A small practice, on purpose
Neural Studio Lab is new, deliberately small, and takes one build at a time. All three of those are things you should know before you hire us.
Take the free assessmentFirms are experimenting with AI. Nothing is changing.
You've tried the chatbots. You've run a few pilots. Maybe you even have a team member who's good at prompting. But Monday still looks exactly the same.
The problem isn't that you don't know what AI can do. It's that you don't know which workflow to start with, and guessing wrong costs more than the build does. Most firms waste months — sometimes years — on the wrong automation before they find the one that actually moves the needle.
We exist to find that one workflow. The one that's actually worth automating. The one that, once it's running, makes your team wonder how they ever worked without it.
Four rules we actually hold to
One workflow at a time
We don't start a second until the first runs in daily use. Two half-built workflows is the most common way these projects fail.
Fixed scope, fixed price
You accept a spec with the trigger, the steps, the tools, the owner and a testable definition of "working". Changes become a separate quote rather than a slow renegotiation.
Built on tools you already pay for
We work on your stack — your CRM, your email, your automation platform. No new licences, no vendor lock-in, no reselling. We never front tool costs.
You own it after handover
Every build ends with a runbook and a training session for a named person on your side. We're the on-ramp to running this in-house, not a dependency — if you need us forever, we did it wrong.
What we will not do
We say no in writing. The list is on the services page, not buried in a contract.
Regulated data workloads
We don't take workloads whose core is protected health information, cardholder data, or privileged discovery. We can't offer the security posture those deserve.
Model training
We don't build, fine-tune, or train models. We use existing models through APIs and automation platforms, not custom ML engineering.
Strategy-only engagements
We don't deliver decks or recommendations without building. If you need a strategy document rather than something running, we're not the right fit.
Worth being blunt about fit
✓ This works when:
You're an owner-led professional services firm. One person can approve a $3,000 decision without a committee. You have a CRM or some shared system of record — not just inboxes and spreadsheets. And there's someone who will own the workflow after we hand it over.
✗ This doesn't work when:
Your core workload is regulated data — patient records, cardholder data, privileged discovery. You want a model built or fine-tuned. You want a strategy document rather than something running. Or you need it live in under a month — we're a small practice and we can't serve urgency honestly.
If that last list is you, we'll say so on the call instead of taking the engagement. Two of those are things we decline outright — the assessment will tell you which.
Who you'd actually be working with
A one-person practice: you deal with the person doing the work, from the first call to the handover. No account manager, nobody to hand you off to.
James Brown
Founder & Principal Consultant
A small AI implementation practice focused on owner-led firms. Every engagement is handled directly, from the initial assessment through to handover and training. One build at a time means each project gets full attention, which suits clients who value precision over speed.
Show, don't claim
We don't have client logos or case studies yet. Instead, we publish reference designs — full teardowns of the workflows we build most often, labelled and honest.
View reference designs