What actually happens when you hire us
No black box. This page walks through each stage of a project: what we do, what you get at the end of it, and where you stay in control.
The same question, asked twice
A generic chatbot answers from whatever it absorbed in training. The systems we build search your own documents first, then answer from what they found. Here's the difference, side by side.
Example · sample data.
Confident tone, no source, and it never saw your policy. If it doesn't know, it guesses.
Cites the document it read. If nothing matches, it says "no record found" instead of guessing, and staff can only query files they're permitted to open.
Screens your staff actually use
A chat window on its own doesn't change how a team works. We put the AI behind ordinary tools, so using it is just part of the job.
Dashboards
The numbers your team checks daily, pulled live from your systems instead of assembled by hand each week.
Approval queues
Anything the AI drafts waits here for a person to approve, edit or reject before it goes anywhere.
Internal apps
Small, purpose-built tools for specific jobs (quoting, intake, scheduling), with the AI handling the tedious part.
Reporting
Reports that assemble themselves from live data, so the end-of-month scramble goes away.
One invoice's journey
An illustrative example of how a project runs. The details differ for every business, but the shape of the work doesn't.
We sit with the person who handles supplier invoices and write down every step - where invoices arrive, what gets re-typed, who approves what.
- · Arrives by email, sometimes paper
- · Re-typed into accounting software
- · Chased for approval in person
- · 9 touches counted, 3 re-typing points
Invoices land in one queue. The system reads each one, pulls out the details, and drafts the entry in your accounting software.
Nothing posts automatically. Staff see the drafted entry with the original invoice alongside, and approve, fix or reject it.
Your team runs it day to day. Code, accounts and documentation are handed over, and we're a phone call away when a supplier changes their invoice format.
- Accounts transferred to you
- Documentation walk-through done
- Team trained on the queue
A typical project, week by week
Stages overlap on purpose: the blueprint starts while we're still mapping, and handover begins before the build ends. Timelines are typical and agreed per project in the blueprint.
Map the work
We sit with your team and trace each workflow end to end.
Blueprint
A written plan with a fixed price. If the numbers don't justify building, it says so.
Build
On top of your existing systems, with working software demoed every week.
Handover
Training, documentation, accounts and code, all transferred to you.
After the blueprint you can stop. It's yours either way. Support afterwards is optional and month to month.
Map the work
We sit with your team and trace each workflow end to end.
Blueprint
A written plan with a fixed price. If the numbers don't justify building, it says so.
Build
On top of your existing systems, with working software demoed every week.
Handover
Training, documentation, accounts and code, all transferred to you.
Your data stays yours
Grounding AI in your business data only works if that data is handled properly. These are the questions owners ask us, with the answers we put in writing on every build.
- Is your data used to train AI models?
- No. Your documents are only read at question time to find answers. Nothing you store is used to train anyone's models.
- Who can see what?
- Access follows your existing staff permissions: people can only query data they'd be allowed to open by hand.
- Where does it all run?
- In cloud accounts you own. Nothing we build depends on infrastructure we control, so nothing breaks if we're not around.
- What gets logged?
- Every query and automated action, so you can always show who asked what, and when. That's useful for both audits and disputes.
- What happens if you stop working with us?
- The systems keep running. Everything already lives in your accounts, and the documentation is written for your next hire, not for us.
A first call is free and commits you to nothing. If we don't think a build stacks up, we'll say so.