Alpha — early access. This page describes an early-stage product; features and pricing may change.
governframe

mechanism

The layer under your team's Claude — not another seat, not a new subscription.

Your team already pays for Claude. Governframe is a hosted MCP/tooling layer their own sessions connect to: shared memory, a curated tool roster per role, and one admin screen that shows who's doing what — without Governframe ever touching anyone's account or credentials.

Governframe architecture: BYO-subscription plus hosted MCP layer A team member's own Claude session connects outward to a hosted Governframe MCP layer, which exposes three components: hosted memory/KB, curated tool roster per role, and an admin isolation and audit view. Credentials never cross into Governframe; they stay inside the user's own session. Your Claude session own account · own credentials MCP connect Governframe — hosted layer hosted memory /KB layer team docs SOPs, context searchable, cited curated tool roster / role admin sets per-role access design ≠ prod DB admin isolation & audit view seats, tool access, activity — one screen

Architecture: BYO-subscription + hosted MCP layer. Credentials: never shared, never held by us. Fits between: Claude Team and Claude Enterprise. Built for: teams of 3–15 already on Pro/Max/Team seats.

$ governframe status
account_link ..... none held (BYO-subscription)
credentials ...... 0 stored, 0 requested
tenants .......... logical separation, tier: standard
band ............. 3–15 connected seats
position ......... between Claude Team and Claude Enterprise

who it's for

Small teams past the point of "everyone figures it out themselves."

Agencies, consultancies, and dev/ops shops of 3–15 people already running multiple people through paid Claude subscriptions for real client work — past the point where one person's ad hoc setup is enough, before the size where Anthropic's Enterprise sales motion makes sense. The buyer is the team lead who's already said, in effect: "I don't want everyone using it willy-nilly."

ProfileFitWhy
Small agencies & consultancies Fit Multiple people on paid Claude seats, delivering client work, no one owns "how we use AI as a team" yet.
Dev/ops shops, 3–15 people Fit Already past Team seats, not ready for (or not sized for) Enterprise's sales-led, compliance-heavy motion.
Solo operators, 30+ seat orgs Not (yet) a fit Below this band there's nothing to govern yet; well above it, Anthropic Enterprise is the better-built answer.

spec

Three pieces, each independently real — bundled the way nobody else has yet.

ComponentWhat it does
Hosted memory / KB layer Your team's own docs, SOPs, and project context — searchable and cited, not re-explained to a fresh session every time.
Curated tool roster, per role The admin decides what each role can call — design doesn't get production-DB access, ops doesn't get billing export.
Admin isolation & audit view One screen showing seats, tool access, and activity — and an honest statement of what "isolated" does and doesn't mean at this price.

pricing

Priced per team, not per seat — the layer is shared infrastructure.

TierPriceIncludes
Starter $149/mo Up to 5 connected seats · hosted memory/KB MCP · curated tool roster (read-only admin view)
Team $299/mo Up to 15 connected seats · full orchestration/skills layer · admin isolation view · usage/audit log
Custom Talk to us 15+ seats or white-label resale
12–299%
wide range for the add-on to a team's already-committed Claude seat spend, not a new budget line — wide because status-quo Team spend itself spans ~12x across the target band
~4 hrs
total team time/month the Team tier ($299) needs to save to pay for itself, at a $75/hr blended rate (unverified)
~25 min
per person, per month — the actual breakeven bar on a 10-person team
Status-quo Claude spend+ Governframe (12–299% add-on)

Worked from a $100–1,250/mo status-quo Team-seat range and $75/hr blended labor rate. Full math is retained in the provenance record. figures cite claude.com/pricing; the $75/hr rate and the 12–299% framing are / (unverified) as marked

walkthrough

What using it actually looks like — simulated, not live.

$ governframe demo --scenes 3
scene 1: team member connects their own Claude session
scene 2: knowledge-base query answered from the team's own docs
scene 3: admin's isolation / audit view

status: SIMULATED — nothing calls a live system
Open the simulated walkthrough →

limits & faq

The five hardest questions, answered straight.

Isn't this just reselling Claude access?

No. BYO-subscription means your own Claude session/account connects to our hosted layer — we never touch, hold, or operate your credentials. Anthropic's own developer docs name exactly this pattern (third-party companies running remote MCP servers a user's own session connects to) as a documented, observed-in-the-wild shape — not a confirmed blessing in so many words, but a real precedent, not a guess.

Why pay for this when I can clone a free GitHub "agent roster" repo myself?

You can — and that proves the demand is real. But that's a config artifact you maintain yourself, forever, alone, with no shared memory layer, no governance/audit view, and no isolation design across a team. We sell the running, maintained, governed service — the free repo is the honest ceiling on what one technically confident person can DIY.

Is our data actually isolated from your other clients?

Honestly: logical per-tenant separation at this price tier, not dedicated infrastructure or a formal compliance certification. Composio and Glama — the closest named competitors — both reserve real (SOC2/VPC-level) isolation for their enterprise tiers too. If you need that level, we'll tell you this isn't the right fit yet rather than imply otherwise.

Why not just buy Claude Team or Enterprise instead?

Team ($20–125/seat/mo) buys governance controls — SSO, admin, billing — but no shared data/memory/orchestration layer. Enterprise adds compliance depth but is a sales-led, seat-minimum motion most 5–15 person teams won't clear and don't need yet. We sit in that gap: the layer on top of Team, without jumping tiers or building it yourself.

What's actually yours to sell here versus just repackaged Claude?

The hosted memory/KB layer (a real, independently priced category — see mem0's pricing), the curated/governed tool roster (Composio and Zapier MCP prove tool-calling access is independently sellable), and the admin isolation/audit view across a team (nobody bundles this with the other two at small-team pricing today). What's NOT ours to sell: the model itself — that's Claude's, through your own subscription.

get in touch

Interested? Book a walkthrough.

Contact routing isn't live yet — this is a concept page. Placeholder only.