What it is
Scribe records what you click and turns it into a written procedure. You hit record in the browser extension or desktop app, do the task once, and it produces a numbered guide with a captured screenshot at every step, the click target highlighted. Editing is where most of the value lands: you rewrite the auto-generated step text, blur a field, drop in a callout, and publish a link.
The company renamed from scribehow.com to scribe.com and now sells three things. Scribe Capture is the documentation product described above. Scribe Optimize, launched alongside the Series C in November 2025, reads the captured workflow data across an org and reports where time actually goes and which processes are worth automating. Scribe Agents and an MCP server followed, exposing that workflow data to agents.
The scale numbers are unusual for the category. Scribe reported crossing $100M ARR on 11 May 2026, roughly six months after a $75M Series C led by StepStone announced 12 November 2025 at a $1.3B valuation, with Amplify Partners, Redpoint Ventures and Tiger Global participating. At the Series C it ran on about 120 employees. It reports 600,000+ organizations and 15 million documented workflows across 40,000+ applications. New York Life, T-Mobile, Klaviyo and Schneider Electric are on the public customer list.
The fork it resolves
Every ops team hits the same wall: the wiki page nobody updates. Someone wrote the refund-approval procedure in Notion in 2024, the billing UI moved two buttons since, and now new hires ask in Slack instead of reading it. The document was never wrong on purpose. It rotted because writing it cost 40 minutes and re-writing it costs 40 minutes again.
Scribe’s answer is to make the cost of the second capture near zero. Re-running the task with the recorder on is the update. That changes who writes documentation — it stops being a project for whoever has slack in their week and becomes something the person who does the task produces as a byproduct.
The second fork is newer and matters more in 2026. If you are pointing an agent at your internal processes, the input format is the whole question. Scribe reports that agents given structured workflow data performed 71% better than baseline, while agents handed full knowledge bases performed 10% worse — vendor-reported, not independently replicated, so treat the direction as the signal and the magnitude as marketing. The direction is consistent with what teams building RAG over Confluence keep finding: a wiki dump is worse than nothing because it retrieves plausible stale prose.
Pricing reality
Published rates, checked 7 September 2026. Basic is free, one seat, browser capture only. Pro Personal is $25 per seat per month billed annually, $35 monthly, from one seat. Pro Team is $13 per seat per month billed annually, $17 monthly, with a 5-seat minimum. Enterprise is quote-only.
Read those two Pro rows again, because the per-seat price inverts. Pro Team at five seats is $65 a month. Three Pro Personal seats are $75. A three-person team that buys the plan named after its size pays 15% more than the same team buying the five-seat minimum and leaving two seats empty. Buy Pro Team from the first paid seat.
Enterprise is where the compliance features live, and it is a real jump. Vendr’s published contract data puts the median Scribe enterprise deal near $8,640 a year; reported quotes for small enterprise teams run to roughly $18,000. Behind that gate: automatic PII and PHI redaction, SAML SSO, SCIM, role-based access control, IP allowlisting, multi-team governance, and BAA execution for HIPAA. The MCP server is Enterprise-only and covers select modules.
The gate placement is the thing to argue about in the negotiation. Auto-redaction is not an admin convenience — for a healthcare or legal-ops team it is the control that makes capture permissible at all, and it sits three tiers up from where the pilot happens.
Best for
Ops leaders and enablement owners at 50-500 person companies standardizing screen-based procedures across a tool sprawl nobody has time to consolidate. The scoped case where Scribe wins outright: onboarding a cohort of new hires onto 8-15 SaaS applications, where the procedures change quarterly and the person who knows each one has 20 minutes, not an afternoon.
It is also the right pick when the deliverable is a shareable link embedded somewhere else — Notion, Zendesk, Confluence, SharePoint — rather than a destination knowledge base you are asking people to visit.
Not for
Teams documenting work that does not happen on a screen. Warehouse procedures, phone scripts, anything with a physical step: the recorder has nothing to capture and you are back to writing.
Also not for teams whose primary need is guiding someone through the app in real time. That is a different product shape — an overlay that walks the user through the live UI — and Whatfix or Tango’s Guide Me do it properly. A Scribe is a document you read beside the app, not inside it.
And it is the wrong pilot if your compliance review requires enforced redaction. On Pro, redaction is manual and any author can skip it. Do not run a PHI-adjacent pilot on Pro and plan to tighten later.
Versus the alternatives
- Tango — the closest direct rival and the main competitive threat. Pro is $15 per user per month annually at 3+ users, $22 at 1-2; the free plan allows 5 shared workflows for up to 10 users. Enterprise adds SSO, automatic PII redaction, SCIM, 15-language translation, 365-day version history and the Guide Me overlay. Pick Tango when in-app guidance is part of the requirement, or when you want a cheaper per-seat rate at three people and can live with a 5-workflow free tier. Pick Scribe when the artifact is the point and you need editing depth.
- Guidde — the fastest-growing entrant. It produces a step guide and a narrated video from one capture, with AI voiceover. Pick it when the audience watches rather than reads, which is most customer-facing enablement and very little internal ops.
- Notion plus screenshots — the status quo you are replacing, and honest about its cost. Pick it when you have under roughly 20 procedures and one person owns all of them. Under that count, a documentation tool is overhead.
If none fit, the real alternative is deleting the procedures nobody reads and writing five that matter. Most SOP libraries are 80% dead weight, and no capture tool fixes that.
Watch-outs
- Redaction sits behind the Enterprise quote, and the pilot is when data leaks. Manual blurring on Pro depends on the author remembering, every time. Guard: run the evaluation against a sanitized test account with synthetic records, or negotiate an Enterprise trial with Smart Privacy Screen enforced at the admin level so authors cannot disable it. Get the redaction requirement into the security review before the pilot, not after — discovering it once 200 guides exist means re-capturing them.
- Screenshot guides rot silently when the underlying app ships a UI change. The text stays plausible and the picture no longer matches, which is worse than an obviously stale document because the reader trusts it. Guard: tag every Scribe with the source application, subscribe to that vendor’s release notes, and re-record on UI change rather than on a calendar. Unlimited version history is Enterprise-only, so on Pro your rollback path is re-capture.
- The Optimize and Agents pitch pulls a $13-per-seat purchase into a six-figure transformation conversation. Optimize is aimed at CIO and transformation budgets, and the sales motion changes accordingly. Guard: scope and sign the Capture purchase on its own terms and its own budget line. Evaluate Optimize as a separate decision after you have 6+ months of real capture data, because without that data it has nothing to read.
- MCP access is Enterprise plus select modules, so do not architect on it from a Pro seat. Teams that prototype agent access on a Pro trial find the connection is not available on the plan they budgeted. Guard: confirm in writing which modules the MCP server covers for your contract shape before any agent work depends on it.