Adapt
Verified Oct 15, 2026 ActiveCompany-wide agent platform reachable from Slack, email and iMessage, with adaptive model routing and an isolated computer per run.
Listed on EffectHub ·
Pricing
- Pricing model
- Freemium
- Paid options from
- $14/seat/mo
- Billing
- Per seat, monthly
Open Source is $0 on your own hardware with your own model keys. Managed is $14 per seat per month plus pay-as-you-go compute ($0.045 per vCPU-hour, $0.0175 per GiB-hour, 50 GB storage and 50 GB egress included) and a 5.5% markup on inference bought through Adapt's keys (0% with your own); it has no free tier and needs a card. Enterprise is custom with committed-use rates.
Use tool ↗About Adapt
Adapt runs one agent for a whole company and answers wherever staff already work: Slack, email, iMessage and a web app, with more channels added as plug-in surfaces. Each run gets its own isolated computer, which can live on Adapt's cloud, Modal, Vercel, Daytona, Browserbase or your own account, and an adaptive router picks between frontier models, open-weight models and your own keys per task. Tools connect through API, MCP or CLI with identity and permissions inherited from your identity provider, and a policy store gates named actions behind approvals and per-team budgets. The runtime, CLI and templates are published as open source for self-hosting; the managed cloud is priced per seat plus metered compute, and an enterprise tier adds SSO, SCIM, audit export and VPC or on-prem deployment. The company is Adapt.com, Inc., a Texas corporation that calls itself made in San Francisco.
- API
- Yes
- Last verified
- Oct 15, 2026
Verified sources
- Pricing page: Open Source $0 self-hosted; Managed $14 per user per month plus pay-as-you-go compute at $0.045 per vCPU-hour and $0.0175 per GiB-hour, 50 GB storage and egress included, 5.5% inference markup via Adapt keys and 0% with own keys, no free tier; Enterprise custom with SSO, SCIM, on-prem and VPC (15 October 2026)adapt.com ↗
- Home page: surfaces Slack, web, email and iMessage; adaptive routing across OpenAI, Anthropic, Gemini, Kimi, GLM and OpenRouter; isolated computer per run on Adapt, Modal, Vercel, Daytona, Browserbase or own cloud; local open source, Adapt Cloud and on-prem deployment; open-source launch dated Jan 4 with waitlist; SOC 2 Type II; 'Made in San Francisco' (15 October 2026)adapt.com ↗
- Terms of service name Adapt.com, Inc., a Texas corporation, governed by Texas law with disputes in Austin; last updated 4/7/26 (15 October 2026)adapt.com ↗
Is Adapt legit?
We don’t label tools safe or unsafe. Below are the records we checked ourselves, each with the date we read it.
- Domain registered
- Sep 7, 1993 (33 years ago)Registry record (RDAP), registrar GoDaddy.com, LLC, looked up Sep 6, 2026
- TLS certificate
- Issued by Let's Encrypt, valid until Dec 15, 2026Seen in our own connection to the site on Oct 15, 2026
- Last checked
- Oct 15, 2026The site answered
- Operated by
- Adapt.com, Inc., United StatesOur research, card verified Oct 15, 2026
Pros & cons of Adapt
Adapt has moved from metered credits to a conventional seat price with usage on top, and published its runtime as open source, so the decision is now self-host for free or pay $14 a seat for the managed tenant layer. The substance is still the routing layer, the per-run isolated computer and the policy store that forces approvals on named actions. The bill is predictable per head but compute and inference float with how busy the agents are, and the self-hosted path comes with community support only.
Pros
- Router chooses between frontier, open-weight and your own models per task rather than pinning one model
- Each run gets an isolated computer, with a choice of sandbox providers or your own cloud
- Open-source runtime, CLI and templates for self-hosting with no markup on your own model keys
- Policy store with approvals on named actions and per-team and per-agent budgets
- SOC 2 Type II, standard DPA and optional zero data retention on the managed plan
Cons
- Managed plan has no free tier and requires a card up front
- Seat fee is only the floor: compute, memory, egress and inference are metered on top
- Inference through Adapt's keys carries a 5.5% markup over provider prices
- SSO, SLAs and on-prem are Enterprise-only, so the mid tier lacks what larger teams usually need
Best for
- Engineering orgs that want one agent reachable from Slack, GitHub, Linear and email
- Teams that need agent runs in isolated sandboxes with scoped permissions and spend caps
- Companies with VPC or on-prem requirements, or who prefer to self-host the open-source runtime
Not for
- Individuals who want a personal chatbot
- Teams expecting a free managed tier to trial before entering a card
Watch out: Pricing changed in autumn 2026 from usage credits (Starter free, Pro $50-5,000) to $14 per seat plus metered usage; older reviews and the previous free Starter grant no longer apply.
Alternatives to Adapt
Adapt head-to-head
Side-by-side fact sheets against the competitors our research or their makers name — pricing, platforms, API and both verdicts on one page.