How we think
Underwriting Data Extraction: ACORD Forms, Submissions and Loss Runs
Underwriting Data Extraction: ACORD Forms, Submissions and Loss Runs
How insurers extract clean data from ACORD forms, broker submissions and loss runs with AI, where extraction breaks, and how to measure accuracy.
Read article
Knowledge Graphs and LLMs: When You Need Both
Knowledge Graphs and LLMs: When You Need Both
Knowledge graphs and LLMs solve different problems. Learn when vector RAG is enough, when to add a graph, and how to combine both in one assistant.
Read article
LLMOps vs MLOps: What Actually Changes
LLMOps vs MLOps: What Actually Changes
LLMOps vs MLOps compared: what carries over from your MLOps platform, and what changes in evaluation, retrieval data, monitoring, cost, security and releases for LLM applications.
Read article
Build vs. Buy vs. Orchestrate: The Real AI Capability Decision
Build vs. Buy vs. Orchestrate: The Real AI Capability Decision
When organizations start looking at AI, the conversation often starts with the classic build vs buy AI question: “Should we build it or buy it?” From the discovery conversations I have been part of, I think that question is becoming too narrow. Today, organizations have a third option: orchestrate. The real question is: What should […]
Read article
The AI Services Market Is Splitting: Headcount Sellers vs. Outcome Engineers
The AI Services Market Is Splitting: Headcount Sellers vs. Outcome Engineers
AI is splitting IT services into headcount sellers and outcome engineers. See how outcome-based pricing works and what to ask your AI services partner.
Read article
AI in Underwriting: Reading ACORD Forms and Loss Runs
AI in Underwriting: Reading ACORD Forms and Loss Runs
AI in underwriting is only as good as the submission it reads. How AI handles ACORD forms, broker emails, spreadsheets and loss runs, where it goes wrong, and how to measure it before you rely on it.
Read article
When Software Gets Cheap, Your Data Becomes the Moat
When Software Gets Cheap, Your Data Becomes the Moat
When anyone can build software with AI, proprietary data becomes the moat. Here’s what a real data moat looks like, and three moves to start building yours.
Read article
If AI Writes Code, Who Tests It?
If AI Writes Code, Who Tests It?
AI can write working code in seconds. The harder question is who makes sure it is right for the customer, the business and the risk you are willing to accept. A QE leader on why quality engineering has to move upstream.
Read article
Enterprise Search That Cites Its Sources: How It Works
Enterprise Search That Cites Its Sources: How It Works
AI enterprise search that cites its sources links every answer to the passage behind it. Here is how the seven-step pipeline works, where it goes wrong and how to measure it before go-live.
Read article
Why Agentic AI Deals Stall Between the Demo and the Statement of Work
Why Agentic AI Deals Stall Between the Demo and the Statement of Work
Agentic AI pilots rarely fail on technology. They stall because the business case was never documented, governance arrived late and the champion was left to sell it alone. Here is how to sequence the next pilot so it reaches production.
Read article
Document AI in 2026: From Messy Documents to AI-Ready Data
Document AI in 2026: From Messy Documents to AI-Ready Data
Document AI turns PDFs, scans and emails into AI-ready data. See how intelligent document processing works in 2026, plus a 90-day roadmap to production.
Read article
Mid-Market AI: The Enterprise Playbook Doesn’t Scale Down
Mid-Market AI: The Enterprise Playbook Doesn’t Scale Down
Why the enterprise AI playbook fails at mid-market companies, and what works instead: AI that runs on imperfect data, needs no dedicated team, has governance built in and pays back within a quarter or two.
Read article