Overview
Quickstart
Sign up, seed the default agency, run an agent, and inspect the trace — in five minutes.
This is the shortest path from "never used ALDO AI" to "watched a multi-agent run replay end to end". You'll need a browser and (optionally) a model API key — local-only models work too.
#1. Create an account
Open the control plane and choose Sign up. Verify your email and you'll land on the welcome screen of a fresh tenant.
#2. Seed the default agency
The control plane ships with a reference organization — the one ALDO TECH LABS uses internally to ship the platform. From the welcome screen click Seed default agency, or run:
curl -X POST https://ai.aldo.tech/api/auth-proxy/v1/tenants/me/seed-default \
-H "Authorization: Bearer $ALDO_API_KEY"You now have 26 agents — principal, architect, engineers, code- reviewer, evaluator-builder, and so on. Each one is described in the agency concept doc.
#3. Run an agent
Open the Playground from the sidebar. Pick the code-reviewer
agent and paste a snippet of code. Click Run.
While the run executes, watch the live event stream on the right. Tool calls, model tokens, guard hits — all stream in order.
#4. Inspect the trace
When the run finishes, click Open run to land on the run detail page. From there:
- Timeline shows every step in order.
- Flame graph visualises duration and parallelism.
- Replay lets you re-execute any step against a different model — see Replayability.
#5. Add an eval gate
Promotion in ALDO AI is gated by eval suites. Open the agent's spec
in the registry, declare a suite under eval.regression, and use
the Sweeps page to run it. The registry refuses to promote a new
version unless the suite passes its declared threshold.
#What's next
- Building an agent — write your own from scratch.
- Capability-class routing — how the gateway picks a model.
- Privacy tier — how
sensitiveruns are physically prevented from reaching the cloud.