Code Sorcery
Approach

Six weeks to something running. Not a slide deck.

Most AI engagements spend their first quarter on strategy. We would rather spend it removing your Monday morning.

  1. 01

    Week 1–2

    We follow the work

    We sit with the people doing the task and time it. No workshops, no maturity models, no readiness assessment. We watch the thing happen, write down where it stalls, and count the hours honestly.

    You end up with

    A ranked list of what is worth automating, what is not, and what each would return.

  2. 02

    Week 3–6

    One process, shipped

    We take the top item and build the smallest version that removes real hours. It goes into production on your data, not into a sandbox. You see it working before anyone asks you to commit to a roadmap.

    You end up with

    One process running unattended, with the exceptions routed to a person.

  3. 03

    Ongoing

    We hand you the keys

    Your team gets the code, the documentation and the runbook, in your repositories. We are happy to stay on for the next process, but not because leaving would break something.

    You end up with

    Everything in your accounts, under your control, with no licence to us.

What we believe

Six positions we will not negotiate on.

01

Hours returned, or it did not work

Model accuracy is an input, not a result. The only number we report on is what changed for the people doing the work.

02

The smallest thing that removes real hours

A narrow automation running next month beats a platform arriving next year. We would rather ship four small things than scope one large one.

03

A person owns the exceptions

Nothing we build decides alone at the edges. Low-confidence cases go to a human with the context attached, every time.

04

We will talk you out of it

If a process genuinely needs judgement, or the volume does not justify the build, we say so. It costs us a project and saves you a year.

05

Your stack, your accounts, your data

We build in your cloud, on your keys. No data leaves your tenancy to train anything, ours or anyone else's.

06

No lock-in by design

Plain code in your repositories, documented well enough for your team to change it without calling us.

Start here

The audit is the cheapest way to find out.

Two weeks, fixed fee, credited against the first build. You keep the findings either way.