FROM THE TRENCHES Issue №354

Practical decisions for AI and software

Technical guides to choose models, verify software and plan delivery.

Latest articles

New comparisons, practical guides and engineering lessons.

Reducto for German Invoices: Measure Human Review Effort AI & Agents

Reducto for German Invoices: Measure Human Review Effort

Evaluate Reducto on German invoice line items, decimal commas and totals. Build a correction-time pilot and separate PDF extraction from XML validation.

Cua for Desktop QA: A Browser-to-Native Test Protocol Delivery & QA

Cua for Desktop QA: A Browser-to-Native Test Protocol

Evaluate Cua on a workflow crossing browser, downloads and a native app. Verify file contents, recovery and machine cleanup rather than clicks alone.

AgentMail for SaaS: Tenant Isolation and Duplicate-Send Recovery Delivery & QA

AgentMail for SaaS: Tenant Isolation and Duplicate-Send Recovery

Design AgentMail email workflows with tenant-scoped keys, durable send intents and recovery after lost responses. Includes a proposed failure-test matrix.

Scope vs Armature: Measure Agent API Adoption Beyond Mentions Product & MVP

Scope vs Armature: Measure Agent API Adoption Beyond Mentions

Compare Scope and Armature for agent discovery and API usability. Evaluate installation, authentication and completed tasks with a proposed buying checklist.

Browser Use vs Playwright: Verify Authenticated Actions After Timeouts Delivery & QA

Browser Use vs Playwright: Verify Authenticated Actions After Timeouts

Choose Browser Use or Playwright for authenticated workflows. Handle saved login state, uncertain submissions and retries with independent acceptance checks.

Turquoise branching paths for an application choosing between defined processing routes AI & Agents

OpenAI Decisions API: Confidence, Refusals and Routing

A routing guide for the generally available Decisions API, with per-question refusal handling, probability thresholds and cached-input billing.

Abstract branching paths illustrating session and request routing in Claude Code AI & Agents

Claude Model Router: What Switches the Model, and When?

A model recommendation can leave your active session unchanged. Compare native controls, model-router hooks and provider gateways, then verify what actually runs.

A shield representing reliable software operations as a competitive advantage Delivery & QA

AI Coding and the Software Moat: Who Runs What You Build?

AI makes more custom software worth building. The next decision is who runs it. Compare build, buy and hybrid options through cost, reliability and ownership.

Abstract branching paths illustrating a choice between Clef, Clef-flash and Jev, not benchmark measurements AI & Agents

Cloudflare Clef vs Jev: Pricing, Benchmarks and Migration

Clef brings open weights and vision to the Jev-style decision API. But price, confidence semantics and Flash results make migration less simple than it looks.

Inbox, without the noise

Get the next field note

One concise email when we publish. No tracking pixels, and no inbox filler.

Facing one of these decisions right now? Talk it through with us →