Skip to content
MAVERICK Labs Contact

How we work

Audit, build, hand over. Each step is its own decision.

You can stop after any step and keep what you have. The aim is a team that can run and change what was built, not a dependency on us.

In short

Fixed scope and fee agreed before each step. Review points written down. Your information stays behind permissions you set.

The engagement

What happens, in order.

Engagements follow this path, and you can stop after any step. Stopping after the audit is a perfectly good outcome if the report shows a build isn't worth it yet.

  1. Audit
    Map one process and agree a first scope

    A Workflow Audit with the people who do the work. You receive a report you keep, whatever you decide next.

  2. Agree
    Fixed scope, fixed fee, written down

    If you want us to build, we agree exactly what will be built, what it will and won't access, where a person approves, and how we will both know it works.

  3. Build
    Build and test on your real process

    We build in small pieces and test against examples you choose. Nothing goes live without your sign-off.

  4. Hand over
    Train the people who will run it

    Training, a plain-language runbook and a walkthrough of how to change the rules. You own the configuration and the documentation.

  5. Support
    Optional, and easy to leave

    Help as the process changes. If you'd rather run it entirely yourselves, the handover is designed for that.

Your information

Agreed before access, limited to the task.

We agree what information a workflow can use, and where it is processed, before we touch any of it.

Scope before accessWe list the folders, inboxes or systems a workflow needs, and nothing more. Access uses your accounts and permissions, so you can remove it at any time.
Where the AI runsEither an approved AI provider under agreed settings, or a model on hardware you control when information has to stay in-house. We explain the trade-off in plain terms. Private AI, explained.
People approve what mattersIn the workflows we design, a step that sends, pays or commits on your behalf waits for a person's approval before it happens, not a review after.
Records you can checkWorkflows keep a record of what they did and why, so a result can be traced back to its source.

Common questions

Straight answers.

If your question isn't here, ask it.

What does it cost?

Each step has a fixed fee agreed in writing before it starts. The audit is quoted after a short call; a build is quoted from the audit's scope. Ongoing support, if you want it, is a separate monthly arrangement.

What happens to my team's jobs?

That is not what we build. We take the repetitive, procedural parts of a process and leave the judgement, relationships and decisions with people. The aim is to free time for work your team is already short of hands for. More on AI and your team.

Is our data safe?

We limit access to what a workflow needs, use your own accounts and permissions, and agree where processing happens before we start. For information that must not leave the business, we set up models on hardware you control. We don't use your information for anything other than your engagement.

Are we locked in to you?

No. You own the configuration and the documentation, the audit report is written so anyone could build from it, and the handover is designed so your team can run things without us.

Which AI models do you use?

Whichever suits the task and your information boundary: an approved commercial model, or an open model running locally. We avoid designs that only work with one provider.

What if the automation gets something wrong?

That is why review points exist. Workflows flag what they are unsure about rather than guessing, and anything consequential waits for a person. Records make it possible to see what happened and fix the rule.

How long does a build take?

It depends on the scope, which is why the audit comes first. The build quote includes the plan and the points where you'll see working pieces.

Next step

Start with one process.

Tell us what comes in, what should come out and where it gets stuck.

Describe one workflow
Signal 01 / If the tool adds friction, it is not finished.