AI & Automation

AI tools wired into work you already do.

A tested assistant or content workflow, or a RAG build on documents you supply. I review and test everything before delivery, and I’ll follow any AI-use rules you set.

Who this is for

This is for a small, named integration — not an unattended enterprise agent platform. There is no public client AI case study on this site yet.

A content or support workflow

Drafting, summarising, or answering from material you already have, with you still reviewing the output.

A site or tool you already run

I integrate into a process or page you own, rather than inventing a new product around the model.

Your rules for AI use

I’ll follow any limits you set on what the model may see or say.

What I don’t claim

I don’t invent client results, accuracy percentages, or “autonomous” outcomes I can’t show.

Packages

Prices are starting points for the listed scope. I’ll confirm the quote before work begins.

Starter

$499 starting at · 7 days · 1 revision

  • One assistant or content workflow
  • Wired to one tool you already use
  • Reviewed and tested before delivery
Request Starter

Custom

$1,500 starting at · quoted · scoped before work starts

  • Larger LLM or agent work
  • Written scope and a quoted turnaround
  • Human review before anything leaves your process, if you want that gate
Request a quote

How the work runs

  1. Describe the job

    Tell me the workflow, the tool or site, and any AI-use rules I should follow.

  2. I confirm the quote

    You get the scope, price, and delivery date in writing.

  3. I build and review

    I wire the integration and test it against the cases we agreed.

  4. You keep the keys

    Model and API credentials stay yours. I don’t keep them after delivery.

Samples

There is no public client AI case study on this site yet. Related studio tooling — a job-and-approval API and a delivery pipeline with a secret scan — is described on the work page as internal tooling, not as a client AI product.

FAQ

Do you use LangChain or LangGraph?

Yes, when they fit the job. I’ll name the stack in the quote.

Will the bot run without my review?

Only if you ask for that and we write it into the scope. I default to reviewed, tested delivery.

Do you train a custom model?

Not in these packages. I work with APIs and your documents. Fine-tuning or hosting a model is a separate quote.

Can you guarantee accuracy?

No. I’ll test the cases we agree, and I’ll tell you what the system still gets wrong.