Skip to content
PlotUProperty intelligence · AI · evidence

Project controls · 9 min read

From WhatsApp threads to a decision trail: project controls that survive scrutiny

The information existed — in diaries, inboxes and camera rolls. What did not exist was a record that could answer 'who decided this, when, on what basis?'

Published

About this case study

This is an anonymised composite drawn from recurring engagements. No client is identified, no figures are invented, and no outcome is guaranteed. The workflow is the substance.

The situation

A developer with several schemes running concurrently was managing delivery through a familiar stack: site diaries in notebooks, instructions buried in WhatsApp, photographs on personal phones, and weekly reports written from memory on a Sunday night. Nothing was dishonest; almost nothing was retrievable.

The details that follow are a composite of recurring engagements of this kind. No client is identified, and nothing here is a claim about a specific company's results.

How the workflow runs

  1. Capture where the work already happens

    The workflow does not ask site teams to change tools. Diaries, messages, photos and instructions are routed into one structured store, dated and tagged by project, area and contractor.

  2. Structure the record daily, not monthly

    AI drafts the daily structured record from the raw inputs: what happened, what was instructed, what was decided, what is blocked. A human reviews and confirms it — the draft is never the record of its own accord.

  3. Log decisions with their basis

    Material decisions get a decision entry: what was known, when, which document or drawing revision it rested on, and who approved it. This is the spine of the PROVE stage of the framework.

  4. Let reporting assemble itself

    The weekly report is drafted from the confirmed records — progress, instructions, issues, decisions — and edited by the person who owns it. Nobody reconstructs the week from memory.

  5. Keep the pack exportable

    For a lender drawdown, an insurer query or a dispute, the evidence pack is a filtered export of records that already exist, not a forensic exercise.

What changed

The record stopped depending on memory. When a question arrives months later — from a funder, an insurer or the other side of a disagreement — the answer is a lookup, dated and sourced.

  • Every material decision carries its date, basis and approver
  • Weekly reporting is edited, not reconstructed
  • Instructions and variations trace back to a drawing revision
  • Evidence packs assemble from existing records rather than a scramble

Where the professional line sits

  • AI drafts; the responsible person confirms every daily record
  • Contractual positions and certifications stay with the professionals appointed to make them
  • Personal data handling follows the confidentiality rules agreed before anything is routed anywhere

FAQ

Common questions

Does this work with WhatsApp, or do we have to change tools?
The design principle is capture where the work already happens. The workflow routes existing channels into a structured record rather than forcing a platform migration that site teams would ignore.
Would this hold up in a dispute?
It is built for that context: dated, sourced, human-confirmed records with version history are materially stronger than reconstructed memory. It is not legal advice, and your solicitor remains the right person to judge any specific position.
Is this a real project's story?
It is an anonymised composite of recurring engagements. We do not publish client identities, and we do not invent metrics to make a workflow sound better than it is.

Keep reading

More case studies

Discuss your workflowAll case studies