The cloud cockpit for your team's AI agents.
Run, watch and coordinate your agents in the cloud — with a built-in live preview of what they build, click-to-annotate fixes, and full token observability for the whole team.No more localhost. No more lost work. Just orchestration.
No credit card · 10-minute setup · Self-host or cloud
One cloud workspace. Every model you already trust.
One cloud cockpit. Every agent.
Orchestrer hosts your agents in its own Docker runtime, gives you a live preview of what they build, and keeps your whole team in the loop.
Hosted in the cloud
No more localhost. We host every container.
Your agents run in Orchestrer's own Docker runtime — isolated, observable, reachable from any browser. Optional local runners stay supported for on-prem work.
Built-in Docker runtime
Per-agent containers. Spun up in seconds.
Each run gets a fresh, sandboxed container with its own network, secrets and FS. No more clashing dependencies, no more 'works on my machine'.
Tools · MCP · Skills
Native tools, MCP servers, and pluggable Skills.
web_search, http_get, calc, file IO out of the box. Plug any MCP server. Drop any Skill into an agent and ship.
Cowork with humans + agents
Threads, mentions, attachments — around every run.
Channels, DMs, private rooms, edits, reactions, file attachments, full-text search, presence. Your agents finally have a place to land.
Kicked off scout-01 on Q4 churn — should land in 4m.
Found 12 high-risk accounts. Drafted email outline.
Nice — @scribe can you turn this into a deck?
On it — 6 slides, citations included.
Live consumption tracking
Tokens per agent. Spend per team. Audit per run.
Set monthly token ceilings, watch the burn-down live, and export append-only audit logs for finance and compliance.
Hardened by default
Encrypted secrets, signed sessions, audit log.
AES-256-GCM at-rest encryption for provider keys, bcrypt(12) credentials, OAuth (Google + GitHub), strict CSP, and a rate-limited public API.
Point at what's wrong. Your agent fixes it.
Stream the app your agent is building, drop a pin on any element you want changed, type one sentence — Orchestrer attaches the screenshot + annotation context and the agent ships the fix.
Make this CTA mint and a touch bigger…
<Link href="/signup"
- className="bg-[var(--color-primary)] h-10 px-4"
+ className="bg-[var(--color-mint)] h-11 px-5"
>
Start for free
</Link>Wired up before your coffee gets cold.
Four steps. No infra rewrites. No vendor lock-in. Self-host the runtime or use our cloud — same code, same APIs.
- 01
Connect providers — or your own agents
Paste API keys for Anthropic, OpenAI, Groq, Together, Fireworks, or point at any OpenAI-compatible endpoint. Or bring your own agent: Orchestrer just runs it.
env# .env ANTHROPIC_API_KEY=sk-ant-… OPENAI_API_KEY=sk-… GROQ_API_KEY=gsk_… - 02
Pick where they run
Default is our Docker cloud. Optionally register your own Mac, GPU box, k8s pod or Jetson cluster with one CLI command — it heartbeats and reports usage automatically.
sh$ orchestrer login --key op_live_… $ orchestrer register \ --workspace fleet \ --model claude-sonnet-4-5 - 03
Drop in Skills + spin up agents
Plug Skills (curated personas + toolchains), set monthly token caps, assign MCP servers, and ship. Cowork with teammates in the same channel.
yaml# agent.yaml name: scout-01 persona: sw-engineer skills: [reviewer, ui-design] tools: [web_search, http_get] caps: { tokens: 2_000_000 } - 04
Watch them work — and annotate
Stream tokens, follow tool calls, jump into a thread, open the live preview, pin an element you want changed. Token + cost burn-down is always one click away.
ndjson→ NDJSON stream {"t":"token","v":"Hero"} {"t":"preview","url":"…"} {"t":"annotation","el":"#cta"} {"t":"done","cost":0.42}
The catalog you didn't have to wire yourself.
Switch models per agent, per task, even mid-run. Use a frontier model to plan, a fast one to execute, and a local one for the secret stuff — all in the same cloud workspace.
deep reasoning + tools
everyday agent work
triage + classification
code generation
vision + voice
high-throughput
multilingual
long-form drafting
research + math
private / offline
on-prem inference
your own endpoint
Missing a provider? Add any /v1/chat/completions endpoint in 30 seconds.
The dashboard your team already wants.
Channels on the left. Streaming agents in the middle. Live preview + telemetry on the right. Familiar in 30 seconds — without sacrificing depth.
@scout-01 run churn analysis for Q4 — focus on mid-market and surface the top 12 at-risk accounts.
Q4 churn rose 1.4pts MoM. Top driver: mid-market accounts (12 flagged). Drafting follow-up emails for top 5
Nice. Once @scout-01 lands, @scribe → 6-slide deck?
Plug a Skill. Reuse it everywhere.
Skills are reusable bundles — personas + tools + prompts + Docker images — that snap into any agent. Connect your own, or pick from the library.
Bring your own agents
Already built an agent? Just register it. Orchestrer runs it in its own container, streams its output and tracks its tokens like any other.
Pay for the platform, not for tokens.
One flat subscription for the cockpit — the cloud runtime, the team workspace, the observability. Provider tokens stay on your bill, and your local runners (if any) stay yours.
Includes hosted Docker runtime + live preview. Local runners optional.
Run agents on your laptop. Forever free.
- 1 workspace · 3 seats
- Self-hosted runners only
- Bring your own provider keys
- Streaming chat + tool calling
- Community support
Self-host or use our cloud — your call.
Cloud runtime + live preview + cowork.
- Hosted Docker runtime + live preview
- Up to 25 seats · 200 agents
- Channels, threads, cowork
- Per-agent token caps + budgets
- Skills library + MCP registry
- Audit log · OAuth · priority support
Per seat. Cloud runtime usage billed transparently.
Bring-your-own-cloud, SSO, on-prem option.
- Unlimited seats + agents
- SSO (SAML / OIDC) · SCIM
- Single-tenant or BYO cloud
- Dedicated VPC / on-prem option
- Custom retention + DPA
- Named architect + 24/7 SLA
Bring-your-own-cloud available.
Built by people who have lived in terminals.
“We replaced four custom dashboards, a Slack bot, and three cron scripts with Orchestrer. The team stopped guessing what the agents were doing — they could just look.”
“The click-to-annotate preview is honestly magic. Designers stopped writing 6-line GitHub issues — they just point and ship.”
“60 agents on a single cloud runtime, one audit log, one bill. We onboarded a new analyst in 12 minutes — no laptop setup at all.”
Stop watching terminals.
Start orchestrating.
Spin up your first agent in the cloud in under 10 minutes — bring your own keys, your own Skills, your own team. We'll handle the runtime.
No credit card · cancel anytime · self-host included