Every job starts with a free, read-only look at the systems you already use. If we do paid work after that, the scope comes from the report.
You can walk away after this. We review how the business actually runs in the tools you already pay for, and we write it down.
Access stays read-only. We don't change a record, move a folder, or download your data. If a step would need more than that, we stop and ask.
You get a write-up of what we reviewed and what we'd do next. That's the whole of this phase. You're not on the hook for the follow-on work.
People describe how work is supposed to move. The files show how it actually moves. We start with the files, because that's where the gaps are.
If you continue, the paid work is based on what we found, not a generic package. That keeps the project from growing past what the data can support.
This comes after the report. We don't resell software, and we don't start with an agent.
One record per thing, with dates and history, and shared IDs across the systems that already run the business. Most AI projects skip this and stall later.
Workflows on the stack you already have. Invoice matching, document assembly, intake, compliance tracking. Built around how the office already works.
Read-only means read-only. If we ever need a working copy later, that's a separate conversation in writing.
No software markup, and no migration for its own sake. The point is to make the tools you already own reliable enough for AI.
Access is for your project. We don't put client records or findings from them on this site as a case study.
If the data can't support it yet, we say so. Better to stop than to run a pilot that was going to fail for a reason we could see on day one.