slack microsoft-365 salesforce ironclad zapier claude
GC AI and Wordsmith reach the same shortlist when a two-to-eight-lawyer in-house department decides its problem is volume rather than difficulty. Both are AI platforms for corporate legal, both review and redline contracts against your playbooks, both run inside Microsoft Word, and both are funded well past the point where either disappears — GC AI raised a $60M Series B in November 2025 at a $555M valuation, and Wordsmith raised a $70M Series B on 3 June 2026 plus a $14M extension on 5 August 2026, taking it to $114M.
The rule everyone repeats about these two is that GC AI is the lawyer’s tool at a published $500 a seat and Wordsmith is the front door that lets non-lawyers self-serve in Slack. That was accurate in the first half of 2026. Six weeks of shipping killed it, and the replacement rule is about metering, not personas.
What broke the old rule
GC AI moved into the front door. Its REST API reached general availability on 29 July 2026, described by the vendor as being “for people without seats, for systems that run without a human starting them, and for AI clients” — with named recipes for Salesforce, Ironclad, Jira, Zapier, Make, and n8n, and a documented Slack triage bot example built on a Zapier workflow. Agent Connectors now reach 20+ apps including Gmail, Outlook, Slack, Teams, Drive, SharePoint, HubSpot, Notion, Confluence, Asana, Linear, and Agiloft, with write actions defaulting to “needs approval” and an admin gate on the whole feature. On 5 August 2026 it launched Contract Intelligence, which queries signed agreements in plain English with cited answers across repositories including Google Drive, SharePoint, OneDrive, Dropbox, and Ironclad — infrastructure it bought, acquiring backend engineering firm Jetify in a deal dated 1 June 2026 and reported on 10 August by Legaltech News.
Wordsmith moved toward the document. Its homepage schema now describes the product as an “AI platform for in-house legal teams to draft, review, research, and manage contracts” that “works natively inside Microsoft Word,” and its published feature list leads with contract review, drafting, and research. Intake is one nav item among eight.
So both do intake and both do redlines. What survived the convergence is narrower and more decision-useful: who assembles the front door, and what the meter counts.
Where GC AI wins
You get a price without a sales call. The Individual plan is published at $500/seat/mo with a 14-day trial and a single-seat minimum. Almost nobody else in this segment gives you a number before a demo, and it means you can model a four-lawyer year one — $24,000 — before you talk to anyone.
Non-seated usage is priced per unit, publicly. The API bills in credits pooled across the organization at $5 per credit: one API chat message costs 1 credit, one playbook check costs 0.1 credit, so a 30-check contract review costs 3 credits, or $15. You can compute the business-side bill yourself instead of negotiating it.
Post-signature querying is in the box. Contract Intelligence answers questions across signed agreements, links amendments to their parents, and renames files to your convention. That is work a CLM usually charges for separately.
Connector permissions are per action, not per app. Each user connects their own accounts, admins never connect on their behalf, and every write action carries approve / always-allow / block, resolvable inside the thread. For privileged material that is the right granularity.
Where Wordsmith wins
The front door is a product, not a recipe. Slack intake through an #ask-legal channel ships with auto-routing by request context — privacy questions to the DPO, commercial questions to Commercial Counsel — conditional escalation rules, and two operating modes: Co-Pilot, which holds every answer for a lawyer’s approval, and Autopilot, which answers low-risk queries instantly. GC AI’s equivalent is a Zapier workflow you build and own.
The record is the point. Receive, route, resolve, record is the product’s own spine, and the audit trail is what a GC shows the board when asked how legal handled 4,000 requests. Building that on top of an API means building the reporting too.
Enterprise proof at size. More than 500 companies including BT, the Financial Times, Canva, Sage, Starling, Trip.com, Deliveroo, Belron, Rakuten, and Trustpilot, under SOC 2 Type 2 and GDPR. Vendor-published outcomes: Belron avoided $100,000+ in external counsel spend in four weeks; Trustpilot cut contract review time by 85%.
Requester volume does not move the bill. Whatever the seat price turns out to be, it is charged for lawyers. Ten thousand business-side questions cost the same as one thousand.
Pricing reality
This is where the decision actually gets made, and the two vendors are asymmetric in a way that matters more than the headline numbers.
GC AI publishes its floor and its per-unit rate. Four lawyers on Individual is $2,000/mo, or $24,000/yr. Team and Enterprise are annual and quoted. Business-side usage then adds $5 per API chat message on top. Run the arithmetic and the crossover is stark: every 100 business-side questions per month costs the same as another lawyer’s seat. At 400 questions a month the API bill equals the entire four-lawyer subscription. At 1,000 a month it is $60,000 a year on its own.
Wordsmith publishes nothing. Its pricing page as of today carries no tiers and no numbers, only a 30-minute demo booking and a free trial, and its FAQ routes full report access to “a pro or enterprise plan.” One disclosure survives: the ROI calculator’s own disclaimer states pricing “from {price}/seat/month” across tiers labelled Starter, Professional, and Enterprise, with the figure rendered client-side. That confirms the billing unit is a seat and the ladder has three rungs. It does not tell you the number. This also contradicts the usage-metered structure with a $200/month default budget that Wordsmith published earlier in 2026 — the model changed inside a single quarter, so treat any older quote as void.
The honest read: GC AI’s cost scales with how many questions the business asks, and Wordsmith’s scales with how many lawyers you employ. Pick the axis that stays flat for you.
Both are agent-reachable now
Wordsmith’s MCP server has been its distinguishing technical claim, exposing ask_question, upload_file, list_playbooks, list_templates, and list_assistants over an endpoint at https://mcp.wordsmith.ai, with a REST API behind it.
GC AI closed that gap. Its MCP server is in public beta with 20 tools across ask, files, projects, playbooks, and skills, over a single authenticated endpoint, plus OAuth 2.0 client-credentials provisioning so a machine client such as Salesforce Agentforce can hold org-scoped access. One design detail is worth checking against your access model: GC AI’s MCP requests run as the signed-in user rather than a service account, so the assistant sees exactly what that person sees and nothing more.
Treat MCP as a tie now, and decide on beta tolerance — GC AI labels its server public beta and says the tool surface will change.
Verdict
Pick GC AI when the lawyers are the users and business-side volume is modest or spiky; when a published price materially helps your business case; when you want post-signature contract querying without buying a second system; or when you have someone who will build the Slack bot on Zapier and would rather pay per message than per seat.
Pick Wordsmith when the business-side request queue is the actual problem — procurement, sales, and HR asking hundreds of questions a month — and you need routing, escalation modes, and an audit record that exist on day one rather than after an integration project. Also pick it when requester volume is growing faster than headcount, because that growth is free here and metered there.
Pick neither when you cannot say how many requests legal received last quarter or what they were about. Neither product fixes an unmeasured queue, and a routing form in the service desk you already own plus a triage rota buys you the measurement in a fortnight. If the real job is the contract repository, Ironclad or Juro is the purchase; if it is firm-grade research and litigation, Harvey; if it is Word-native redlining and nothing else, Spellbook is cheaper per seat; and if you want a configurable front door that will later carry HR and finance intake on the same tenant, Checkbox — see Streamline AI vs Checkbox.
Default pick when you genuinely cannot decide: GC AI. The cannot-decide case is one where nobody has counted the inbound requests, and in that state Wordsmith’s structural advantage is unmeasured while GC AI’s is checkable today — a published seat price, a published per-message rate, and a trial you can start without a procurement cycle. The condition that flips it is specific and you will recognise it: a counted business-side queue above roughly 400 requests a month, at which point GC AI’s meter costs more than its seats and a flat per-lawyer price wins.
Two guards for either negotiation. First, on GC AI, get Agent Connectors’ plan placement in writing — the marketing pricing page lists 20+ Agent Connectors under the $500 Individual plan, while the vendor’s own documentation places them under Team as an org-gated feature. Those two pages disagree, and connectors are load-bearing for the non-lawyer use case. Second, on Wordsmith, get the seat count definition and the year-three price in the same document as year one, since the published model changed once already this year. On both sides, put the model provider, subprocessor list, and a written no-training commitment in the DPA before privileged material moves — see legal intake and matter management for the underlying model.
GC AI and Wordsmith reach the same shortlist when a two-to-eight-lawyer in-house department decides its problem is volume rather than difficulty. Both are AI platforms for corporate legal, both review and redline contracts against your playbooks, both run inside Microsoft Word, and both are funded well past the point where either disappears — GC AI raised a $60M Series B in November 2025 at a $555M valuation, and Wordsmith raised a $70M Series B on 3 June 2026 plus a $14M extension on 5 August 2026, taking it to $114M.
The rule everyone repeats about these two is that GC AI is the lawyer’s tool at a published $500 a seat and Wordsmith is the front door that lets non-lawyers self-serve in Slack. That was accurate in the first half of 2026. Six weeks of shipping killed it, and the replacement rule is about metering, not personas.
What broke the old rule
GC AI moved into the front door. Its REST API reached general availability on 29 July 2026, described by the vendor as being “for people without seats, for systems that run without a human starting them, and for AI clients” — with named recipes for Salesforce, Ironclad, Jira, Zapier, Make, and n8n, and a documented Slack triage bot example built on a Zapier workflow. Agent Connectors now reach 20+ apps including Gmail, Outlook, Slack, Teams, Drive, SharePoint, HubSpot, Notion, Confluence, Asana, Linear, and Agiloft, with write actions defaulting to “needs approval” and an admin gate on the whole feature. On 5 August 2026 it launched Contract Intelligence, which queries signed agreements in plain English with cited answers across repositories including Google Drive, SharePoint, OneDrive, Dropbox, and Ironclad — infrastructure it bought, acquiring backend engineering firm Jetify in a deal dated 1 June 2026 and reported on 10 August by Legaltech News.
Wordsmith moved toward the document. Its homepage schema now describes the product as an “AI platform for in-house legal teams to draft, review, research, and manage contracts” that “works natively inside Microsoft Word,” and its published feature list leads with contract review, drafting, and research. Intake is one nav item among eight.
So both do intake and both do redlines. What survived the convergence is narrower and more decision-useful: who assembles the front door, and what the meter counts.
Where GC AI wins
Where Wordsmith wins
#ask-legalchannel ships with auto-routing by request context — privacy questions to the DPO, commercial questions to Commercial Counsel — conditional escalation rules, and two operating modes: Co-Pilot, which holds every answer for a lawyer’s approval, and Autopilot, which answers low-risk queries instantly. GC AI’s equivalent is a Zapier workflow you build and own.Pricing reality
This is where the decision actually gets made, and the two vendors are asymmetric in a way that matters more than the headline numbers.
GC AI publishes its floor and its per-unit rate. Four lawyers on Individual is $2,000/mo, or $24,000/yr. Team and Enterprise are annual and quoted. Business-side usage then adds $5 per API chat message on top. Run the arithmetic and the crossover is stark: every 100 business-side questions per month costs the same as another lawyer’s seat. At 400 questions a month the API bill equals the entire four-lawyer subscription. At 1,000 a month it is $60,000 a year on its own.
Wordsmith publishes nothing. Its pricing page as of today carries no tiers and no numbers, only a 30-minute demo booking and a free trial, and its FAQ routes full report access to “a pro or enterprise plan.” One disclosure survives: the ROI calculator’s own disclaimer states pricing “from
{price}/seat/month” across tiers labelled Starter, Professional, and Enterprise, with the figure rendered client-side. That confirms the billing unit is a seat and the ladder has three rungs. It does not tell you the number. This also contradicts the usage-metered structure with a $200/month default budget that Wordsmith published earlier in 2026 — the model changed inside a single quarter, so treat any older quote as void.The honest read: GC AI’s cost scales with how many questions the business asks, and Wordsmith’s scales with how many lawyers you employ. Pick the axis that stays flat for you.
Both are agent-reachable now
Wordsmith’s MCP server has been its distinguishing technical claim, exposing
ask_question,upload_file,list_playbooks,list_templates, andlist_assistantsover an endpoint athttps://mcp.wordsmith.ai, with a REST API behind it.GC AI closed that gap. Its MCP server is in public beta with 20 tools across ask, files, projects, playbooks, and skills, over a single authenticated endpoint, plus OAuth 2.0 client-credentials provisioning so a machine client such as Salesforce Agentforce can hold org-scoped access. One design detail is worth checking against your access model: GC AI’s MCP requests run as the signed-in user rather than a service account, so the assistant sees exactly what that person sees and nothing more.
Treat MCP as a tie now, and decide on beta tolerance — GC AI labels its server public beta and says the tool surface will change.
Verdict
Default pick when you genuinely cannot decide: GC AI. The cannot-decide case is one where nobody has counted the inbound requests, and in that state Wordsmith’s structural advantage is unmeasured while GC AI’s is checkable today — a published seat price, a published per-message rate, and a trial you can start without a procurement cycle. The condition that flips it is specific and you will recognise it: a counted business-side queue above roughly 400 requests a month, at which point GC AI’s meter costs more than its seats and a flat per-lawyer price wins.
Two guards for either negotiation. First, on GC AI, get Agent Connectors’ plan placement in writing — the marketing pricing page lists 20+ Agent Connectors under the $500 Individual plan, while the vendor’s own documentation places them under Team as an org-gated feature. Those two pages disagree, and connectors are load-bearing for the non-lawyer use case. Second, on Wordsmith, get the seat count definition and the year-three price in the same document as year one, since the published model changed once already this year. On both sides, put the model provider, subprocessor list, and a written no-training commitment in the DPA before privileged material moves — see legal intake and matter management for the underlying model.