SERVICES

I build AI agents that do real work for a business

CAT-01 and VoiceCall are the working proof. Start with a ready-built lead agent live in 7 days, from $299, or ask me to build a different one. You get a clear number after a 30-minute call.

Book a scoping call See the ready-built agent

Ready-built agents

A fixed scope at a fixed price. It runs on your own n8n and you own the workflow. It is not a SaaS subscription.

Custom Speed-to-Lead Deployment

A bespoke, self-hosted lead qualification engine live on your stack 7 days after you give full access, then 14 days of fixes.

Timeline: 7 days from full access

Founding pilot$299First 3 clients only. You give a review, case-study rights and before and after numbers.
Standard build$1,500One time. A second channel or a CRM other than Twenty adds $500 each.
Care plan$149 / monthOptional. Monitoring, fixes, a monthly AI cost report and one small change a month. AI usage: use your own OpenRouter key, or I pass it on at cost (about $1.50 per 1,000 leads). You pay your own hosting.

1 lead source: a web form or an email inbox
An instant reply written from your approved fact sheet
Each lead scored on your criteria
An owner alert on Telegram, WhatsApp or Slack
The lead written to Twenty or Google Sheets
Runs on your own n8n, and you own the workflow
14 days of fixes after launch

Not included voice calls, ad spend and custom apps. Those are custom work.

See it live Book a call

Need a different agent? I build it

Three ways to work with me when the ready-built agent is not the right fit.

Architecture Sprint

A blueprint your team can build from, in weeks not months.

Timeline: 2-4 weeks
Price: Scoped on a call

What changes the cost

How many systems and teams the design must cover
How much existing code and data I must read first
How many review rounds your team needs

Technical architecture document
Agent topology diagram
Model selection rationale
Infrastructure & cost projection
Implementation roadmap

Best for teams with engineering capacity who need architectural direction

Discuss Architecture Sprint

Full-Stack Build

A working, monitored AI system in production, built end to end.

Timeline: 6-12 weeks
Price: Scoped on a call

What changes the cost

How many agents and tools the system needs
How many systems it connects to
How much monitoring and handover you need

Production-deployed AI system
Full source code with documentation
Monitoring & observability setup
Runbook for operations team
30-day post-launch support

Best for organizations that need a production system, fast

Discuss Full-Stack Build

Advisory Retainer

A fractional AI architect on call while your team builds.

Timeline: Ongoing monthly
Price: Scoped on a call

What changes the cost

How many hours of calls each week
How fast you need answers
How many teams I advise

Weekly architecture review sessions
Async Slack/email access
Code & design review
Vendor & tool evaluation
Quarterly strategy alignment

Best for teams building AI capabilities in-house who need senior guidance

Discuss Advisory Retainer

More agents I build on request

Each one is scoped and priced on a call, and built into your setup. None is a live product yet.

CAT-02 Built on request

Voice receptionist and missed-call text-back

Customers ring a small home-service firm, nobody picks up because the team is on a job, and the caller rings the next firm.

A call nobody answers in 20 seconds gets a text from the firm within 30 seconds.
The caller's reply is qualified and answered from the firm's own facts sheet, and the owner gets Send, Edit and Stop links.
A second tier adds an AI voice that says it is an AI assistant, answers from the same sheet, and transfers to a person on request.

Missed-call text-back from $299
Voice receptionist from $1,500
Care from $149 a month plus call minutes at cost

US and UK numbers only. This agent never makes outbound marketing calls.

Book a call

CAT-03 Built on request

Support chatbot on your own documents

A small business answers the same customer questions every day, and a generic chatbot invents answers it never gave.

The bot finds the matching parts of your pages and PDFs and answers only from them.
Every answer names the page it came from, and a second model removes sentences the page does not support.
When the answer is weak, the bot says a person will reply and passes the visitor to you as a lead.

One website and one document set from $499
With CRM hand-off from $1,500
Care from $149 a month

Book a call

CAT-04 Built on request

Invoice and document extraction

A small firm types supplier invoices into its accounting tool by hand, and a typing mistake puts a wrong amount in the books.

The agent reads each invoice PDF and fills the fields against a fixed schema.
Code checks the arithmetic and a second model checks the fields against the page.
It writes a draft bill only, never a final one, and sends anything doubtful to a review queue.

One document type to a sheet from $499
To QuickBooks or Xero with a review queue from $1,200
Care from $99 a month plus AI cost per page

Scope: supplier invoices in English, up to 500 a month.

Book a call

CAT-05 Built on request

Outreach first-line writer

Cold emails read like templates, and a first line that sounds personal but is wrong does more harm than a template.

The agent reads each company's own home and about pages and drafts one first line naming one true thing.
A second model confirms the fact is on the page, and you review every line in a sheet.
It never sends email. Your own sending tool sends, so unsubscribe and sending limits stay with you.

From $499 for the build
Care from $99 a month

B2B only. The run refuses a list with no stated source.

Book a call

How it starts

30-minute scoping call
You walk me through the problem, the team, and the constraints.
A clear number
You get the model that fits, its scope, and its price.
Work begins
You decide whether to go ahead. The call itself commits you to nothing.

Questions buyers ask

What does a project cost?

Each custom model below shows its budget level and what changes its cost. After a 30-minute scoping call you get one clear, written number for your scope before any work starts. If the scope changes later, we re-scope and I re-price before the change is built.

What is in the Speed-to-Lead Deployment?

The CAT-01 lead engine, custom-built for your sales pipeline: intake from web forms and WhatsApp; qualification against your BANT criteria; fixed rules for brand, pricing and safety; qualified leads written to your CRM (Twenty today; other CRMs and email-inbox intake on request); and instant Telegram alerts to your team.

What does the Speed-to-Lead Deployment cost?

The founding pilot costs $299 for the first 3 clients, who give a review and case-study rights in return. The standard build costs $1,500 one time. A second channel or a CRM other than Twenty adds $500 each. An optional care plan costs $149 a month. You pay your own LLM and hosting bills.

How long does it take?

The Speed-to-Lead Deployment goes live 7 days after you give full access, and then you get 14 days of fixes. An Architecture Sprint takes 2–4 weeks and a Full-Stack Build 6–12 weeks. The Advisory Retainer runs month to month.

Do I own what you build?

Yes. The Speed-to-Lead Deployment and the Full-Stack Build hand over the full source code, so there are no per-seat SaaS fees. An Architecture Sprint hands over the blueprint your team builds from.

What happens after the call?

You get the model that fits, its scope and its price. Then you decide whether to go ahead; the call itself commits you to nothing.

I already have a development team. Can you work with them?

Yes. I design and build the AI layer alongside your engineers, not instead of them. Teams with their own capacity usually pick the Architecture Sprint or the Advisory Retainer.

Not sure which fits? Tell me the problem and I will recommend one.