Blog.

The harness is not the model. What you actually pick when you install an agentic CLI

Who really welds the harness to the provider, what the Claude Agent SDK actually locks, and the five layers a model swap rewrites.

AI Futures: OpenAI asks who holds power when AI stops needing people

A good argument, the same one Anthropic published seven months earlier, and one citation that leads back to OpenAI.

The AI GTM stack: five stages you can buy, four handoffs you define

The four handoffs between funnel stages that no vendor sells, what an agent can run alone, and where a person still has to stand.

A public report written with AI: nine errors you can see in the file

A number that contradicts its own footnote, one methodology in three versions, and nine checks that stop all of it.

Mutation testing: can you trust AI-generated tests?

Coverage measures execution, mutation score measures assertion. How mutation works, what it found in production code, and how to run it in CI.

AI training for a non-technical team: how to run it

One task instead of twenty tools, the sceptic against the blocker, and six failures from outside the programme. Built on a session for the TVP Television Academy.

What an AI agent costs and what actually drives the quote

Six line items a quote is made of, two of which only show up after launch. How to count a deployment in days of work instead of licences.

Prompt injection: an instruction hidden in content the agent fetches itself

How prompt injection works, five risks specific to agents, and a checklist your team can run before the agent reaches production.

AI agents on a team: what training gives you and who is ready

Agents are one of the last steps of AI adoption, not the first. What a team learns in two days, what it will not learn, and who should skip it.

Workflow Studio: observe and design Claude Code workflows

A local dashboard that reads Claude Code's own run artifacts, plus a 12-block canvas that compiles to a runnable Workflow script.

Claude Certified Architect certification: the complete guide

What the CCA Foundations exam tests, the five domains and weights, how it works, and how I prepared and passed 833/1000.

Claude Certified Architect practice exam: 60 free questions

60 original questions I wrote while preparing for CCA Foundations. Study mode, exam mode and per-domain scoring.

I passed Claude Certified Architect: decisions, not definitions

The proctoring step by step, why the exam tests decisions not definitions, and how to prepare.

12 rules for building agents

Notes from prepping for the Anthropic certification. The rules that decide whether an agent survives production.

Claude Partner Badge — Claude Code: what it is

Anthropic partner badge: 8 courses, a capstone, and real rollout from demo to production.

awesome-ai-gtm: a map of the autonomous GTM funnel

138 AI tools arranged into 7 funnel stages. Not a link list, but a blueprint.

Model self-evaluation: the agent that grades itself

The most common agentic engineering pitfall — and the fix.

Multi-agent consensus isn’t verification

Three models agree and fail the same way. What to do instead of voting.

RATCHET: iterating against a frozen yardstick

Keep-if-better, else revert — a pattern you can’t game.

Silent failures in agentic systems

Silent retry and fallback return “OK” and hide the errors.

Position bias: the first answer wins on order

An adversarial panel with rotation instead of a single judge.

A confidence score isn’t correctness

Confidence is a tone of voice. Validate it from the outside.

Bounded Autonomy: How much freedom to give an agent?

A design pattern for safe AI agent autonomy.

5 questions before deploying AI

A checklist that will save you months and thousands.

AI Governance: How not to lose a million

A governance framework that works in practice.

Multi-Agent Systems: When one agent isn't enough

Orchestrating multiple agents and communication patterns.

Hybrid AI: Combining models for better results

When and how to combine different AI models in one system.

AI in logistics: A deployment case study

How we built an agent for route optimization.

From pilot to scale: Lessons learned

What changes when an agent hits production.

AI Stack for your company: My tool choices

The stack I use in every agent project.

Agentic AI in 2026: From experiments to production

62% of companies are experimenting with AI agents. Only 11% have them in production.

Why AI strategies end up in a drawer

Strategy documents vs. real change in the organization.

ROI from AI: How to measure it without fooling yourself

The metrics that actually show the value of a deployment.

Legacy Systems vs AI: How to bridge old and new

Integrating AI agents with existing infrastructure.

6 roles of a CTO in the AI era

How the role of the tech leader is changing.

AI Ethics Lead: A new role on the team

Why every AI team needs someone owning ethics.

Workforce 2026: AI and the future of work

How to prepare your team to work alongside AI agents.