AI Business Automation.
The repetitive work your team does every week, handed to software. Reading inbound email, pulling data out of documents, drafting the same reply, moving records between tools. I find the workflows worth automating, build them, and hand you something your non-technical staff can actually run.
Six workflows worth handing to software.
These are the ones that show up in almost every business I audit. Yours will look like one or two of them, and the audit is how we find out which.
Inbound email triage
Reads what arrives, classifies it, drafts the reply, and routes anything unusual to a person.
Document data entry
Invoices, forms, and PDFs turned into structured rows in your database, with a confidence score on every field.
Report assembly
The weekly number-gathering somebody currently does by hand, run on a schedule and delivered to Slack.
Record sync between tools
The copy-and-paste between your CRM, spreadsheet, and billing system, done on a trigger instead of a Tuesday.
Meeting and call notes
Transcripts turned into decisions, owners, and follow-ups that land where the work actually happens.
Review and feedback digests
A pile of free text reduced to themes and counts, so the pattern is visible without reading all of it.
- AutomateHigh frequency, low judgment, currently done in a spreadsheet.
- AssistThe draft is automated, a person approves before it leaves.
- Leave itRare, high-stakes, or so nuanced the review costs more than the task.
What's included.
- +A workflow audit: what your team repeats, how long it takes, what it costs
- +A shortlist of what to automate and, just as importantly, what to leave alone
- +Document, email, and form processing with AI where it earns its place
- +Integrations with the tools you already run: email, CRM, Sheets, Slack, your database
- +Human review steps on anything with consequences, so nothing ships unchecked
- +Plain-language dashboards and runbooks so your team owns it after handover
What you can expect.
- ▶Hours back each week on work nobody wanted to be doing
- ▶A cost per run you can see, budgeted against the salary hours it replaces
- ▶Automations your ops team can trigger and inspect without an engineer
- ▶A clear audit trail: what ran, what it decided, and who approved it
How an engagement runs.
We sit with the people doing the work and time it. The best candidates are high-frequency, low-judgment, and currently done in a spreadsheet.
One workflow, end to end, running against real data with a human approving every result. We measure it against the manual version before scaling.
Error handling, retries, escalation to a person, and a kill switch. An automation that fails loudly beats one that fails quietly.
Documentation, training, and dashboards. If your team needs me to run it, I built the wrong thing.
Best for.
- ◇Teams with someone spending hours a week on copy, paste, and re-typing
- ◇Operations that live in email and spreadsheets and have outgrown both
- ◇Businesses that want AI doing something useful rather than something impressive
Engagement & pricing.
The audit is fixed-fee and stands on its own. Most single workflows are 2 to 4 weeks to build and hand over.
- Claude API
- Next.js
- PostgreSQL
- Vercel
- Slack / Gmail / Sheets APIs
- Custom dashboards
Recent projects in this lane.
Ready to start?
Send a one-paragraph brief.
What you're building, the rough timeline, and one constraint that matters. I read every message and reply within one business day.