Two months from build to first live SOP.

It was never planned in a product meeting. It grew out of client work. At Zenith Rubber, a performance review project stalled because nobody could see whether work was being done, step by step and on time. We rebuilt the SOPs to fix the process, and then met the question every business eventually asks: once the process is written down, how do you know it is followed?

No tool answered that the way a factory needed, so in July 2026 we decided to build one. The build started in August and the first SOP went live at Zenith’s Dombivli factory in October. New SOPs now go live at about two a week.

That blind spot is expensive even though it never shows up as a line item. A quotation waits days because a hand-off has no owner. The same mistake returns every quarter because the fix never became anyone’s step. And the real way the business runs lives in two or three people’s heads.

We saw all of it up close. Performance reviews had paused because the data behind them did not exist, and the SOPs we were rebuilding kept running into the same wall. The tools we looked at either stored an SOP as a document or tracked a task’s start and end. None followed every step and milestone in between, each with its own deadline, and raised a flag before the delay landed.

The pace comes from a simple, repeatable cycle. Each SOP is built on the platform, tested on real transactions, demoed to the team that will run it, refined from their feedback and rolled out. A new SOP moves from paper to live in days, and a new business can have its first function live in four to six weeks.

Where AI fits. Once every step carries proof, there is a great deal of it to check. Optional AI can review the proof and flag what looks wrong, draft a first version of an SOP from an existing document, spot quiet patterns such as steps that slip at month end, draft root-cause notes, and answer plain-language questions about performance. It is switched on only when a business wants it, and can run on the business’s own model.

What was not working. Quality check data was hard to track in Business Central, and so were holds and dispositions on material. Customer complaints moved between people without a clear owner at each stage, so a complaint could sit unnoticed until a customer chased it. And because none of it was recorded step by step, the performance reviews had nothing reliable to measure.

How it rolled out. Customer complaints first — every complaint from first report to closure, with an owner and a deadline at each stage and an escalation if a stage runs late. Quality checks next, each with an owner and a deadline and its result recorded against the step rather than in a separate register. Then holds and dispositions, tracked to closure rather than left in the ERP where they were hard to follow. Once every Quality SOP was live, EXIM and Purchase followed. Production Planning is next.

Zenith kept Business Central. It still records the transactions; 100 Ops runs the steps that were long or awkward there.

What it needs from you. One senior internal owner, named up front. About four hours a month of leadership time for the first three months, then about two. And a start of two or three functions, adding the rest once people are using it.

Proof over ticks. A step closes on evidence, not a tap. Clean data about real work is worth more than tidy data about nothing.

Functions first, people later. Scoring starts with departments and processes. Individual scoring is a choice a business makes later, openly, once the system is trusted.

One internal owner, named up front. Every rollout needs a senior person inside the business who owns it. It is the strongest single predictor that the system will still be in use a year later.

Start small, then grow. Two or three functions first, then the rest once people are using it. Our first factory started with one SOP and now adds about two a week.