Is Prompt Caching Actually Lowering Your AI API Bill?
Check whether prompt caching reduces cost per completed task, accounting for cache writes, retries, review effort and the charges on your provider's bill.
AI automation
From Supplier PDFs to Product Data: Where AI Needs a Second Check
Supplier files need more than extraction. Here’s how to check coverage, match SKUs, resolve unclear units and prices, and test product data before a catalogue import.
Not Every AI Job Needs an Instant Answer: Batch the Backlog
Move delay-tolerant AI work into dependable batch queues to cut processing costs without compromising quality, data controls, or urgent workflows.
Background AI Tasks Need Queues, Not Longer API Calls
A practical guide to reliable long-running AI work using job records, queues, idempotency, approvals, webhooks, retries, and recovery.
The Riskiest Part of an AI Workflow Pilot Is Often the OAuth Screen
Before approving an AI workflow, review its OAuth scopes, identities, spending controls and shutdown plan with this practical pilot checklist.
Agentic AI: What It Is, How It Works, and When to Use It
A practical guide to agentic AI: how it differs from chatbots and automation, where it creates value, and how to pilot it without losing control.
OpenAI Evals: Acceptance Tests for AI Workflow Releases
Learn how to use OpenAI evals, graders, red-team cases, and release gates before prompt, model, tool, or routing changes go live.
OpenAI's Guardrails and Run State Make Internal Agent Rollouts a Paid Approval-and-Audit Job
OpenAI's agent documentation points to a practical reality for internal automation: once an agent can update records or trigger actions, the valuable work shifts to approval design, run-state logging, observability, and staged rollout governance.
OpenAI Computer Use: Browser Agents Need Credentials, Not Demos
A practical guide to taking OpenAI computer use beyond the demo with safer credentials, approval gates, isolated browsers, and recoverable operations.
OpenAI File Search: Make Internal Docs Worth Trusting
Make OpenAI File Search reliable for internal knowledge: organise documents, enforce access, test citations, and manage updates, retention and running costs.
Written recommendations from Trafik og Veje, Aarhus Municipality (2011) and AgroTech (2010).
