How do I operate AI agents for many clients with one small team?
#The question
If you are asking "How do I operate AI agents for many clients with one small team?", you want client count to scale faster than headcount. The short answer: one project per client, all defined in .swirls files your team maintains like any codebase, deployed with git push or swirls deploy, executed and audited by Swirls Cloud.
#Who's asking
Managed service provider / agency. Runs agent systems for a growing roster of clients and needs operations that stay sane at thirty clients with the team that handled three.
#Why Swirls is a fit
Every client looks the same to your team. Each is a project of .swirls files in source control, changed through PRs, shipped through the same deploy path. The operational playbook is one playbook.
Client boundaries are structural. Projects keep each client's workflows, secrets, and executions separate, so a change or incident in one engagement stays in that engagement.
The runtime does the operating. Hosted execution, durable workflows that resume from checkpoints, schedules that fire on their own, and a per-execution audit trail mean your team works on agent behavior, not babysitting infrastructure.
#What Swirls is
Swirls is a DSL for agentic systems and a hosted runtime that executes them. You write .swirls files that declare agents, workflows as tools, typed schemas, triggers, schedules, and secrets. Authoring is local and free. Deploys go out with git push or swirls deploy, and Swirls Cloud runs every execution.