Run & Operate

Once your agent is live, someone has to watch it. Here are the five realistic options — with the honest tradeoffs on each. You choose.

Book a Free Scoping Call See the options

What happens after go-live

Four things every live agent needs watched.

AI agents need ongoing maintenance. Models update, costs shift, edge cases appear, and integrations change. None of it is dramatic — but all of it needs someone tracking it. These are the four areas that need attention after go-live.

Model Drift

Accuracy changes as the underlying model is updated by the vendor.

Cost Creep

Token usage shifts with volume and needs active cost management.

Edge Cases

New situations emerge that fall outside the original scope.

Integration Breakage

Connected tools update their APIs and need reconnecting.

The five options

Five ways to run an AI agent. You pick.

Every option on this page is honest. Every option has a best-fit scenario and a real tradeoff. We'll walk you through them on the call, recommend the one that fits your business, and we'll never push you toward the one that bills us the most.

Option 01

Full Handover

We hand it over. Your team runs it. We're done.

Best for

Teams with engineering capacity and experience running production systems. Businesses that already have observability, on-call rotation, and technical bench depth.

You'll need

An internal operations owner, monitoring tooling, credentials management, and someone who picks up the phone when something breaks.

Watch out for

Model drift and cost creep can go unnoticed for weeks without deliberate monitoring. "Set and forget" is where silent failures live.

Option 02

Handover + Training

We hand it over, but your team is trained to run it properly first.

Best for

Teams with capacity but limited AI operations experience. Businesses that want in-house ownership and are willing to invest in building the skill set.

You'll need

One or two team members willing to attend a structured training programme over the course of a week, plus access to the build environment.

Watch out for

Training covers the specific build we shipped, not AI agent operations in general. Deep cross-build expertise comes from running multiple agents over time.

Option 03

Run by Us

We run it. You get monthly reports. Ongoing retainer.

Best for

Teams that want zero operational overhead. Businesses where the workflow is too critical to risk in-house mistakes or too specialised to justify hiring for.

You'll need

An escalation contact, occasional business context updates, and a quarterly review meeting. We handle everything else — monitoring, drift, cost, model updates.

Watch out for

This is the highest-investment option. It's also the highest-leverage one — which is why it's the most common choice for revenue-critical workflows.

Option 04

Run by Your Vendor

Your existing IT partner or MSP runs it. We hand over and brief them.

Best for

Teams with an existing IT vendor, managed service partner, or agency relationship who already handle operational technology for the business.

You'll need

Your vendor's written agreement to take on the work, a formal handover session, and shared documentation access. We run the handover at no extra cost.

Watch out for

Not every vendor has AI agent experience. Before we hand over, we'll tell you honestly whether your vendor can handle it — and what to watch for if they can't.

Option 05

Hybrid

Your team runs day-to-day. We're on retainer for escalations and reviews.

Best for

Most mid-market businesses. The quiet default that balances cost, control, and access to specialised expertise when it's actually needed.

You'll need

An internal operations owner for day-to-day, plus a lighter monthly engagement from us for escalations, optimisation cycles, and quarterly strategy reviews.

Watch out for

Works best when the roles and escalation paths are clearly defined upfront. Blurred boundaries are where hybrid models quietly stop working.

Side by side

The options, compared.

A 10-second comparison across the five dimensions that actually matter. Scannable, honest, no asterisks.

Option	Day-to-day	Drift & cost	Model updates	Time to start
01 · Full Handover	Your team	Your team	Your team	Immediate
02 · Handover + Training	Your team	Your team	Your team	1 week training
03 · Run by Us	Us	Us	Us	Immediate
04 · Run by Your Vendor	Your vendor	Your vendor	Your vendor	1–2 weeks handover
05 · Hybrid	Your team	Shared	Us	Immediate

How to choose

Three questions. One honest answer.

If you're not sure which option fits, these three questions get you 90% of the way there. We'll walk through them on the call — but you can do the thinking ahead of time.

Question 01

Do you have production ops experience in-house?

Monitoring, on-call rotation, observability, deployment pipelines — the boring stuff that keeps production systems alive.

If yes → Option 01 or 02
If no → Option 03 or 05

Question 02

How critical is this workflow to your revenue?

If it stops working on a Friday evening, is your Monday morning affected? Is revenue, customer experience, or compliance at risk?

Mission-critical → Option 03 or 05
Nice-to-have → Option 01 or 02

Question 03

How fast does the stack underneath change?

Model vendors ship updates every few weeks. If your workflow sits on top of a rapidly-changing stack, someone needs to track it.

Changes often → Option 03 or 05
Stable stack → Option 01, 02, or 04

Questions

Frequently asked

What if I pick the wrong option?

You won't lock yourself in. Every option on this page is reversible. You can move between full handover, retainer, hybrid, or vendor-run as your business changes. The decision you make at go-live isn't a life sentence — it's a starting point.

Can I switch options later?

Yes. Many clients start with Full Handover and move to Retainer or Hybrid once they see what ongoing operation actually involves in practice. Others start with Retainer and graduate to Handover once their team has built capacity. The direction can go either way, and we handle the transition.

What if I built the agent with someone else — will you still help?

Yes. We take on agents built by other teams or vendors regularly. We start with a short audit to understand what we're inheriting — the stack, the documentation, the gaps — then run it under whichever of the five options fits your situation.

Do you publish pricing for the retainer option?

No. Retainer pricing is scoped to the number of agents, their volume, their autonomy level, and their risk profile. We confirm it on the call so it matches what you actually need rather than a number that won't fit your reality. Same philosophy as Build pricing — honest scoping over false precision.

Book a free call

Not sure which option fits? That's what the call is for.

30 minutes, no pitch, no obligation. Walk us through your business and we'll tell you honestly which of the five options makes the most sense — including the ones that don't involve us.

Book a Free Scoping Call

30 min · Honest recommendation · Zero pressure