Your DORA Metrics Are Lying Now That the Team Uses AI
Deployment frequency and lead time still matter, but AI-assisted development broke how most teams measure them. Here is what to instrument instead.
- Devops
- Engineering metrics
- Dora
Deployment frequency and lead time still matter, but AI-assisted development broke how most teams measure them. Here is what to instrument instead.
Model access is a commodity your competitors buy at the same price. Your labelled evaluation set is not purchasable at any price. Build it first.
Before you build the AI feature you promised the board, test four things: data reality, permissions, evaluation and failure economics. In that order.
Internal AI chat tools spike at launch and decay. Embedding AI inside the workflow on a defined trigger is what actually holds adoption.
Where AI agents genuinely work in delivery pipelines, where they must not be trusted, and the permission boundary that makes the difference.
Eight testable competencies that correlate with mobile engineering performance, and the specific question that probes each one in an interview.
What a product discovery phase must produce for engineers to act on it: domain model, integration map, risk register and an explicit exclusion list.
Agent assist is a controlled proving ground that generates labelled evidence of which intents are safe to automate. Here is the full sequence.
Insurance platforms connect to rating providers, payments, reinsurance, brokers and regulators. Each brings its own protocol, contract and failure mode.
When an agent calls your API for a user through a platform, identity has three answers. Most gateways were not designed for that.
Build an LLM pre-check for brokerage documents: narrow prompts, evidence-cited findings, per-check evals, false positive budgets and shadow mode.
How JSI, Fabric, and Turbo Modules closed the gap, and why on-device AI still pulls some apps back to native. A dev-first 2026 guide.
A technical walkthrough of an unstructured data pipeline for RAG: parsing, chunking, embedding versioning, ACL-aware retrieval and evaluation sets.
An unconstrained tutor gives learners the answer on request. Here are the three constraints that turn a chatbot into something that actually teaches.
Why dashboard queries wreck production databases, and the migration path from live queries to rollup tables without rewriting your whole application.
Treat interruption, bad networks and mid-range Android as design inputs: resumable flows, optimistic updates, permission priming and perf budgets.
Thanks for subscribing.