The report should be clearly formatted.”

Example response:

Screenshot of the response generated by the agent analyzing AI investments and positioning for Microsoft and NVIDIA. — Response generated by the agent analyzing AI investments and positioning for Microsoft and NVIDIA.

The result is a comprehensive report that executed quickly and contains detailed information about each of the companies. 26,802 total tokens were used to research and respond to the request with a significant percentage of them used for the final response (2,893 tokens or ~11%). These results were also reviewed by a human to verify their validity.

Screenshot of snippet indicating total token usage for the task — Snippet indicating total token usage for the task

Example task: “Create a report on analyzing the risks disclosed by the various financial companies in the DOW. Indicate which risks are shared and unique.”

Example response:

Screenshot of part 1 of a response generated by the agent on disclosed risks. — Part 1 of response generated by the agent on disclosed risks.

Screenshot of part 2 of a response generated by the agent on disclosed risks. — Part 2 of response generated by the agent on disclosed risks.

Similarly, this task was completed in 42.7 seconds and used 31,685 total tokens, with 3,116 tokens used to generate the final report.

Screenshot of a snippet indicating total token usage for the task — Snippet indicating total token usage for the task

These results for both fact-finding and complex analysis tasks demonstrate that the pyramid approach efficiently creates detailed reports with low latency using a minimal amount of tokens. The tokens used for the tasks carry dense meaning with little noise allowing for high-quality, thorough responses across tasks.

Benefits of the pyramid: Why use it?

Overall, we found that our pyramid approach provided a significant boost in response quality and overall performance for high-value questions.

Some of the key benefits we observed include:

Reduced model’s cognitive load: When the agent receives the user task, it retrieves pre-processed, distilled information rather than the raw, inconsistently formatted, disparate document chunks. This fundamentally improves the retrieval process since the model doesn’t waste its cognitive capacity on trying to break down the page/chunk text for the first time.
Superior table processing: By breaking down table information and storing it in concise but descriptive sentences, the pyramid approach makes it easier to retrieve relevant information at inference time through natural language queries. This was particularly important for our dataset since financial reports contain lots of critical information in tables.
Improved response quality to many types of requests: The pyramid enables more comprehensive context-aware responses to both precise, fact-finding questions and broad analysis based tasks that involve many themes across numerous documents.
Preservation of critical context: Since the distillation process identifies and keeps track of key facts, important information that might appear only once in the document is easier to maintain. For example, noting that all tables are represented in millions of dollars or in a particular currency. Traditional chunking methods often cause this type of information to slip through the cracks.
Optimized token usage, memory, and speed: By distilling information at ingestion time, we significantly reduce the number of tokens required during inference, are able to maximize the value of information put in the context window, and improve memory use.
Scalability: Many solutions struggle to perform as the size of the document dataset grows. This approach provides a much more efficient way to manage a large volume of text by only preserving critical information. This also allows for a more efficient use of the LLMs context window by only sending it useful, clear information.
Efficient concept exploration: The pyramid enables the agent to explore related information similar to navigating a knowledge graph, but does not require ever generating or maintaining relationships in the graph. The agent can use natural language exclusively and keep track of important facts related to the concepts it’s exploring in a highly token-efficient and fluid way.
Emergent dataset understanding: An unexpected benefit of this approach emerged during our testing. When asking questions like “what can you tell me about this dataset?” or “what types of questions can I ask?”, the system is able to respond and suggest productive search topics because it has a more robust understanding of the dataset context by accessing higher levels in the pyramid like the abstracts and recollections.

Beyond the pyramid: Evaluation challenges & future directions

Challenges

While the results we’ve observed when using the pyramid search approach have been nothing short of amazing, finding ways to establish meaningful metrics to evaluate the entire system both at ingestion time and during information retrieval is challenging. Traditional RAG and Agent evaluation frameworks often fail to address nuanced questions and analytical responses where many different responses are valid.

Our team plans to write a research paper on this approach in the future, and we are open to any thoughts and feedback from the community, especially when it comes to evaluation metrics. Many of the existing datasets we found were focused on evaluating RAG use cases within one document or precise information retrieval across multiple documents rather than robust concept and theme analysis across documents and domains.

The main use cases we are interested in relate to broader questions that are representative of how businesses actually want to interact with GenAI systems. For example, “tell me everything I need to know about customer X” or “how do the behaviors of Customer A and B differ? Which am I more likely to have a successful meeting with?”. These types of questions require a deep understanding of information across many sources. The answers to these questions typically require a person to synthesize data from multiple areas of the business and think critically about it. As a result, the answers to these questions are rarely written or saved anywhere which makes it impossible to simply store and retrieve them through a vector index in a typical RAG process.

Another consideration is that many real-world use cases involve dynamic datasets where documents are consistently being added, edited, and deleted. This makes it difficult to evaluate and track what a “correct” response is since the answer will evolve as the available information changes.

Future directions

In the future, we believe that the pyramid approach can address some of these challenges by enabling more effective processing of dense documents and storing learned information as recollections. However, tracking and evaluating the validity of the recollections over time will be critical to the system’s overall success and remains a key focus area for our ongoing work.

When applying this approach to organizational data, the pyramid process could also be used to identify and assess discrepancies across areas of the business. For example, uploading all of a company’s sales pitch decks could surface where certain products or services are being positioned inconsistently. It could also be used to compare insights extracted from various line of business data to help understand if and where teams have developed conflicting understandings of topics or different priorities. This application goes beyond pure information retrieval use cases and would allow the pyramid to serve as an organizational alignment tool that helps identify divergences in messaging, terminology, and overall communication.

Conclusion: Key takeaways and why the pyramid approach matters

The knowledge distillation pyramid approach is significant because it leverages the full power of the LLM at both ingestion and retrieval time. Our approach allows you to store dense information in fewer tokens which has the added benefit of reducing noise in the dataset at inference. Our approach also runs very quickly and is incredibly token efficient, we are able to generate responses within seconds, explore potentially hundreds of searches, and on average use (this includes all the search iterations!).

We find that the LLM is much better at writing atomic insights as sentences and that these insights effectively distill information from both text-based and tabular data. This distilled information written in natural language is very easy for the LLM to understand and navigate at inference since it does not have to expend unnecessary energy reasoning about and breaking down document formatting or filtering through noise.

The ability to retrieve and aggregate information at any level of the pyramid also provides significant flexibility to address a variety of query types. This approach offers promising performance for large datasets and enables high-value use cases that require nuanced information retrieval and analysis.

Note: The opinions expressed in this article are solely my own and do not necessarily reflect the views or policies of my employer.

Interested in discussing further or collaborating? Reach out on LinkedIn!

Stay Ahead

Explore More Insights

Stay ahead with more perspectives on cutting-edge power, infrastructure, energy, bitcoin and AI solutions. Explore these articles to uncover strategies and insights shaping the future of industries.

Eni North Galal continues contracting for Kutei North Hub field development offshore Indonesia

Eni North Ganal Ltd., a subsidiary of Searah Ltd., has let a contract to SLB OneSubsea to deliver a steel tube umbilical system for the Kutei North Hub field development project offshore East Kalimantan, Indonesia. SLB OneSubsea will engineer, procure, and manufacture 94.6 km of umbilical for water depths of

AI job titles expand beyond tech as IT hiring remains strong

Separately, according to CompTIA’s latest Tech Jobs Report, employers posted more than 280,000 new technology job postings in June, marking the sixth consecutive month of growth. Active technology job postings approached 600,000, while employment in tech occupations increased by 47,000 positions. The unemployment rate for tech occupations fell to 2.9%,

Broadcom will sell more US-made components to Apple

Apple has agreed to buy up to $30 billion-worth of additional chips and other wireless components from Broadcom over the next few years. The new agreement could lead to Broadcom making up to 15 billion more chips in the US, and enable it to modernize its manufacturing facilities in Fort

AI tie-in accelerates quantum usefulness, early adopters say

The Quantum Tech World conference showcases this ecosystem, she says. According to conference organizers, more than 1,300 people attended this year, and there were more than one hundred sponsors. Among them were multiple quantum computer makers, including Quantum Computing Inc., a maker of room-temperature photonic computers, which ran a real-time

PTTEP achieves Thailand’s first wellhead platform reuse in Gulf of Thailand

PTT Exploration and Production Public Co. Ltd. (PTTEP) has completed Thailand’s first total wellhead platform reuse project by redeploying an entire decommissioned petroleum wellhead platform as a complete structure in Funan field in the Gulf of Thailand. The reuse project comes as part of PTTEP’s program to maximize value and extend utilization of wellhead platforms that remain structurally sound and safe after depleting resources at a location by redeploying the platform as a complete structure. The first implementation was carried out at the Jakrawan K wellhead platform (JKWK), in Funan field under the G1/61 Project. As part of the project, PTTEP adopted the wet-tow method to relocate the jacket, helping curb energy consumption and minimize impacts on marine life attached to the platform structure, supporting a balance between energy production and marine environmental stewardship. The topside, jacket, and selected pile sections were relocated and reinstalled for use within the same field, reducing the overall construction and installation period to only 6 months, down from about 20 months for a newly built platform. Additionally, the approach cut construction costs by about 35–50% compared with construction of an entirely new wellhead platform. PTTEP said it expects the initiative to also reduce greenhouse gas emissions by about 3,270 tonnes of CO2e/platform by limiting the use of steel and other equipment required for construction of new platforms. PTTEP is operator of the G1/61 project (60%) with partner Mubadala Investment Co. (40%).

Trump declares Iran ceasefire over; oil surges on renewed supply risk

US President Donald Trump said the ceasefire and memorandum of understanding (MOU) reached with Iran last month is effectively over following a fresh exchange of strikes, reigniting supply concerns and sending crude prices sharply higher. Speaking alongside NATO Secretary-General Mark Rutte at the alliance’s summit in Ankara, Pres. Trump said Washington no longer sees value in maintaining the ceasefire framework with Tehran, though he left open the possibility of continued talks. He added that further US military action against Iran remains likely after strikes overnight. Stay updated on oil price volatility, shipping disruptions, LNG market analysis, and production output at OGJ’s Iran war content hub. The escalation was triggered by alleged Iranian attacks on three commercial vessels transiting the Strait of Hormuz on July 7. US Central Command said it responded with strikes on more than 80 Iranian targets, including air defense systems, command-and-control infrastructure, anti-ship missile capabilities, and over 60 Islamic Revolutionary Guard Corps (IRGC) fast boats operating in and near the strait. US Central Command described the tanker attacks as a clear violation of the June 17 agreement. Iran’s Foreign Ministry called the US strikes a breach of the MOU and said Tehran would continue to defend its sovereignty. The IRGC said it retaliated with drone and missile strikes targeting US military facilities in Bahrain and Kuwait. Authorities in both countries reported intercepting incoming projectiles, with no material damage confirmed. Trump said on July 8 the US is considering reinstating a naval blockade targeting Iranian ports and vessels. He also raised the possibility of strikes on civilian infrastructure, including electric plants and desalination facilities, as well as a potential move to take control of Kharg Island, home to the bulk of Iran’s crude export infrastructure. He said Tuesday’s strikes had reached the island but had not targeted its

US EIA forecasts declining oil prices as supply disruptions ease

In its July 7 Short-Term Energy Outlook (STEO) report, the US Energy Information Administration (EIA) said it expects global oil prices to decline as supply disruptions linked to the Strait of Hormuz ease and production recovers. On June 18, the US and Iran signed a memorandum of understanding to end the conflict and reopen the strait, which had been largely closed since Feb. 28. The disruption to this critical oil transit chokepoint constrained global flows, driving major price volatility. Brent crude averaged $85/bbl in June, down $22/bbl from May and $32/bbl below its April peak. Prices fell below $70/bbl on July 1 as tanker traffic through the strait increased sharply, easing supply concerns. EIA now expects most shut-in crude production to return to near pre-conflict levels by yearend, with full restoration largely to be completed by first-quarter 2027. Despite the recovery in flows, global inventories remain significantly depleted following earlier draws. EIA estimates oil inventories declined by an average of 5.1 million b/d in second-quarter 2026 and will fall by a further 2.2 million b/d in third-quarter 2026, as much of the recent tanker movement reflects previously stranded cargoes. As a result, the market is expected to remain relatively tight through most of third-quarter 2026 before shifting back into oversupply. EIA forecasts global oil consumption will decline by 1.2 million b/d in 2026, led by a 0.8 million b/d drop in non-OECD demand, particularly in the Asia Pacific. Demand is expected to rebound in 2027 as prices ease and supply normalizes, with consumption rising by 2.0 million b/d to 104.8 million b/d. As supply growth outpaces demand, inventories are projected to build by 2.7 million b/d in fourth-quarter 2026 and by 5.0 million b/d in 2027. This shift is expected to place sustained downward pressure on prices. EIA forecasts Brent

Eni lets EPCI contract for Kutei North Hub field FPSO

Eni North Ganal has let an engineering, procurement, construction, and installation (EPCI) contract to a joint venture between PT Saipem Indonesia and PT Tripatra Engineers and Constructors for a floating production, storage, and offloading (FPSO) unit for the Kutei North Hub Field Development Project in Kutei basin, offshore Indonesia, about 70 km off East Kalimantan. The project execution, with an estimated duration of 48 months, includes project management, engineering, procurement of materials, fabrication, construction and installation activities, as well as commissioning and start-up of the FPSO unit. The contract is valued at about $2 billion for Saipem’s share. The Kutei FPSO project is part of the Kutei North Hub Development, which comprises a subsea development tied back to the new FPSO, a dedicated gas export pipeline to the Bontang LNG plant, and domestic gas users via the existing East Kalimantan System. Eni North Ganal is controlled by Searah Ltd., which was formed through a strategic partnership between Eni and Petronas.

EIA: US crude inventories up 3.0 million bbl

US crude oil inventories for the week ended July 3, excluding the Strategic Petroleum Reserve, increased by 3.0 million bbl from the previous week, according to data from the US Energy Information Administration (EIA). At 411.4 million bbl, US crude oil inventories are about 6% below the 5-year average for this time of year, the EIA report indicated. EIA said total motor gasoline inventories decreased by 1.9 million bbl from last week and are about 6% below the 5-year average for this time of year. Finished gasoline inventories and blending components inventories both decreased last week. Distillate fuel inventories decreased by 5.0 million bbl last week and are about 12% below the 5-year average for this time of year. Propane-propylene inventories decreased by 800,000 bbl from last week and are 29% above the 5-year average for this time of year, EIA said. US crude oil refinery inputs averaged 17.0 million b/d for the week ended July 3, which was 173,000 b/d less than the previous week’s average. Refineries operated at 95.8% of capacity. Gasoline production decreased, averaging 9.7 million b/d. Distillate fuel production decreased, averaging 5.2 million b/d. US crude oil imports averaged 5.6 million b/d, up 351,000 b/d from the previous week. Over the last 4 weeks, crude oil imports averaged about 5.4 million b/d, 11.4% less than the same 4-week period last year. Total motor gasoline imports averaged 423,000 b/d. Distillate fuel imports averaged 87,000 b/d.

Marubeni closes deal to acquire Barnett-focused EagleRidge Energy

Marubeni Corp. has closed a deal to acquire EagleRidge Energy II LLC, a natural gas operator based in Dallas, Tex., advancing its effort to expand energy operations and natural gas assets in North America. EagleRidge, now a wholly owned subsidiary of Marubeni, has focused its operations on the Barnett shale in North Texas. The company operates over 3,500 wells and produces 300 MMcfed over 450,000 gross acres across 16 counties. The deal, which was announced in June, increases Marubeni’s production capacity in the Barnett shale to about 170 MMcfed. With the transaction closed, Marubeni said in a July 6 update, the EagleRidge management and operational team remains in place, with updates to executive leadership. Tom Ashton and Sam Miller will share the role of co-presidents. Hiroki Shima has been appointed chairman, and Michael Ronca transitions to vice-chairman.

Infoblox acquires Kentik, adding network observability to its DNS and DDI platform

Infoblox announced today that it has entered into a definitive agreement to acquire Kentik, combining Infoblox’s authoritative DNS, DHCP, and IP address management (IPAM) data with Kentik’s network observability platform. Financial terms were not disclosed. Kentik was founded in 2014, originally as CloudHelix before rebranding the following year, and has raised more than $100 million in venture funding to date. The platform provides real-time visibility into network traffic and ingests flow data, routing intelligence, and device telemetry across data centers, cloud environments, WANs, and the public internet. In recent years, the company has enhanced its platform with an AI advisor that helps to accelerate investigations. Infoblox has spent more than two decades managing the DNS, DHCP, and IPAM services enterprises rely on to stay connected. In 2024, it first launched its Universal DDI SaaS platform for managing DNS, DHCP, and IP addresses from a single place, expanding in 2025 to more providers. DDI refers to the trio of core network services in IP networks: DNS, which turns domain names into IP addresses; DHCP, which assigns IP addresses to resources; and IPAM, which manages the network’s IP address infrastructure.

Building the AI Optical Layer: Connectivity, Standards, and the Future of AI Infrastructure

As AI data centers push past the limits of traditional compute architecture, the industry’s attention is moving deeper into the physical layer. GPUs, accelerators, power systems and cooling platforms still dominate the headlines, but the network fabric that connects those systems is becoming just as critical. A wave of recent announcements points to the same conclusion: future growth will depend not only on more compute, but on faster, denser, more efficient and more scalable optical connectivity. A new multi-source agreement is bringing together major technology companies to standardize expanded beam optical connectivity for AI data centers. University of Arizona research is powering a new optical switching technology designed to reduce the energy consumed by data center networks. STL is planning to invest up to $100 million in U.S. manufacturing capacity to support AI data center and telecom customers with optical connectivity products. Those developments are now being reinforced by a broader series of moves across the optical ecosystem: Corning’s major AI infrastructure partnerships with NVIDIA and Amazon, GlobalFoundries’ push into co-packaged optics, Sivers’ laser-array collaboration with GlobalFoundries, Wiwynn’s co-packaged optics demonstration at Computex, Credo’s acquisition of DustPhotonics, and emerging near-packaged optical interconnect designs from LightSpeed Photonics. Taken together, these announcements highlight a maturing market around the optical layer of AI infrastructure. The value is not simply faster data movement. It is about reducing deployment complexity, lowering operating overhead, supporting higher-density clusters, improving energy efficiency and strengthening the domestic supply chain behind AI-ready networks. Let’s drill down into what these announcements mean. Standards for the AI Optical Layer The launch of a new coalition focused on expanded beam optical, or EBO, connectivity reflects a practical challenge facing AI deployments: as clusters grow larger and more bandwidth-intensive, physical connections become harder to deploy, maintain and scale. 3M announced that it has joined

Data Center Jobs: Engineering, Construction, Commissioning, Sales, Field Service and Facility Tech Jobs Available in Major Data Center Hotspots

Each month Data Center Frontier, in partnership with Pkaza, posts some of the hottest data center career opportunities in the market. Here’s a look at some of the latest data center jobs posted on the Data Center Frontier jobs board, powered by Pkaza Critical Facilities Recruiting. Looking for Data Center Candidates? Check out Pkaza’s Active Candidate / Featured Candidate Hotlist CFD Engineer – Data Center Mechanical DesignNew York, NY (remote)This position is also available as a remote role anywhere in the US, in addition to key markets such as Cedar Rapids, IA; Kansas City, MO or White Plains, NY. Our client is an engineering design and commissioning company that has a national footprint and specializes in MEP critical facilities design. They provide design, commissioning, consulting and management expertise in the critical facilities space. They have a mindset to provide reliability, energy efficiency, and sustainable design expertise when providing these consulting services for enterprise, colocation and hyperscale companies. This career-growth minded opportunity offers exciting projects with leading-edge technology and innovation as well as competitive salaries and benefits. Electrical Commissioning Engineer New Albany, OH (limited travel)Non-Traveling CxA positions available in: Indianapolis, IN; Cedar Rapids, IA and Austin, TX. Traveling CxA Roles: New York, NY; White Plain, NY; Morristown, NJ; Dallas, TX; Richmond, VA; Ashburn, VA; Montvale, NJ; Charlotte, NC; Atlanta, GA; Phoenix, AZ; Salt Lake City, UT; Kansas City, MO; Omaha, NE; Chesterton, IN or Chicago, IL. *** ALSO looking for a LEAD EE and ME CxA Agents and CxA PMs. *** Our client is an engineering design and commissioning company that has a national footprint and specializes in MEP critical facilities design. They provide design, commissioning, consulting and management expertise in the critical facilities space. They have a mindset to provide reliability, energy efficiency, sustainable design and LEED expertise when providing these consulting services

H5 Data Centers’ 325 Hudson: A Manhattan Carrier Hotel with SoHo DNA – DCF Tours

A Carrier Hotel Reimagined for Modern Colocation H5 formally announced its expansion into 325 Hudson, a 225,000 square-foot mixed-use building comprised of office, lab and data center uses, in 2021 through a partnership with real estate investment firm DivcoWest, describing its new location as a data center and carrier hotel in one of the world’s largest communications markets. At the time, H5 founder and CEO Josh Simms framed the move as an opportunity to expand an already-established interconnection ecosystem while supporting growing demand from cloud providers, content delivery networks, and communications carriers. That vision now appears fully realized inside the building. The facility today operates as both a traditional carrier hotel and a modern enterprise colocation environment. H5’s infrastructure footprint supports high-density deployments, A/B UPS power architecture, N+1 emergency generators, N+1 CRAC systems, and energy-efficient in-row cooling with cold aisle containment. The building also reflects the physical realities of Manhattan infrastructure engineering. Operators work within vertical constraints rather than sprawling horizontal campuses. Freight access, riser strategy, structured cabling pathways, and efficient floor utilization become critical operational variables. H5 highlighted several features tailored for those realities, including 13-foot slab-to-slab heights, 150 pounds-per-square-foot floor loading capability, secure loading access, and extensive pre-built conduit infrastructure.

AI Infrastructure Demands a New Operating System for Project Delivery

For much of the data center industry’s history, project management has largely been viewed as an execution discipline: a collection of schedules, milestones, spreadsheets, and status meetings designed to shepherd individual facilities from groundbreaking to commissioning. The AI era is rapidly rendering that model obsolete. As hyperscalers, developers, utilities, EPC firms, telecom providers, equipment suppliers, and local governments converge around increasingly complex AI campuses, the challenge is no longer simply delivering projects on time. Rather, it is orchestrating an infrastructure manufacturing process that stretches from land acquisition and permitting through construction, operations, and ultimately asset modernization years later. That changing reality was a central theme during Data Center Frontier’s conversation with Sitetracker at Fiber Connect 2026. The company’s perspective reflects a broader shift underway across digital infrastructure: project management is evolving into lifecycle management, where financial planning, regulatory coordination, supply chain visibility, and operational readiness become inseparable parts of the same platform. Complexity Begins Before Construction Much of the attention surrounding AI infrastructure focuses on GPU deployments, liquid cooling, and power availability. Yet Sitetracker argues that many of today’s greatest operational headaches begin much earlier in the development process. According to Reilly McClure, Sr. Product Marketing Manager – Digital Infrastructure with Sitetracker, operators are increasingly seeking help with land acquisition, parcel management, and site identification as AI infrastructure expands into new markets. “There are so many variables that we need to track,” he explained. “The demand and the growth and the build-out for where all that infrastructure is going is becoming increasingly complex. They’re finding it just cannot be done on a spreadsheet.” That observation resonates across the industry. Finding suitable sites now requires simultaneously evaluating power availability, transmission timelines, fiber access, permitting requirements, environmental studies, municipal approvals, and community considerations; all while competing developers race to secure the same

Gartner: Data center electricity consumption to grow 26% in 2026

AI-optimized servers are a relatively new phenomenon but they have rapidly gained uh traditional data centers in terms of power use. Gartner estimates AI-optimized server adoption will account for 31% of data center power consumption in 2026, and that by 2027 their power consumption will surpass that of conventional servers. “Surging demand for compute-intensive AI workloads is driving unprecedented data center power growth, while AI capacity is now constrained by power availability, making data center power security the new battle ground for scaling and protecting margins in the global AI race,” said Wang in a statement. Wang said of the 565TWh consumed this year, the U.S. will account for about 204TWh, or 36% of the total amount consumed. And of the 204TWh consumed this year, dedicated AI data centers will consume 68TWh, or one-third of the total. So in just five years, AI data centers have gone from zero to of the total power consumption in the US.

Microsoft will invest $80B in AI data centers in fiscal 2025

And Microsoft isn’t the only one that is ramping up its investments into AI-enabled data centers. Rival cloud service providers are all investing in either upgrading or opening new data centers to capture a larger chunk of business from developers and users of large language models (LLMs). In a report published in October 2024, Bloomberg Intelligence estimated that demand for generative AI would push Microsoft, AWS, Google, Oracle, Meta, and Apple would between them devote $200 billion to capex in 2025, up from $110 billion in 2023. Microsoft is one of the biggest spenders, followed closely by Google and AWS, Bloomberg Intelligence said. Its estimate of Microsoft’s capital spending on AI, at $62.4 billion for calendar 2025, is lower than Smith’s claim that the company will invest $80 billion in the fiscal year to June 30, 2025. Both figures, though, are way higher than Microsoft’s 2020 capital expenditure of “just” $17.6 billion. The majority of the increased spending is tied to cloud services and the expansion of AI infrastructure needed to provide compute capacity for OpenAI workloads. Separately, last October Amazon CEO Andy Jassy said his company planned total capex spend of $75 billion in 2024 and even more in 2025, with much of it going to AWS, its cloud computing division.

John Deere unveils more autonomous farm machines to address skill labor shortage

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More Self-driving tractors might be the path to self-driving cars. John Deere has revealed a new line of autonomous machines and tech across agriculture, construction and commercial landscaping. The Moline, Illinois-based John Deere has been in business for 187 years, yet it’s been a regular as a non-tech company showing off technology at the big tech trade show in Las Vegas and is back at CES 2025 with more autonomous tractors and other vehicles. This is not something we usually cover, but John Deere has a lot of data that is interesting in the big picture of tech. The message from the company is that there aren’t enough skilled farm laborers to do the work that its customers need. It’s been a challenge for most of the last two decades, said Jahmy Hindman, CTO at John Deere, in a briefing. Much of the tech will come this fall and after that. He noted that the average farmer in the U.S. is over 58 and works 12 to 18 hours a day to grow food for us. And he said the American Farm Bureau Federation estimates there are roughly 2.4 million farm jobs that need to be filled annually; and the agricultural work force continues to shrink. (This is my hint to the anti-immigration crowd). John Deere’s autonomous 9RX Tractor. Farmers can oversee it using an app. While each of these industries experiences their own set of challenges, a commonality across all is skilled labor availability. In construction, about 80% percent of contractors struggle to find skilled labor. And in commercial landscaping, 86% of landscaping business owners can’t find labor to fill open positions, he said. “They have to figure out how to do

2025 playbook for enterprise AI success, from agents to evals

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More 2025 is poised to be a pivotal year for enterprise AI. The past year has seen rapid innovation, and this year will see the same. This has made it more critical than ever to revisit your AI strategy to stay competitive and create value for your customers. From scaling AI agents to optimizing costs, here are the five critical areas enterprises should prioritize for their AI strategy this year. 1. Agents: the next generation of automation AI agents are no longer theoretical. In 2025, they’re indispensable tools for enterprises looking to streamline operations and enhance customer interactions. Unlike traditional software, agents powered by large language models (LLMs) can make nuanced decisions, navigate complex multi-step tasks, and integrate seamlessly with tools and APIs. At the start of 2024, agents were not ready for prime time, making frustrating mistakes like hallucinating URLs. They started getting better as frontier large language models themselves improved. “Let me put it this way,” said Sam Witteveen, cofounder of Red Dragon, a company that develops agents for companies, and that recently reviewed the 48 agents it built last year. “Interestingly, the ones that we built at the start of the year, a lot of those worked way better at the end of the year just because the models got better.” Witteveen shared this in the video podcast we filmed to discuss these five big trends in detail. Models are getting better and hallucinating less, and they’re also being trained to do agentic tasks. Another feature that the model providers are researching is a way to use the LLM as a judge, and as models get cheaper (something we’ll cover below), companies can use three or more models to

OpenAI’s red teaming innovations define new essentials for security leaders in the AI era

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More OpenAI has taken a more aggressive approach to red teaming than its AI competitors, demonstrating its security teams’ advanced capabilities in two areas: multi-step reinforcement and external red teaming. OpenAI recently released two papers that set a new competitive standard for improving the quality, reliability and safety of AI models in these two techniques and more. The first paper, “OpenAI’s Approach to External Red Teaming for AI Models and Systems,” reports that specialized teams outside the company have proven effective in uncovering vulnerabilities that might otherwise have made it into a released model because in-house testing techniques may have missed them. In the second paper, “Diverse and Effective Red Teaming with Auto-Generated Rewards and Multi-Step Reinforcement Learning,” OpenAI introduces an automated framework that relies on iterative reinforcement learning to generate a broad spectrum of novel, wide-ranging attacks. Going all-in on red teaming pays practical, competitive dividends It’s encouraging to see competitive intensity in red teaming growing among AI companies. When Anthropic released its AI red team guidelines in June of last year, it joined AI providers including Google, Microsoft, Nvidia, OpenAI, and even the U.S.’s National Institute of Standards and Technology (NIST), which all had released red teaming frameworks. Investing heavily in red teaming yields tangible benefits for security leaders in any organization. OpenAI’s paper on external red teaming provides a detailed analysis of how the company strives to create specialized external teams that include cybersecurity and subject matter experts. The goal is to see if knowledgeable external teams can defeat models’ security perimeters and find gaps in their security, biases and controls that prompt-based testing couldn’t find. What makes OpenAI’s recent papers noteworthy is how well they define using human-in-the-middle