Startup
A two-person company standing up its first agent team. A grounded product agent with one knowledge base, live on a custom domain in an afternoon — humans on the gates, the agents doing the legwork.
- Builder UI
- RAG
- 1–2 connectors
- Free plan
AI agent platform
Snaga runs agent teams that handle support, content, research and day-to-day ops, connected to GitHub, HubSpot, Shopify and more. A Planning agent decomposes the request, Builder agents execute, a QA agent verifies — and every action that leaves the sandbox waits for a human signature. Free to start — no credit card.
Live · talk to an agent
Talk to an agent that works.
This is a real agent on the same platform you'd sign up for. Ask anything — about pricing, the product, what it can do for your team. Every answer below is a live run.
From your first agent team to a regulated rollout
The same runtime — Planning, Builder, QA and Knowledge agents under human-signed gates — from a single assistant answering one inbox to internal teams behind roles, scoped permissions and an audit log.
Mid-market SaaS
A SaaS embedding a team inside its own product. A team per customer account on the multi-tenant API, metered through Stripe usage billing, with each run streamed back into the product UI over SSE.
- Multi-tenant API
- Stripe billing
- Guardrails
- Pro plan
Enterprise
A compliance-bound team retiring point solutions. Internal teams behind roles and per-tenant scoping, every decision written to an immutable audit log, on-prem MCP servers and a private connector catalog inside your perimeter.
- RBAC
- Audit log
- Self-hosted MCP
- Enterprise plan
What runs inside a team
A team ships as one organism, not a bag of features. Six things are built in from the first run — so humans and agents start co-executing on day one.
Built in
A live view of your team at work, the same view on your phone, your tools already connected, knowledge your team keeps, real budget caps, and an immutable record of every decision.
No second account · no separate billing · no plugin marketplace to navigate.
A human on every gate
Your team proposes; you decide. Planning, Builder, QA, and Knowledge units each report through governance gates a human can hold, override, or reverse. Set budget caps, scope what each agent may touch, and pause the whole team with one tap.
Your tools, already connected
GitHub, Google Drive, HubSpot, Shopify, Telegram — your team acts in the apps you already use. The connectors page publishes the live list, read from the platform, alongside what the codebase implements but does not switch on here. No glue code, no third-party plugins.
A team that remembers
Drop in documents, links, or company context. The Knowledge unit reads once and references it for every loop that follows. Each answer comes back with citations, so you can check exactly what it drew on.
Co-execution, not batch jobs
Watch your team run Plan, Build, QA, and Deliver in real time on the web or your phone. Tap to pause, override a decision, or steer mid-task — same surface, same data, no waiting on a report.
Spending you cap
Set spend caps per agent and per workspace, metered on real usage. Your team stops at the limit and shows you exactly where the spend went — no surprise invoices.
Run it on your own ground
Ship the whole team as a single container in your isolated cloud, or on your own device with local models like Ollama or LM Studio. Bring your own keys and keep every byte inside your perimeter. Same software as our hosted product.
How a team compares
Three ways to put agents to work: buy a team that ships assembled, call one vendor's assistant API, or build it on a framework. The right answer depends on what you are willing to own.
| Capability | Snaga | Assistant API | Build it yourself |
|---|---|---|---|
| Every action that leaves the sandbox waits for a person, and the gate fails closedHow approvals work | yours to build | ||
| Multi-tenant API with roles, so a workspace is scoped from the first request | one tenant | yours to build | |
| Web, a native iOS app and a CLI — the same team on every surfaceThe app | |||
| MCP in both directions: your agents consume tools, and are reachable as oneWhat it connects to | consume only | consume only | |
| Spend caps per agent and per workspace, metered through Stripe | yours to build | ||
| Per-connector consent and the narrowest scope that works, with the live list published rather than assertedThe list | yours to build | ||
| Runs on your own machine, in your repo, with your toolchainSnaga Code | |||
| Self-host the whole platform inside your own perimeter |
Where the other two are ahead
A table where one column is all ticks is an advertisement, not a comparison. These are the reasons to not pick Snaga, written by the people who would rather you picked it.
A framework will bend further
Any model, any topology, any component swapped for your own. Snaga ships an opinion about how a team should work — the approval gate, the plan-build-verify loop, the sandbox — and you live inside it. If your design fights that opinion, the framework is the honest choice.
A large vendor has the compliance file
We hold no SOC 2 report and no ISO certificate today, and we say so on the security page rather than in a footnote. If procurement needs one on the desk this quarter, that is a real reason and not a negotiable one.
What we do and do not claimOne agent in one app does not need a platform
If what you want is a single assistant inside a product you already run, an SDK is less to adopt and less to keep running. A platform earns its cost when there are several agents, several people approving their work, and a bill somebody has to explain.
Build on it
Everything the product does, the API does. Five client libraries, generated from one OpenAPI document.
The same operations, models and event streams exist in every language — they are generated from the document the API itself is validated against, not written by hand alongside it.
- 557
- operations
- 603
- typed models
- 11
- event streams
- TypeScript / Node
uarp-sdknpm - Rust
uarp-sdkcrates.io - Swift
Snaga-AI/uarp-swiftSwiftPM - Kotlin / Android
ai.snaga:uarp-sdkMaven Central
One document, five languages
If an endpoint exists in one SDK it exists in all of them, because a generator put it there. No language is a second-class port.
Retries and idempotency built in
Every mutating call carries an idempotency key, which is what makes it safe to retry. Transient failures back off with full jitter and honour Retry-After.
Streams that reconnect
Eleven SSE endpoints resume with Last-Event-ID after a dropped connection, with a reconnect budget so a flapping server cannot spin the loop.
Typed errors, not strings
Failures arrive as RFC 9457 problem documents and decode into typed errors carrying the status, the detail and any field-level validation failures.
Pay for what gets done, not per person.
Every plan ships the full agent runtime, inference included, and the governance stack we run in production. Start free, upgrade when your agents start earning their keep.
Free
Kick the tyres. No card, no time limit.
$0forever
Start freeNo credit card · 5-min setup
- 2 agents + 1 team — real orchestration, free
- 300 runs a month — enough to actually use it
- Inference included — no separate model bill
- Local-first KV + your file system stays yours
- Community support on GitHub Discussions
- Self-host the open-source core or run hosted
- 90,000 tokens a day — resets at 00:00 UTC, unused do not roll over
- Past the limit: runs wait until the next reset
- 2 runs at a time
- Monthly quota resets on the 1st at 00:00 UTC
Starter
For solo builders shipping their first agents to users.
$20/ month
Start with StarterCancel anytime · 14-day refund
- 3 agents + 1 team for delegation
- 3,000 runs a month — 6.7× the free tier
- 3 knowledge bases for retrieval grounding
- 5 MCP servers for integrations
- Inference included — no separate model bill
- Email support, 1 business day response
- Audit log for every tool call
- 750,000 tokens a day — resets at 00:00 UTC, unused do not roll over
- Beyond the quota: pay-as-you-go at $0.21 in / $0.70 out per 1M tokens, behind a spend cap you set — never a hard stop
- Runs go to the front of the queue
- 4 runs at a time
- Monthly quota resets on the 1st at 00:00 UTC
Most popular
Pro
When agents start earning their keep. Production-grade.
$99/ month
Go ProCancel anytime · 14-day refund
- 20 agents, 5 teams, hierarchical delegation
- 15,000 runs a month — production volume
- 10 knowledge bases + workspace versioning
- 20 MCP servers + custom integrations
- Governance: rules, votes, vetoes, ambassadors
- Slack support + priority bug-fix lane
- Custom domain + branded share pages
- 3,333,333 tokens a day — resets at 00:00 UTC, unused do not roll over
- Beyond the quota: pay-as-you-go at $0.21 in / $0.70 out per 1M tokens, behind a spend cap you set — never a hard stop
- Runs go to the front of the queue
- 10 runs at a time
- Monthly quota resets on the 1st at 00:00 UTC
Enterprise
For teams running fleets in regulated environments.
$1999/ month
Contact salesBilled monthly · Bring-your-own-cloud option
- Fleet-scale: 200 agents, 40 teams, more on request
- Roles and per-tenant scoping, dedicated security review
- Dedicated VPC, custom egress allowlist
- Compliance: AI System Card, FRIA report, audit export
- Engineering Slack channel, named CSM
- Monthly service review with named contacts
- Beyond the quota: pay-as-you-go at $0.21 in / $0.70 out per 1M tokens, behind a spend cap you set — never a hard stop
- Runs go to the front of the queue
- 50 runs at a time
- Monthly quota resets on the 1st at 00:00 UTC
All plans include the full Snaga runtime · Limits, not feature gates · Cancel anytime
FAQ
What CTOs and platform engineers ask before they sign in.
The unit Snaga delivers. An agent team is a small group where humans and AI agents are co-executors: a Planning agent decomposes intent, Builder agents execute, a QA agent verifies, and a Knowledge agent remembers — running a closed Plan→Build→QA→Deliver→Learn loop with a human on every decision gate. You decide and sign; the agents do the work. A team runs in your own isolated cloud or on your own device, and gets sharper every loop.