Service
Ongoing Support for Your AI Employees
An AI employee should not be installed and forgotten. I keep it reliable, useful and aligned with your business as your people, processes and tools change.
Useful systems need looking after
Your prices change. Your team learns a better way to handle something. A supplier updates its software. An AI employee needs to move with the business rather than slowly becoming another workaround.
Ongoing support covers the practical work: checking what happened, improving the instructions, handling unusual cases and keeping connections working when other platforms change.
You are never locked in. The point is to keep the system useful and dependable, not to make you reliant on a mystery box nobody else understands.
What ongoing support includes
Monitoring
Check that work is completing properly and that anything unusual is being surfaced rather than silently missed.
Small improvements
Refine wording, rules and handovers as the team learns what makes the system more useful.
Changing processes
Update the AI employee when your prices, services, responsibilities or internal processes change.
Software changes
Keep integrations healthy when email, accounting, booking or customer platforms update their systems.
Team feedback
Use feedback from the people doing the work to remove friction and make the next version better.
Clear reporting
Give you a straightforward view of what the system handled, what needed help and where time was saved.
New responsibilities
Add the next sensible task only after the existing job is working reliably and earning its place.
Human control
Keep approvals, permissions and escalation points aligned with the level of control your team wants.
A real example: Dan OS — Email Drafting
Dan OS — Email Drafting is an internal capability I built and run myself. With explicit Google authorisation it reads my incoming mail, identifies invoices and extracts their details for review, and prepares draft replies to customer emails — draft-only, so nothing sends without my approval. That includes drafting from the correct business address, which required getting the Gmail OAuth configuration genuinely right rather than approximately right.
It's a useful example because it shows the shape of this work honestly. The interesting parts weren't the AI. They were the permission scopes, the review step, the handling of edge cases, and making sure a draft always lands somewhere a human sees it before a customer does.
I've written up the Gmail authentication work in full, including the part the documentation glosses over.
How I keep AI employees trustworthy
- Grounded in your own documents and data, not general guesswork
- Narrow, explicit permissions on every system it can touch
- A human review step wherever the output reaches a customer
- Clear escalation rules for anything outside its remit
- Logging of what was asked, retrieved and produced
- An off switch that doesn't require a developer
On performance claims
I deliberately don't publish percentage figures for time saved or enquiries handled. Those numbers vary enormously between businesses, and quoting someone else's would tell you nothing useful about yours. During discovery I'll estimate the realistic saving for your specific volumes, and I'd rather that estimate be conservative and correct.
FAQ