Field notes on building AI systems that hold up in real work. Written by Teja Sagiraju, an AI engineer with 14 years in data science and machine learning. Every system here is built end to end, in public.
17 notes · 2 systems · 5 planned · RSSStart here
Nothing found isn't the same as nothing checked
My AI judges agreed with themselves. Not with each other.
Latest
The invoice said £244.80. Why did the AI tool show £0?
Who chose zero? I did. Four rules for a failed AI request, replayed on 81 tasks: what each one lets through, and what it costs.
Semantic caching, explained: when an AI can reuse an answer
A semantic cache reuses an AI answer when a new question is close in meaning to an old one. The real test is whether that answer is still right for this customer, today.
Version 1 let an AI agent decide. Version 2 lets written rules decide.
A better model removes the model's excuses. What's left is the design: if a check has no way to say it didn't run, it will be reported as clean.
I didn’t need the expensive model. I needed to know when to call it.
On a £100.01 refund, a cheap AI asked Finance for a ticket twice, and once just said the job was done. I tested two ways to tell when an answer needs a second opinion.
Before you quote a UK company, know what wasn’t checked
An underwriter needs a route and its reasons before quoting a company, plus an honest list of what wasn’t checked. A tool that hides its gaps is worse than none.
Built in public
Each system has its own page, code and notes, and says plainly what it doesn’t do yet.
PolicyTrace
Reads a UK motor policy pack and gives a claims handler one checked record, with every value traced to its line in the PDF.
Last run on the demo pack (18 Oct): 59 fields · 46 traced to the PDF
A first look at a UK company for a commercial underwriter: rules decide, agents explain, and every check it couldn’t run is shown.
What’s planned RoadmapDocuments, vision and testing.
Working in insurance? The insurance systems and notes are on one page.
