I start with hours, not tools.
Most automation projects begin with a product decision and work backwards to a problem. I do it the other way round. Before anything is built I sit with the people doing the work and count where the hours actually go — not where a process map says they go.
That count is what tells me whether a workflow is worth automating. Volume, variability, how clear the rules are, and what it costs when the output is wrong. High volume with clear rules is worth building. Low volume with heavy judgement usually is not, and I will say so.
Then I build in short cycles against one workflow at a time, inside your environment and under your data-handling rules. Nothing goes live until the people who use it agree it is better than what they had.
Where it lands
Four places this usually pays off
- Process automationRepetitive internal workflows, replaced end to end.
- Records and documentsExtraction, classification, and routing at volume.
- Readiness reviewsWhere AI will pay off, and where it will not.
- IntegrationsYour existing systems, talking to each other properly.
Before you ask
Common questions
Which tools and platforms do you build on?
Whatever fits your environment and your constraints, rather than whatever I have a partnership with. In practice that usually means your existing stack plus a small amount of glue. If a workflow can be solved with a rule and a scheduled job, I will not put a language model in front of it — models are expensive, non-deterministic, and harder for your team to reason about later.
Our systems are old and have no API. Is that a blocker?
Usually not, but it changes the shape of the build and I will say so in the audit rather than discovering it halfway through. Plenty of useful automation sits around a system rather than inside it — at the intake, at the handoff between teams, or in the reporting that comes out the other end. If the only viable approach would be brittle, I will tell you that too.
Read the full answerBring me the process that wastes the most time.
A first call takes thirty minutes. You will leave it knowing whether automation is worth doing at all in your case.