Your documents already know how your business works

You run a professional services firm. Two, maybe five senior people. Between you, you're running fifteen projects.

Every week you write the same things.

Fee proposals. Progress claims. Cost reports. Tender packages. Notices. Compliance packs. Each one rebuilt from scratch — despite the fact that last month one of you assessed a progress claim on the same contract type. Same logic, same structure, different numbers. That's two days each of you didn't spend on site, or with clients, or doing the work you actually charge for.

This is what most professional services firms look like from the inside. And it's not a scheduling problem, or a discipline problem, or a tools problem.

It's that nobody has ever asked the documents what they know.

Six document types run your entire practice

Firms think they produce dozens of document types. They don't.

Sort everything you've produced this quarter by purpose instead of by project, and you'll find six classes.

1. Fee proposals — the thing you write to win work.
2. Tender & procurement — the thing you run to get the client a builder or supplier.
3. Contract administration — notices, progress claims, variation assessments. The thing you do every week once execution starts.
4. Cost & financial reporting — budgets, cost-to-complete, cash flow. The thing the client calls about on the 15th of the month.
5. Project reporting — monthly reports, risk registers, dashboards. The thing that keeps everyone informed.
6. Compliance & approvals — authority submissions, DAs, OC packages. The thing between the client and the regulator.

That's it. Everything else is general admin. These six are what clients pay for and what takes your week.

The three layers inside every document

Open any one of those six. Whatever you open, it has three layers underneath.

Rules. Things that come from outside — contract terms, statutory obligations, fee scales, insurance minimums, regulatory thresholds. Not yours to change. You follow them or you're wrong.

Conventions. Things your firm always does — your formatting, your section order, your standard clauses, your preferred exclusions list, your way of presenting numbers. Identical across every project, but unwritten. "The way we do it."

Judgement. The parts that genuinely require a senior human thinking about the specific situation — interpreting a client's risk appetite, calling a scope boundary, weighing competing obligations. The bit that deserves your time.

A fee proposal runs roughly 30/50/20. A compliance submission, 35/50/15. A cost report, 50/40/10. Judgement never clears a fifth.

The first two layers — rules plus conventions — are the same every time.

You're spending most of your week recreating the parts that don't change.

A worked example — what 81 documents said

Here's what happened when we did.

One of an architectural practice's fee proposals had the wrong PI cover specified — $1M on a $2.8M project. Someone caught it in review. But the rule (PI cover must scale with project value) was never written down anywhere as a rule. It lived in one partner's head. Once that rule is explicit in the canonical structure, the wrong answer is structurally impossible.

We got there by taking eighty-one documents from that practice — fee proposals, scopes, contracts, certifications, correspondence across multiple projects — and running them through a pipeline that classified them by purpose and looked for semantic proximity.

The map found four fee proposals sitting in one cluster. Near-identical. Each one built from scratch. Each one consumed senior time. The overlap was 71% — same structure, same clauses, same boilerplate. Four authors had spent roughly ten senior days reproducing work that already existed in the firm's own filing system.

After the extraction, a compliant fee proposal takes thirty-eight seconds to generate and twenty minutes to review. Two and a half days compressed to half an hour. Same output quality — arguably better, because correctness is now built into the structure instead of checked after the fact.

The firm didn't gain a new skill. It made visible what was already true.

What to do about yours

This isn't a pitch to build something. It's a suggestion for a kind of looking.

Take a week's worth of documents. Every fee proposal, every progress claim, every report that left the office. Don't sort them. Don't tidy them. Just put them together.

Then ask three questions:

1. How much of this did we actually author this week? (Measure against last month's equivalents.)
2. Which parts follow rules we never wrote down? (The unexamined conventions.)
3. Which parts genuinely required judgement? (The bit that deserved the time.)

The ratio of those three tells you what your practice is really spending senior time on. Most firms, when they do this honestly, find they're spending sixty to eighty percent of senior hours on the first two — and they tell themselves it's the third.

Your documents already know

The canonical structure of your practice is not something you need to invent. It's already there, buried in the work you've already done.

Nobody's asked it.

---

Karl Howard · Reforged · 31 March 2026