Wordsmith

What it is

Wordsmith is an AI platform for in-house legal teams built around the department’s front door instead of the document. Requests arrive in the channels the business already uses — Slack, Microsoft Teams, Outlook, email, Salesforce — and Wordsmith captures, triages, resolves, and records each one: agents answer routine questions against the company’s own playbooks and templates, redline commercial contracts, and hand off to a lawyer when the request needs judgment, with every step written to an audit trail. Founded in Edinburgh in 2023 by CEO Ross McNairn, CTO Volodymyr Giginiak, and COO Robbie Falkenthal, Wordsmith raised a $70M Series B on 3 June 2026 led by Highland Europe and Index Ventures, then a $14M extension on 5 August 2026 led by Intact Private Capital with FT Ventures joining — $114M raised across seed, Series A, and Series B. More than 500 companies run it, including BT, the Financial Times, Canva, Sage, Starling, Trip.com, Deliveroo, Belron, and Rakuten.

Why it shows up in Legal Ops stacks

• The unit of work is the request, not the file. Ironclad and Juro start from the contract record; Harvey and CoCounsel start from the research question. Wordsmith starts from the inbound ask — procurement pasting a vendor MSA into Slack, a rep asking whether a counterparty’s redline is acceptable, HR checking an offer letter. Intake, triage, resolution, and the record sit in one pipeline rather than a form that feeds three other tools.
• It ships an official MCP server. The endpoint at https://mcp.wordsmith.ai exposes ask_question, upload_file, list_playbooks, list_templates, list_assistants, and async job-status polling, over a REST API at https://api.wordsmith.ai/api/v1 with bearer auth and webhook callbacks. Legal review becomes a step another team’s agent can call — from Claude Code or any MCP client — instead of a destination app someone has to open.
• The ROI argument is outside-counsel spend, not seat productivity. Wordsmith sells against the law-firm invoice, and its published customer numbers follow: Belron avoided $100,000+ in external counsel spend in the first four weeks and averaged 238 minutes saved per task; Trustpilot reports contract review time down 85%. Both figures are vendor-reported.

Pricing reality

There is no per-seat list price. The pricing page publishes three tiers and attaches a number to none of them: Free (self-signup, no card, evaluation usage, all primitives unlocked), Production (usage-metered — “pay only for what your agents ship” — with a default $200/month budget, self-set spend caps, and a hard pause at 100% instead of silent overage, plus SSO, RBAC, SCIM, and BAA/zero-data-retention options), and Enterprise, branded Wordsmith Black (dedicated infrastructure, custom retention and residency, a named deployment engineer, procurement and DPA support). The vendor’s framing is “three tiers, one billing surface, no per-seat tax.”

In practice that splits the buying motion in two. You can meter a pilot against a $200/month ceiling with no purchase order and no demo — genuinely cheap to test. The department-wide deployments those 500 logos represent still land through a 30-minute demo and an annual contract Wordsmith does not price publicly. Budget the pilot to the dollar and treat the rollout as an unknown until you have the quote.

Best for

Legal ops managers and GCs at 200–5,000-person companies whose bottleneck is request volume from procurement, sales, and HR — not litigation, not contract storage. The specific trigger: a two-to-six-lawyer department where Slack DMs and a shared legal@ inbox are the intake system, nobody can say how many requests came in last quarter, and the outside-counsel invoice keeps absorbing work that should have been answered internally. It is also the catalog’s strongest pick if you want legal review callable over MCP from agents your own team builds.

Versus the alternatives

Ironclad is the incumbent when the object of record is the contract — pick it when you need repository, obligations, and renewals, and intake is a form that feeds the CLM rather than the product itself. Onit is the enterprise legal management pole — pick it when matter management and outside-counsel spend controls are the actual purchase and legal already runs a mature ops function. Checkbox is the fastest-growing entrant in the front-door segment ($23M Series A in January 2026) — pick it when you want a no-code canvas you configure yourself and intend to run non-legal workflows on the same platform; pick Wordsmith when you want prebuilt legal behavior and would rather not staff the configuration. GC AI is the in-house assistant with a published $500/seat price — pick it when the lawyers want a better tool and the business-side intake problem does not exist. If none of them fit, the honest fallback is a routing form in the service desk you already own plus a triage rota; that costs nothing and fixes measurement, which is often the real gap.

Watch-outs

• Every efficiency number is vendor-reported. The 85% review-time cut, the 238 minutes per task, the $100K in four weeks — all Wordsmith’s, none independently audited. Guard: run the free tier on one request type for 30 days, and gate the annual commit on a deflection rate and an outside-counsel spend delta you measured yourself.
• “No published price” is not “cheap.” The $200/month figure is a default spend cap on metered platform usage, not a subscription price, and it tells you nothing about the enterprise contract. Guard: get the rollout quote in writing before the pilot ends, and ask specifically what the metered bill does when agent volume triples.
• Intake tools fail on adoption, not accuracy. A front door only works if the business walks through it; the failure mode is procurement continuing to DM the GC directly. Guard: launch with one requester group and one request type, leave the old channel open in parallel for a quarter, and track the share of requests arriving through Wordsmith as the go/no-go metric.
• It is not a CLM and not a matter-management system. Wordsmith resolves and records requests; it does not own your executed agreements or your legal spend. Guard: keep the repository (Ironclad, Juro) and spend management (Brightflag) in place, and scope Wordsmith to the intake-through-resolution layer.
• EU data residency is an Enterprise-tier item. Custom retention and residency sit in Wordsmith Black, so the self-serve Free and Production path will not satisfy a hard EU-only requirement. Guard: if residency is non-negotiable, confirm it in the contract before piloting, because the pilot path does not prove it.