← All thinking

Should a systems project ever really end?

Why "done" is the wrong frame for a live estate: what has to be watched, measured and rehearsed for it to hold.

Operations By a Delivery Director

After go-live an estate drifts and degrades as APIs change and data drifts, unless a watching layer (SLOs, synthetic checks, drift detection, AI Ops) holds it steady and improving.A LIVE ESTATE, AFTER LAUNCHreliabilitytimeGO-LIVESLOssynthetic checksdrift detectionAI Opswatched: holds & improvesAPI changeschema driftcert expiry
Go-live isn't the finish line. Left alone an estate quietly degrades; a watching layer of SLOs, checks, drift detection and AI Ops is what keeps it holding and improving.

A foundation is never done. The systems a business runs on keep changing, the expectations placed on them move faster, and the day a platform goes live is the day it starts drifting from the world around it. The honest question isn’t when a systems project ends. It’s what’s actually watching it once the invoice closes.

”Done” is a project myth

Software projects like to declare victory at launch, because that’s when the deliverable is complete on paper. But a connected estate is a live system: integrations age, data drifts, a vendor ships a breaking change on a Tuesday. None of that stops at go-live, it just stops being anyone’s job if no one stays. The lock-in people actually fear isn’t a long relationship. It’s a system nobody understands any more, because the people who built it left with the knowledge.

What actually keeps a live estate from failing over

Staying has to mean something specific, or it’s just a retainer in name. Service-level objectives on the integrations that matter, so “slow” and “down” have a number attached instead of a feeling. Synthetic checks that exercise the real flow on a schedule, so a silent failure gets caught before a customer reports it. A runbook for the foreseeable failure modes, and a rehearsed incident process for the ones that aren’t. None of it is glamorous. All of it is the difference between an integration that degrades quietly for a month and one that pages someone in five minutes.

It’s also where AI Ops comes in: proactive, agentic AI that watches every platform against its SLOs and catches drift before your customers feel it, so a maintenance retainer becomes monitoring that works a step ahead of you. The best systems work doesn’t end at launch. It compounds after it, in the SLOs and the runbooks nobody sees until the day they’re the reason nothing went down.

More thinking

See the work
01

Your platforms can talk, but does your business actually work as one?

Integration vs connectivity: why "the systems talk" is the wrong test, and what a governed event contract actually proves.
02

How do you build a connected core without a leap of faith?

What a connected core changes for your business, and the staged path that gets you there without a blind commitment.

Let’s start something that lasts.

Bring the problem you’re weighing: a migration, a re-platform, a build that has to hold. You’ll talk to the people who’d do the work, not a sales desk.

Book a meeting

Let’s talk

Book a meeting

A 30-minute video call with the people who’d actually do the work. Pick a time, tell us what you’re weighing, and we’ll bring the right minds, not a sales desk.

  1. Date
  2. Time
  3. Reason
  4. You

Which day works?

Times shown in your timezone.

Pick a time

What’s it about?

Last thing — who are you?