Health agents as code.

Or no code at all.

One script tag.

index.htmlHTML

<script src="https://loops.ollie.health/embed.js" data-agent="YOUR_AGENT_ID" async></script>

One POST starts an agent.

terminalShell

curl -X POST https://api.ollie.health/loop/$ORG_ID/$LOOP_ID \
  -H "Authorization: Bearer $OLLIE_LOOPS_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"customerId": "cus_123", "reason": "reorder-due"}'

From your terminal.

terminalShell

curl -fsSL https://raw.githubusercontent.com/Ollie-Health/Health-CLI/main/scripts/install.sh | sh
health init sleep-agent --channel whatsapp
health code --hosted --prompt "a protocol agent that checks in nightly and hands side effects to a person"

An agent is a class.

programme-agent.tsTypeScript

export class HealthLoop extends Agent {
  async onChatMessage({ messages, ctx }) {
    const user = await ctx.USER.getUser({ id: ctx.body.userId });

    // Another loop, offered to the model as a tool.
    const checkIn = ctx.workflow({
      loopId: "scan-check-in",
      description: "Start a wellness scan check-in for this user",
      inputSchema: z.object({ userId: z.string() }),
    });

    const result = await ctx.OLLIE.chat({
      thinking: "medium",
      messages,
      user,
      tools: ctx.tools({ loops: { start_check_in: checkIn } }),
      options: { useLatestScan: true },
    });

    return result.toUIMessageStreamResponse();
  }
}

Questions

What language are agents written in?

TypeScript, or none: describe it and it drafts.

How is an agent different from a model API?

It keeps state for days, waits, retries and records every step.

Which models?

Open-weight models you pick per agent.

Can we test before going live?

Yes. Playground, mock scans and SMS test mode.