An AI Workforce is a team of AI crews and agents run as one system. Where a single crew routes a handful of prompts to answer one request, a workforce composes whole crews and agents — each an independently built, runnable listing — into an org that runs a whole function: a crew-of-crews under a top-level manager.
It's the top rung of the Sigrix composition ladder: prompts compose into crews, and crews compose into a workforce.
How an AI Workforce works
A workforce sits one level above a crew. A crew has a manager persona and 1–5 member prompts for a single task. A workforce keeps that cap and instead composes entire crews and agents as its units, with a top-level manager coordinating them.
- Units are whole listings. Each unit is a published crew or agent that already runs on its own — so it stays separately sellable, reviewable, and updatable. A workforce is not "a bigger crew"; it's a crew of crews.
- A manager coordinates. A top-level manager routes work across the units and, where the job needs it, hands one unit's output to the next.
- It runs a function, not a task. A crew answers a request; a workforce runs a role — a support desk, a research operation, a content team — end to end.
The four ways units work together
A workforce (like a crew) can orchestrate its units in a few shapes:
| Mode | Who works | Good for |
|---|---|---|
| Route | the manager picks the one unit that fits | a mixed inbox — send each item to the right team |
| Sequence | units run in order, each building on the last | a pipeline — research → draft → edit |
| Parallel panel | several units answer the same input at once, then a synthesizer combines them | a review board — QA and architect both weigh in |
| Hierarchical org | the manager delegates across sub-crews | a whole department run as one system |
Route and sequence each answer one request; the parallel panel is the one people miss — every unit answers at once and their answers are merged, not one picked.
AI Workforce vs a crew vs an agent
- An agent runs one task for you, autonomously, and hands back a finished result.
- A crew routes 1–5 prompts through a manager to answer one request across a small bench of specialists.
- An AI Workforce composes whole crews and agents into a system that runs a whole function.
Reach for a crew when a handful of specialists cover the job; reach for a workforce when the job is a whole role made of several teams. Both sit on top of the 5 AI building blocks.
AI Workforces on Sigrix
On Sigrix an AI Workforce listing is a composed listing: it names the published crews and agents it's built from, and buying the workforce entitles every one of them — each sub-team lands in your library, and each unit's seller is paid when a workforce that includes them sells. The deliverable is a runnable project: the composed crews become sub-crews under a top-level manager, wired to run in the mode the workforce was built for, and you run it on your own model key — the same bring-your-own-key posture as a crew. Every listing shows a captured real run and an estimated token cost, so you can see it work before you buy.
The AI Workforce is the newest and largest tier, and it's rolling out gradually. While it opens up, the crews and agents that compose one are already here to browse.
FAQ
What is an AI Workforce? An AI Workforce is a composition of whole AI crews and agents — each a runnable listing in its own right — coordinated by a top-level manager to run a complete function, like a crew-of-crews or an AI org. It is the tier above a single crew on Sigrix.
How is an AI Workforce different from a crew? A crew routes 1–5 member prompts through one manager to answer a single request. A workforce does not raise that cap — it composes entire crews and agents as its units, so each stays independently sellable and reviewable, and a manager coordinates across them to run a whole function.
What do I get when I buy an AI Workforce? You get an entitlement to every crew and agent the workforce composes — each lands in your library — plus a runnable crew-of-crews you deploy on your own model key. Each unit's original seller is credited when the workforce sells.
Can the teams work in parallel? Yes. Alongside routing and sequencing, a workforce can run a parallel panel: several units answer the same input at once and a synthesizer merges their answers — the pattern you want when, say, a QA reviewer and an architect should both comment, not just one.
New to the layers below it? Read what an AI agent is and the 5 AI building blocks, or browse the crews and agents a workforce is built from.
One practical essay, every other Tuesday.
Patterns, sample listings, and short essays from the Sigrix editorial team. One-click unsubscribe, no noise.
0 Comments
Join the discussion. Be kind and stay on topic.