Appearance
Agent Chat
Ask the ErgoPlanner agents a question and watch the answer arrive as it is written.
Open the chat with the Ask a chatbot button at the bottom right of any page (or go to /chatbot). The left rail lists your conversations, newest first; New chat starts a fresh one. A conversation is only saved once you send its first message.
Sending a message
Type in the composer and press Enter (or Send). Shift+Enter adds a new line.
Each message is routed to the agent that can answer it. Sizing, compliance, scheduling and similar requests go to a specialist agent; anything else is answered by the general ErgoPlanner assistant. The name under each reply says who answered.
Streaming replies
Replies stream in: the text appears token by token while the agent is still writing, with The agent is typing… under it. Before the first words arrive you see The agent is replying… instead.
- A specialist agent's answer, and an answer from a provider that cannot stream (for example a self-hosted command-line model), arrives all at once when it is ready.
- While a reply is streaming the Send button becomes Stop.
Stopping a reply
Press Stop to cut the reply short. The agent stops generating straight away (the AI call on the server is cancelled, so the rest of the reply is never generated). The text you already saw stays in the conversation and is kept when you reopen it later. Your message stays too.
When there is no reply
| What you see | Why | What to do |
|---|---|---|
| AI is not configured bar | No AI provider is set up for your organisation | A SuperAdmin adds one under Admin → AI Providers (AI-Provider-Setup) |
| "Your organisation's AI allowance is used up…" | The managed AI allowance for this billing period is spent; the message is not sent | Top up under Settings → Billing, then send again |
| An error banner, your text back in the composer | The message could not be sent (network, expired session) | Fix the cause and press Send again |
| An error banner after part of a reply | The AI provider failed mid-reply | Send the message again |
Related
- AI-Provider-Setup — choosing the model the chat uses
- API:
POST /api/v1/agents/sessions/{sessionId}/messages/stream(Server-Sent Events:delta,done,error) — see the API reference