You're already using AI. You just don't know if you're doing it right.
One-to-one mentoring for business owners who have been figuring out AI on instinct. You have built things that worked once, then broke, and you cannot reproduce them. We fix the foundation, then you build on it properly. Every system already tested in a real business every week.
Sound Painfully Familiar?
The things every self-taught AI user eventually runs into.
Four Tools and No Map
ChatGPT for writing. Claude for anything technical. Something for scraping you set up months ago. A workflow tool you have not opened since. Each one holds a piece of how you work. None of them share it, and none of it is written down.
Compliance Exposure You Can't Quantify
EU AI Act and GDPR obligations are real, and they apply to a business of six exactly as they apply to a business of six hundred. Whatever you put in place has to be something you will actually use.
No Methodology At All
AI results are inconsistent, quality varies wildly, and there's no way to scale what's working.
Speed Without Strategy
Blog posts, outreach, SEO, events, a new tool you read about last week. More is getting done, but it is drifting from what your customers actually respond to. Faster isn't better if it's amplifying the wrong message.
Your Competitors Are Compounding
People with a system build capability that compounds. Without a system, your productivity gains stay individual and temporary.
Built It Once, Lost It Twice
You solved something clever six months ago and it worked. You cannot find it now, or it broke and you do not know why, and you have quietly rebuilt the same thing three times in three different tools. That is not an AI problem. It is an architecture problem, and better prompting will not fix it.
Not sure where you are? The AI Readiness Assessment takes 4 minutes →
The frame we build toward
Which of your activities require a human to be physically present? Everything else is agent-ready. That's what we build toward.
What we build together
Proof that I build what I teach. Every client engagement uses these same tools and patterns.
Built by Duncan
Call Trainer
A digital twin of your prospect, built from scraped data. Practise before the real meeting. A live coaching layer runs during the call, updating deal stage, talking points, and objection handling every turn.
Built by Duncan
Practitioner Hub
The operating system I run engagements from. Governance audits, maturity assessments, programme designs, ROI reports. Auth-gated.
Built by Duncan
Prospect Intelligence Platform
Syncs with your CRM. Dual-AI research returns a confidence-scored dossier per account: company overview, use-case matching, discovery questions, and software signals. Territory-scale intelligence without a research team.
Ask about any of these when we speak.
Methodology
The framework behind every engagement
Prompt → Refine → Deliver → Repeat is how we build AI systems together, not how I build them for you.
PROMPT
Design prompt systems grounded in your customer's actual language, extracted from sales calls, support tickets, and reviews. No generic AI outputs.
REFINE
Test, iterate, and tighten every workflow with you until outputs consistently match your brand and your audience's expectations.
DELIVER
Deploy governed, documented AI systems into your real day-to-day. Not theory, not demos. Working tools you actually use.
REPEAT
Build repeatable systems that compound over time. You own the capability. Productivity gains don't walk out the door.
The published methodology
Read Prompt. Refine. Deliver. Repeat.Meet Duncan
A Practitioner. Not a Consultant.
I'm Duncan Hendy, an AI mentor and practitioner, and author of Prompt. Refine. Deliver. Repeat. The systems I teach are the systems I run.
I ship AI tools daily, including the systems I bring to every client engagement. This isn't advice I read about. It's what I do in production, every week.
The difference: I build alongside you, not for you. When the engagement ends, the capability stays with you. You own it, run it, and extend it.
The Playbook
Get the Playbook
The second edition of the published methodology behind every Duncan Hendy AI Mentoring engagement. 29 chapters of practical AI workflows for B2B product marketers, built on Claude. No fluff, no theory. The prompts, patterns, and systems that work.
Prefer a physical copy? Get the book on Amazon for €11.99
Frequently Asked
The questions people ask
before they engage.
Not at all, and it usually means you will move faster than someone starting cold. The problem is not that you have been experimenting. It is that experiments accumulate. Tools that do not connect, prompts you cannot find, workflows nobody documented, and data going places you have not audited. The first job is always to see clearly what you have actually got, before building anything new on top of it.
There is no fixed length, because there is no fixed programme. Most people start with a couple of sessions to define the problem properly and agree a sequence, then the pace is theirs. Some work intensively for a few weeks. Others go quiet for a month while they build, then come back when they hit something. Both are fine.
Training tells you what AI can do. Mentoring builds systems you actually use. We build together using real tools on real problems, so the capability stays with you when the engagement ends. No dependency on continued consultancy. You own everything.
The people I work with tend to be running their own business, expert in what they do, and self-taught with AI. They have got real results already and they have also got a mess they cannot untangle. Size does not matter much; one person or twenty makes little difference to how we would work. What matters is that you want to build the thing yourself rather than have someone build it for you and leave. If you would rather hand it over and receive a finished product, I am the wrong person and I will say so early.