Skip to main content

Our Thinking

Notes from the build

Practical takes on software development, AI integration, and shipping products that hold up.

Page 5 of 9

Product 2026-08-23

The Full-Time CTO Job Description We Talked Three Founders Out Of: A Line-by-Line Duty Audit

Three founders came to us with a drafted full-time CTO posting. We scored each responsibility bullet by real hours consumed, whether it needed continuity or judgement, and which ones a fractional arrangement covers. Plus the two roles we told them to hire full-time instead.

Read Article
AI 2026-08-22

Shipping AI Features Without an ML Engineer: The Four Ownership Roles We Assign Instead

None of our three clients running AI features had an ML hire. Here is how we split model behaviour into four named roles (Prompt Owner, Eval Owner, Cost Owner, Fallback Owner) and handed them to existing product and backend staff.

Read Article
Product 2026-08-21

Scoping a Custom Software Project With an Agency: The Five Discovery Artifacts We Hand Over Before a Contract Is Signed

The exact documents a software agency should produce during pre-contract discovery: entity/event model, integration inventory, must-ship-first list, non-goals, and assumptions register. Plus four real scope traps that cost us money.

Read Article
AI 2026-08-20

The AI Retrofit Log: Six Features We Added to Products Already in Production (and the Two We Turned Off)

A feature-by-feature account of retrofitting AI into live products: where the call was inserted, latency budgets, cost per call, fallbacks, human review gates, and the two features we shipped and then removed.

Read Article
Engineering 2026-08-19

The Back Office Work We Deleted: A Task-by-Task Automation Audit from MOVE & MEET and Plodovi

A concrete inventory of the recurring manual admin tasks two small operations ran every week, which ones we automated, which ones we left alone on purpose, and the three automations we reverted.

Read Article
AI 2026-08-18

OpenAI vs. Anthropic vs. a Local Model: The Selection Matrix We Ran for Three Client Features

How we chose between OpenAI, Anthropic, and a self-hosted model for three production features: consent-bound health data, marketplace product copy at scale, and support triage. The constraints that actually decided it, and what broke when we switched providers.

Read Article

Let's Build Together

Your vision,
our expertise.

From AI integration to full-stack development, we turn ambitious ideas into products that perform.