Chat once. A whole team mobilizes.
Every conversation in Orchestror can spawn researchers, writers, and auditors — each reporting back live in the thread, each waiting at the gates you define.
Request Accessgsc-opportunities1,240 queries analyzedspawn: seo-analystscoring 14 candidatesspawn: content-writerrefreshing /ac-repair-cost/Live delegation
Ask for a content refresh and watch the orchestrator spawn researchers, writers, and reviewers — every tool call and API response streamed into the thread.
Permission gates
Risky actions pause the run and wait for your Approve or Deny. Disconnect or timeout defaults to deny — nothing slips through unattended.
Verified output
Every claim an entity makes is checked against the database after the fact. "Published to WordPress" without a matching run gets flagged, not buried.
Entity rooms
Orchestrator, Strategy, Research, Content, and Audit rooms — each entity with its own toolset and specialty, addressable from one chat.
Full traceability
No black boxes. Every step an entity takes is visible in the thread and logged to the per-client changelog with date, action, and target.
Client context built in
Entities load your client’s config, brand voice, and work history before acting. They know what was touched last week — and won’t redo it.
- ✓File attachments — drop files, images, and documents into any message
- ✓Paste a YouTube URL → an original article in the client’s voice
- ✓Background entities — long tasks keep running behind the chat with a live progress feed
- ✓Run lookups — reference a past run by name and the entity retrieves it directly, no re-exporting or re-attaching
- ✓Multi-entity teams collaborating in the same thread
An army of entities,
at your service.
140+ specialized entities and skills, each an expert in exactly one job. One inspects how Google indexes a URL, another mines Reddit threads, another draftsreview replies or designs statistically validA/B tests. Every one is summonable from chat.
Straight answers
What exactly is an "entity"?
Any autonomous worker in Orchestror — an orchestrator that coordinates, a specialist that audits or writes, or a skill that executes one task. They all speak the same language and can delegate to each other.
Can entities act without my approval?
Only within the boundaries you define. Sensitive actions — publishing, deleting, spending — hit a permission gate and wait for your explicit sign-off. Everything else runs autonomously but fully logged.
How do I know an entity actually did what it says?
The verification layer cross-checks every claim against real evidence: database runs, files on disk, API responses. Claims without evidence are flagged in the thread automatically.
Do I need to write prompts?
No. You describe outcomes in plain language — "find quick wins on this site and refresh the top opportunity." The entities figure out the how, and you watch them do it.
Which model does an entity run on?
Whichever you pick — Claude or Codex today, selectable per entity and per workflow. An entity is its instructions, skills, and client context, not the model underneath: when a new model ships, it plugs in as one more runner. See how it works.
What do entities spend tokens on?
Thinking, not plumbing. When an entity needs data, a deterministic tool pulls and crunches it in plain code — the model only reads the computed result. The same job that costs a browser agent ~200,000 tokens costs an entity about 1,500.
Stop Prompting.
Start Orchestrating.
Tell us about your operation. We'll walk you through the platform, agree on pricing, and hand you the keys to your own dedicated instance.