AI Agents
New

AI that does the
work, not the demo.

The interesting part was never the chatbot. It is what happens when research, qualification, drafting and follow-up stop being the bottleneck — and a two-person team starts shipping like a ten-person one. I build agents that do a specific job properly, not ones that do everything badly.

Book a discovery call See the work
What you get

The part that
actually matters.

  • Qualify before you call

    An agent reads every inbound enquiry, enriches it, scores it against your criteria and routes it — so you spend time on the real ones.

  • Draft at volume

    Proposals, replies and campaign copy drafted in your voice from your own material, then edited by a person before anything goes out.

  • Research on demand

    Market, competitor and prospect research synthesised into a brief you can act on the same day, not next week.

  • Wired into your stack

    Agents that read and write to the systems you already run — CRM, inbox, calendar, sheets — rather than another tab to check.

Deliverables

In the box.

  • Workshop to find the tasks genuinely worth automating
  • Agent built and tuned against your real data
  • Integration with your CRM, inbox or database
  • Human review step on anything client-facing
  • Guardrails, logging and a documented failure path
  • Team training on supervising it
  • 30 days of tuning after go-live

Best suited to

Teams drowning in repeatable knowledge work — qualification, drafting, research, triage. Not a fit for anything where a wrong answer is unrecoverable and nobody will be checking the output.

Built with
ClaudeChatGPTGrokAPIs & WebhooksCRM IntegrationWorkersAutomation
Process

How working together goes.

No mystery, no month-long silences. Five steps, and you can stop after any of them.

  1. Discovery call

    20 minutes, free

    We work out what you are actually trying to fix, what you have already tried, and whether I am the right person for it. If I am not, I will say so and point you somewhere better.

    You get: A clear read on whether this is worth doing

  2. Scope and proposal

    Within 3 days

    You get a written scope: what is included, what is explicitly not, the timeline, and a fixed price. No hourly surprises, and nothing starts until you have agreed to it.

    You get: A fixed-price proposal you can say no to

  3. Build in the open

    The bulk of the project

    A staging link is shared from the first week and updated as work lands. You see progress as it happens instead of waiting a month to find out we understood each other differently.

    You get: A live staging URL and weekly check-ins

  4. Review and launch

    Final week

    You go through it properly against the original scope, we fix what needs fixing, and then it goes live — with analytics connected from the first day, not bolted on later.

    You get: Live, measured, and handed over with training

  5. Support and iterate

    30 days, then optional

    Thirty days of support is included for anything broken or confusing. After that you either run it yourself with the training, or keep me on for ongoing work. No lock-in either way.

    You get: A system you own and can actually run

Questions

Straight
answers.

Will it make things up?

Any language model can. That is exactly why everything client-facing goes through a human review step, and why agents are built against your own material rather than left to improvise. AI sets the pace; it does not sign off on the work.

Is my data used to train models?

Not on the setups I build. Business API tiers do not train on your inputs by default, and I will show you the setting rather than ask you to take my word for it.

What does it cost to run?

Usage-based, and usually smaller than people expect — most agents cost a few dollars a month in tokens. I will estimate it from your actual volume before you commit.