AI discovery platform for operational excellence teams

Process documentation that reflects reality — and stays that way.

ClearWork runs AI interviews with the people who actually do the work, reads the SOPs and spreadsheets you already have, and generates process maps, SOPs and requirements that flag themselves for review when something changes.

Start free — 14 days Book a 30-minute call
$50 per user / month Everyone you interview is free
One source. Your document and your process map, updated in step Living documentation
Why documentation goes wrong

Everyone already has process documentation. Almost none of it is true.

Your SOP was accurate the month it was written. Here's the month-end close, as documented and as run.

What the SOP says

Five steps. Signed off in 2023.

Sub-ledgers closed Accruals posted Reconciliations complete Review & sign-off Results reported

Clean, linear, and wrong in five places since the system upgrade.

What the team actually does

The same close, with what nobody updated.

Variance fails → reopen ledger Sub-ledgers closed Accruals posted Review & sign-off Results reported RECONCILIATION — TWO WAYS Shared svcs in the ERP Regional in a spreadsheet Two accrual templates in use the SOP references a retired one Variation — two teams, two methods Neither is documented. Nobody has compared them. Journals over £50k need the CFO not the controller named in the SOP One analyst holds the close checklist on her own desktop

Every dotted line is real, load-bearing, and known only to the person doing it.

Nobody let the documentation rot. Updating it was never anyone's job.

The system changed, someone built a workaround, a second team started doing it differently — and the SOP kept saying what it said in 2023. So the next initiative starts by asking the same questions again. And when the analyst who holds the close checklist leaves, her share of the process goes with her.

Before you commit to anything

One process. One week. Then decide.

Process tools become shelfware for a predictable reason: they ask for a programme before they return anything. Roll out to every team, licence every user, map the whole operating model — and then find out whether it was worth it.

Start with one process instead. The messy one. The one where you already know the documentation is wrong.

If a week doesn't produce something your process owner would actually use, you've spent a week and a trial finding out. That's the whole risk.

Day 1

Upload what already exists

The outdated SOP, the half-finished process map, last year's workshop notes, the spreadsheet someone maintains on the side. ClearWork reads it and drafts a baseline.

Days 2–5

The people who run it respond

Interview links go to the six or seven people who actually do the work — across shifts, sites and time zones. They answer by voice or chat, or record their screen and narrate it. No meetings to schedule.

End of week

A current-state map, an exception list, and a list of what nobody could answer

Plus the conflicts between what two teams told you — surfaced rather than averaged into something that's true for neither.

You licence the handful of people running the work. Everyone who answers an interview is free, with no cap — so "who do we have to buy seats for" never becomes the reason a rollout stalls at five people.

How it gets the truth out

Three ways in — and you don't have to book a single meeting.

ClearWork works out what's missing from what you already have, then sends targeted asynchronous interviews to the people who can close each gap.

Existing SOPs, policies and spreadsheets imported into ClearWork
01 Start here

Everything you already have

SOPs, policies, training guides, process maps, spreadsheets, workshop notes, recordings. Many teams never use AI interviews at all — they run discovery as they always have and upload what comes out. That works completely.

Structured questions with a screen-recording request, as the process owner sees it
02 Targeted

Structured questions & screen walkthroughs

Questions generated from the actual gaps, by role. Plus a narrated day-in-the-life recording — so the person shows you the real workflow instead of describing it from memory.

An async interview by voice or chat, following up on the answers given
03 Conversational

Async voice and chat interviews

A conversation by voice or chat that follows up on ambiguity and asks the second question — whichever the person prefers, on their shift rather than yours.

This augments live discovery, it doesn't replace it. Run your workshops. Async interviews reach the second-shift supervisor, the regional team, and the person who was on leave that week — and whether your organisation takes to them depends on its culture. Some teams use them constantly; others upload transcripts and never touch them. Both produce the same record.

Human review

Nothing becomes documentation until a process owner says so.

Findings arrive as proposed changes — what it is, where it came from, what it affects. Your owner accepts, edits or rejects.

Owner approves
Document review showing proposed changes with their sources, awaiting accept, edit or reject
01
Variations
The same process, run differently across teams, sites or shifts — surfaced side by side, not averaged.
02
Conflicts
Two sources that disagree. Both shown with their evidence so an owner decides, not the software.
03
Unknowns
Nobody has said it yet. Tracked as an open gap rather than filled with a plausible guess.
After the document exists

The hard part isn't writing the SOP. It's the eighteen months after.

Drawing a process map is a solved problem. Keeping it true once the system changes, the team reorganises, and the workaround appears is not.

Every artifact stays connected to the sources behind it. When one of those sources changes — a corrected interview answer, a new policy, an updated procedure — every document built on it flags itself for review, with the change summarised and linked to what caused it.

A person approves it. The version history records who and when. Documentation doesn't go stale quietly, and it doesn't rewrite itself behind your back either.

Source-linked outputs

Every map, SOP and requirement traces to the knowledge that produced it.

Version history

What changed, when, and who reviewed it.

Process owners

Assign owners and reviewers so documents don't become orphaned.

Review cycles

Scheduled re-checks, plus flags whenever a source moves.

Process map and artifacts generated from the validated record
 
A diagramming or SOP tool
ClearWork
Where it starts
A blank template. Someone has to already know the answer.
Your existing documents plus interviews with the people who run the work.
What the diagram knows
The steps. Not why they exist, who owns them, or what the exceptions are.
Steps stay connected to owners, systems, exceptions, requirements and risks.
When something changes
Nothing happens. Someone notices months later, if anyone does.
Dependent documents flag for review with the cause linked.
Who can challenge it
Nobody — there's no evidence behind the box.
Anyone. Click a line and see the interview, page or timestamp it came from.
The same initiative, twice

From scattered context to a record people trust.

How it usually goes

Six weeks in, still no source of truth

The team gathers old SOPs, partial maps and workshop notes. Everyone contributes something. Nothing reconciles.

  • ×An SOP last touched three system versions ago
  • ×A process map that stops at the happy path
  • ×Two teams describing the same step differently, unresolved
  • ×Workarounds nobody wrote down
  • ×No owner, no review cycle, no version history
With ClearWork

The same work, as a governed record

Existing material goes in, interviews close the gaps, owners approve what's true, and the artifacts generate from there.

  • ✓Current-state process map, including exception paths
  • ✓Swimlane view by role, team and system
  • ✓A documented exception and variation list
  • ✓Requirements, risks and open questions with sources
  • ✓Owner-approved documents with review cycles and history
Who it's for

Teams accountable for how work actually happens.

Operational excellence

Map current state, find waste and rework, standardise across teams, and keep the exception list honest.

Internal transformation

Understand how work happens before you redesign the operating model or deploy a new system.

PMO & programme teams

Keep discovery, requirements, risks, decisions and stakeholder input connected across workstreams.

Process owners

Maintain SOPs, maps, controls and training material for the processes you're accountable for.

Straight answers

What we can and can't tell you about security today.

You'll have to take this through IT and procurement. Better you know where we stand now than after a security review.

Live now

  • No AI training on your data. Contractually excluded at the provider level.
  • Role-based access. Scope people to specific projects and process areas.
  • Human approval on every change that enters the record.
  • Full version history — what changed, when, and who reviewed it.
  • Project-level isolation between workstreams.
  • Owner and reviewer assignment so documents don't become orphaned.
In progress

SOC 2. Underway. Ask us on a call and we'll give you a real status rather than a reassurance.

If your security review needs a control or certification you don't see here, ask on the call. We'll tell you where it actually stands — weeks, quarters, or not on the roadmap.

Pricing

You licence the coordinators, not the organisation.

$50per user
per month

That covers the people running discovery, reviewing findings and maintaining the record. Everyone else is free.

Start free — 14 days See full pricing
  • Everyone you interview is free — no cap, however many SMEs you need to reach
  • The whole platform — no tiers, no feature gates
  • Add and remove users monthly as initiatives start and finish
  • Monthly, cancel anytime. No annual commitment
Questions

For internal teams, answered.

How do we roll this out without it becoming shelfware?

Don't roll it out. Start with one process — ideally the one where you already suspect the documentation is wrong. Upload what exists, send interview links to the handful of people who run it, and see what a week produces.

Shelfware happens when a tool needs a programme before it returns anything. This doesn't.

Do we have to licence everyone we interview?

No, and this is the part most teams don't expect. Anyone who answers an interview, records a walkthrough, or reads finished material is free, with no cap. You licence the coordinators — the people running discovery and maintaining the record.

It's deliberate. If reaching one more person costs money, teams stop reaching people, and the whole point is to hear from those who never make the workshop.

We already have Copilot. Why not just use that?

An assistant can draft a process document from files you point it at. That was never the hard part.

Two things are. It can only work with what someone already wrote down — it can't work out that your regional team reconciles in a spreadsheet and then go ask them about it. And a chat thread doesn't persist: ask twice and you get two different answers.

Correct something here and it becomes a fact the system honours from then on — owned, versioned, reviewable, and still true when the next initiative starts.

Does this replace our workshops?

No, and we'd push back if you tried. Live sessions are where you read the room, resolve disagreements and build buy-in.

Async interviews extend the reach — the second-shift supervisor, the regional team, the person on leave that week. Then they keep the record current after the workshop ends, which is where the documentation usually starts drifting.

Can we keep our own SOP format and approval process?

Yes. You keep your structure, template and sign-off route. What changes is that the content is populated from real process knowledge and flags itself for review when a source changes, rather than sitting in a folder going quietly out of date.

How does documentation stay current after the project ends?

Every artifact stays linked to the sources behind it. When a source changes, the documents built on it flag for review with the change summarised and the cause linked. An owner approves, and version history records who and when.

Nothing rewrites itself silently. The point isn't automatic documents — it's that going stale becomes visible instead of invisible.

Can we compare how different teams or sites run the same process?

Yes, and it's one of the more useful things it does. Where two teams do the same work differently, both versions are surfaced side by side with their evidence rather than averaged into something true for neither. You decide what should be standardised and what variation is legitimate.

What can it generate?

Process maps and swimlane diagrams, SOPs and work instructions, business and system requirements, user stories, current-state narratives, risk and gap registers, open questions, improvement backlogs, and executive summaries.

Documents export to Word and PDF; diagrams to Visio and draw.io.

Bring us one messy process

We'll show you what it turns into.

Start with one complex workflow, one outdated SOP, or one initiative that's about to start. Fourteen days, your own materials, and a real answer at the end of it.

Start free — 14 days Book a 30-minute call
$50 per user · Everyone you interview is free