Quickstart

From zero to a deployed Echo agent in five minutes.

1. Open the canvas#

Visit your tenant's ScaiFlow canvas (your operator will give you the URL — typically https://scaiflow.<your-domain>). Sign in with ScaiKey.

A fresh canvas opens with an empty flow named Untitled Flow.

2. Drop an entry and an LLM block#

• From the left palette under Entry points, drag API Entry onto the canvas.
• From LLM blocks, drag Flexible Prompt onto the canvas to the right of the entry.

3. Wire the edge#

Drag from the API Entry's out_request port to the Flexible Prompt's in_message port. ScaiFlow creates a sequential edge.

4. Configure the LLM step#

Click the Flexible Prompt node. In the right-side property panel:

• Goal: Echo the user's message back verbatim.
• Model role: primary (the default — comes from the flow's model registry).
• Output schema: leave as {} for now.

5. Save + deploy#

• Click Save. The flow is persisted to your tenant.
• Click Deploy. ScaiFlow compiles the flow to YAML and POSTs it to ScaiGrid. After a few seconds you'll see a toast: Deployed core_id=core_abc123.

You now have a running ScaiCore Core. To invoke it:

1
2
3
4

curl -X POST "https://scaigrid.scailabs.ai/v1/modules/scaicore/cores/${CORE_ID}/invoke" \
  -H "Authorization: Bearer ${SCAIGRID_TOKEN}" \
  -H "Content-Type: application/json" \
  -d '{"input": {"message": "hello"}}'

Make it a chat model#

In the flow's property panel (click empty canvas to deselect), check Publish as chat model. Save + deploy again. The Core is now reachable via the OpenAI-compatible chat API at slug scaicore/<your-tenant>/echo-agent:

1
2
3
4
5
6
7

curl -X POST "https://scaigrid.scailabs.ai/v1/chat/completions" \
  -H "Authorization: Bearer ${SCAIGRID_TOKEN}" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "scaicore/'"$TENANT"'/echo-agent",
    "messages": [{"role": "user", "content": "hello"}]
  }'

What just happened#

The canvas serialized your two-node graph to a FlowGraph JSON. The backend compiler turned that into a YAML manifest with one api trigger and one llm_turn step. The deploy step POSTed that YAML to /v1/modules/scaicore/cores on ScaiGrid, which returned a core_id. The optional publish step registered a FrontendModel against your tenant.

Read Concepts: Architecture to understand the pieces that just moved.

Next steps#

• Tutorial: Customer support flow — branch on intent, call a knowledge plugin, route to a human reviewer.
• Tutorial: HITL approval — pause execution at a checkpoint and resume via ScaiQueue.
• Reference: REST API — automate flow management from your CI.