SUPPORT & CARE · BUSINESS HOURS · OURS OR SOMEONE ELSE'S
We watch your AI systems, fix what breaks, and make them better every month. Support is shaped around each project, and we look after systems other people built too.
Website chatbot
answers sampled daily
AI voice line
test call every morning
Invoices → Xero
every run checked
CRM sync
health check every 5 min
Lead follow-ups
queue watched hourly
Everything running. Nobody needs to look, which is the point.
example status board and incident
001 / WHAT WE WATCH
AI systems rarely crash loudly. They drift: a token expires, a model changes, answers slowly get worse. That's what we watch for.
Is it running?
Health checks on every workflow, voice line and integration, with alerts to a person when something stalls.
Is it right?
Real conversations and outputs sampled and reviewed, so quality doesn't quietly slip.
Are the connections alive?
Expiring tokens, changed APIs and permissions, found before they break a run.
Is it costing what it should?
Model and platform costs tracked monthly, with savings when cheaper options appear.
Is it safe?
Access reviewed, secrets rotated, dependencies updated and privacy settings kept current.
Are the models current?
Providers retire and change models often. We test and move you over before it matters.
002 / NO OFF-THE-SHELF PLANS
Every system is different, so every support arrangement is too. We agree what we watch, how fast we respond and what we improve, in writing, before it starts.
What's running
A single chatbot needs different care from a voice line with 29 workflows behind it.
How critical it is
Systems that touch customers or money get closer watching than internal helpers.
How much changes
Busy businesses add services, prices and staff; the AI has to keep up.
Who's on your side
Some teams want us to run everything, others want help when they're stuck. Both work.
003 / WHEN SOMETHING BREAKS
During business hours you reach the people who know your system. Response times are agreed per project and written into your support arrangement.
- 01
Alert
Monitoring flags it, or you tell us by email, phone or Slack.
- 02
Someone picks it up
One of the team who knows your system, not a call centre.
- 03
Safe first
Work is paused or queued so nothing is lost or sent twice.
- 04
Fixed
Repaired, tested and replayed, and you're told what happened in plain English.
- 05
Stopped from happening again
The cause is noted and a check added, so the same problem doesn't come back quietly.
004 / BETTER EVERY MONTH
Monthly review
A short report: what ran, what was fixed, what changed and what we suggest next.
Answers tuned
Questions the AI struggled with are added to its knowledge and tested.
New services added
Prices, products, staff and policies kept up to date in your AI.
Next automation
Ideas for the next task worth handing over, based on what we see in your systems.
Docs kept current
Runbooks and diagrams updated as things change, so they're always ready to hand over.
Training for new staff
Short walkthroughs when people join, so the system doesn't depend on one person.
005 / BUILT BY SOMEONE ELSE?
A chatbot from a previous agency, n8n or Zapier workflows a former staff member set up, a voice agent nobody quite understands any more. We start with a review, document what's there, fix the urgent things, then look after it like our own.
- 1Review: what's running, what it touches, what's fragile
- 2Document it, so it no longer lives in one person's head
- 3Fix the urgent issues and close security gaps
- 4Agree ongoing care, or hand it back to your team
006 / QUESTIONS
Anything else is quicker to answer on a call. Thirty minutes, free, with the people who would build it.
It depends on the project: what's running, how critical it is and how much it changes. We scope support for each client and agree it in writing before it starts, so there are no surprises.
Tell us what you've got. We'll review it, tell you what's fragile, and suggest support that fits.
Book a free call→