Every Company's Slowest Department Has the Same Name: Legal

01 /The Concept

In mid-size and large companies, everything involving external partners or clients tends to slow to a crawl – because contracts and significant documents have to pass through legal. Lawyers draft, review incoming documents, and then re-review every negotiated change. The bottleneck is structural.

Wordsmith wants to turn that bottleneck into the fastest department in the company.

The core product is a specialized legal AI assistant that employees can query around the clock.

The assistant answers questions – not just general legal questions, but ones tied to the company's own internal policies. Plugging in the company's legal document repositories and policy libraries makes those available as context.

It can help employees prepare for negotiations: "Help me explain our standard contract terms to a client."

Send it an incoming document from a counterparty and it will review it, flag how it compares to company policy, and color-code clauses by risk level – "this is highly problematic," "this is risky but potentially acceptable," "this is fine."

It drafts legally sound responses to customer complaints and compliance questionnaires – the kind typically required for RFP submissions.

Wordsmith has tuned its AI specifically for legal documents. In its own benchmarks, it outperforms general-purpose models from Google, OpenAI, and Anthropic on legal tasks. It trains specialized bots on specific legal domains in partnership with firms like DLA Piper.

But the technology isn't the real play.

The real play is what Wordsmith explicitly doesn't try to do: replace in-house lawyers. The AI handles the groundwork – the steps that can be accelerated. Complex questions the AI can't confidently answer, or is explicitly blocked from answering, are escalated to a human attorney. Documents requiring a human signature still get reviewed by a human – but the reviewer sees the AI's markups, risk flags, and preliminary analysis, which sharply cuts the time needed to reach a decision.

The rules governing what the AI can handle, what gets escalated, and how employee interactions with legal are routed are all set by a platform administrator.

Wordsmith was founded in Scotland in late 2023. Despite its early stage and geography, it's already signed technology company clients including TrustPilot, and just raised $5M in a round led by Index Ventures and General Catalyst.

02 /Why It Matters

The most interesting thing about Wordsmith is the product it most resembles – which is, unexpectedly, corporate procurement software.

Before modern procurement platforms existed, every purchase request went through a buyer manually. Then self-service procurement arrived: employees could get automatic approvals for purchases that matched company policy – right department, right budget, right spend category. Anything outside policy routed to a human approver. The result was dramatically faster purchasing and a much lighter load on the procurement team.

Zip, [covered here](/review/rynok-gde-za-2-goda-mozhno-stat-milliardnoj-kompaniej), became a billion-dollar company within two years of launching exactly this kind of platform, raising $181.2M. Lumos, covered last month, focused specifically on software procurement and raised $65M.

Wordsmith applies the same model to legal. Whatever can be self-served, employees self-serve. Only what genuinely requires a lawyer reaches a lawyer. The result is the same: appropriate control maintained, operational friction reduced.

The business case is solid. Over the past few years, companies have increasingly brought legal work in-house rather than outsourcing it. Between 2020 and 2022, the number of in-house attorneys at US companies grew 77% – to the point where in-house headcount now roughly equals the attorney count at the top 500 US law firms. Even startups are hiring their first general counsels earlier than ever.

Two converging pressures explain it. Outside legal services are getting expensive fast – average legal spend at US companies jumped 30% in 2023, driven by a supply-side squeeze: senior attorneys are retiring and younger people aren't entering the field in sufficient numbers. And speed has become a critical competitive variable that outsourced legal simply can't deliver.

In-house legal teams are growing. Their priorities are using technology to cut manual work, and pulling more work in-house that technology frees up capacity to handle. Wordsmith is building exactly that technology – and timing it well.

The legal technology market is already at $31B and projected to reach $36B by 2027. So there's plenty of both hassle and money here.

03 /Opportunities

Honestly, Wordsmith is hard to improve upon conceptually – the most direct play is building an equivalent platform for the same or adjacent markets.

The deeper observation is structural: the self-service-plus-escalation model Wordsmith applies to legal is really a general pattern for any department that handles high volumes of structured, policy-governed decisions.

IT departments are an obvious candidate – large portions of their time go to answering standard employee requests that follow predictable patterns. HR and finance face the same dynamic, where policy compliance governs most routine decisions. The model works anywhere the ratio of policy-answerable requests to genuinely complex ones is high – and in most large organizations, that ratio is surprisingly favorable.