--- title: "Blog" url: "https://predictes.com/blog" description: "Practical writing on getting AI into production: framing, ontology, data, agents, deployment and what happens after go-live." --- # Blog Practical writing on getting AI into production: framing, ontology, data, agents, deployment and what happens after go-live. September 4, 2026·3 min read ## [Your AI use-case list has thirty items and no owner](https://predictes.com/blog/ai-use-case-list-with-no-owner) Long use-case lists feel like progress and function as a way of avoiding a decision. What replaces them is shorter, harder to write and actually gets built. - AI strategy - prioritisation - use cases [Read article →](https://predictes.com/blog/ai-use-case-list-with-no-owner) August 28, 2026·3 min read ## [Adoption is a budget line, not a hope](https://predictes.com/blog/adoption-is-a-budget-line) Every transformation plan has a technology budget and an adoption paragraph. That imbalance predicts the outcome more reliably than anything about the technology. - change management - adoption - programmes [Read article →](https://predictes.com/blog/adoption-is-a-budget-line) August 21, 2026·3 min read ## [An agent is only as safe as the tools you gave it](https://predictes.com/blog/an-agent-is-as-safe-as-its-tools) Most attempts to make agents safe happen in the prompt. Almost all of the real safety comes from the tool layer, which is an engineering decision. - AI agents - security - tool design [Read article →](https://predictes.com/blog/an-agent-is-as-safe-as-its-tools) August 14, 2026·2 min read ## [Design for the review, not for the answer](https://predictes.com/blog/design-for-the-review-not-the-answer) The metric that decides whether an AI feature saves anybody time is not accuracy. It is how long it takes a person to confirm or correct what it produced. - UX - AI interfaces - review [Read article →](https://predictes.com/blog/design-for-the-review-not-the-answer) August 7, 2026·3 min read ## [Build the evaluation harness before the feature](https://predictes.com/blog/build-the-evaluation-harness-first) Without a way to score changes, an AI codebase gets slower every month. Teams feel it around month three and usually blame the wrong thing. - evaluation - testing - LLM [Read article →](https://predictes.com/blog/build-the-evaluation-harness-first) July 31, 2026·2 min read ## [Most retrieval failures are definition failures](https://predictes.com/blog/most-retrieval-failures-are-definition-failures) When search returns something technically related and practically wrong, the usual response is to tune the retriever. The usual cause is elsewhere. - ontology - retrieval - master data [Read article →](https://predictes.com/blog/most-retrieval-failures-are-definition-failures) July 24, 2026·2 min read ## [The answer was not wrong. The document was three months old](https://predictes.com/blog/the-answer-was-wrong-because-the-document-was-old) Freshness is the quietest failure in an AI system. Nothing errors, nothing alerts, and the answers are perfectly consistent with a version of reality that has moved on. - data quality - freshness - pipelines [Read article →](https://predictes.com/blog/the-answer-was-wrong-because-the-document-was-old) July 17, 2026·2 min read ## [The AI policy people actually follow](https://predictes.com/blog/the-ai-policy-people-actually-follow) Most AI policies are written to satisfy a risk register and are ignored within a month. The version that works is shorter, more permissive and considerably more useful. - AI policy - governance - training [Read article →](https://predictes.com/blog/the-ai-policy-people-actually-follow) July 10, 2026·3 min read ## [AI systems do not crash. They drift](https://predictes.com/blog/ai-systems-do-not-crash-they-drift) Conventional monitoring watches for things that stop. The failure mode that matters here produces no error, no alert and perfectly formatted answers that are quietly worse. - monitoring - drift - SLA [Read article →](https://predictes.com/blog/ai-systems-do-not-crash-they-drift) July 3, 2026·2 min read ## [Go and watch a shift before you write a line of code](https://predictes.com/blog/go-and-watch-a-shift-first) Field systems are rejected for reasons that never appear in a requirements document, and every one of them is visible within two hours of standing where the work happens. - field deployment - requirements - adoption [Read article →](https://predictes.com/blog/go-and-watch-a-shift-first) June 26, 2026·2 min read ## [When agents write the code, review becomes the bottleneck](https://predictes.com/blog/when-agents-write-the-code-review-is-the-bottleneck) Generation got ten times faster and review did not. Teams that measure only the first half accumulate under-reviewed code, and the debt arrives all at once. - AI coding - code review - spec-driven [Read article →](https://predictes.com/blog/when-agents-write-the-code-review-is-the-bottleneck) June 19, 2026·3 min read ## [The night shift knows things nobody wrote down](https://predictes.com/blog/the-night-shift-knows-things-nobody-wrote-down) Unattended production is rarely blocked by robotics. It is blocked by the judgement three experienced people carry, which exists nowhere a system can read. - dark factory - tacit knowledge - context [Read article →](https://predictes.com/blog/the-night-shift-knows-things-nobody-wrote-down)