Goran StankovskiGoran Stankovski··5 min read

Why We Built OpsChain

We have watched change management become theatre.

LimePoint has spent more than twenty years in enterprise operations, in the banks, utilities and government estates where a bad change makes the evening news, and the pattern has always been the same. The change record said one thing. The terminal did another. The CAB approved a description of a change, written days earlier by someone who would not be the one running it. Then a person with credentials and a deadline did whatever needed doing, and everyone assumed the record matched.

We have sat in post-incident reviews where the approved change record was read aloud, line by line, while the engineers in the room avoided eye contact, because the record described a tidy, rehearsed procedure and the outage logs described something else. We have watched audit teams spend weeks reconstructing what actually ran from shell histories, ticket comments and memory, for changes that had every approval in the book. And we have seen accountable owners sign off on changes knowing that the honest answer to "how do you know that is what was executed?" was that they trusted the person who ran it.

Trust is not a control. It took this industry a long time to say that out loud.

The ticket is not the change

The uncomfortable truth underneath all of it is that the ticket is not the change. The change is what ran, on which host, with which credentials, by whose hand. Everything else is narrative.

The industry's answer to that gap has mostly been more narrative. More fields on the change form. More approvers on the CAB. More process between the person who wants the change and the person who runs it. Meanwhile the execution itself, the only thing that ever touches production, stayed as governed as it was in 2003: a trusted human, a terminal, and hope.

That gap is where outages, audit findings and the 2am argument about who ran what and why all live.

The road through automation

LimePoint was founded in 2012 as a hands-on engineering consultancy. We implement and operate enterprise platforms. We do not observe from a distance. Out of that work came MintPress, our automation platform. If the manual keystrokes were the risk, the answer was to automate them away: define the environment, and let the machine build and configure it, repeatably and at scale.

Automation solved a real problem, and a decade of running it inside serious enterprises taught us something we did not expect: execution was never the hard part. We could make the right thing run reliably. What we could not hand an auditor, and what nobody could, was proof that only the right things could run at all. Automation made changes faster and more consistent, but governance stayed where it had always been, in the paperwork, one layer removed from reality. A pipeline with production credentials is a very fast trusted human.

OpsChain began as the answer to that. Earlier generations of the platform put real controls around real execution, and the estates running them proved the model. What came next turned a product line into a conviction.

Then the hands stopped being human

For twenty years the gap between record and reality was survivable, because the hands were human and humans are slow. A reviewer could, in principle, watch.

That era is ending. The hand executing changes is increasingly an AI agent working at a pace no reviewer can follow, given credentials because it was Tuesday and the backlog was long. Every organisation we speak to is somewhere on the path from "we would never let an agent touch production" to "an agent touched production last sprint and we found out in the retro."

If change governance was theatre when humans ran the changes, it becomes farce when the hands are synthetic, unless the governance moves to where the execution is. That is why we rebuilt the platform from the ground up for v5: one engine that governs the execution itself, whoever or whatever is doing the executing.

Govern the execution, not the paperwork

The principle has not moved since the first commit: govern the execution, not the paperwork.

  • Policy decides what may run and where, before it runs, as an enforced decision. It is not a guideline in a wiki.
  • Approvals gate the actual run. The thing the approver approves is the thing that executes.
  • Every action lands in an evidence ledger as it happens. Nothing is reconstructed afterwards from logs that were never designed to be evidence.
  • The same governance holds for every kind of hand: an engineer at a terminal, a pipeline, or an AI team member with its own identity, explicit grants, and an emergency stop that stops it.

When the auditor asks you to prove that an unauthorised change cannot reach production, you do not assemble a binder. You show them the refusal.

Where it goes from here

OpsChain v5 is now generally available. It governs human changes, pipeline changes and AI team members under one policy engine, one approval model and one evidence ledger, because production does not care who typed the command, and your regulator will not either.

We will be demonstrating the AI team member live in a webinar on Friday 27 November, 11am AEDT: onboarding it, bounding it, putting it to work under the same approvals as any human, and hitting the emergency stop. Register for the webinar. Joining details go to everyone registered.

The gap between the record and the reality was the defining operational risk of the last twenty years. Closing it is the work of the next twenty.

Every change governed. Every action proven.

Ready to see OpsChain in action?

Book a personalised demo and see how OpsChain can transform your operations.

Book a Demo
Goran Stankovski
Goran Stankovski

Founder & CEO, LimePoint

Goran is the founder of LimePoint and the creator of OpsChain. He is passionate about helping enterprises automate and govern their operations at scale.