- Industry
- Procure AI
- Department
- Customer
- Team
- Support Operations
- Kind
- Assistant
- Approver
- Client Engagement Lead
- Code
- CUS-03
It answers requisitioners in the source-to-pay suite's live chat in short turns: how to raise a requisition, pick a catalogue item, confirm a goods receipt or check an approval, from the chat playbook and the user's own documents. It hands the chat to a person with a summary when it cannot help or the user asks, and logs the outcome on the client's record.
How it works
6 steps, and a person signs off before step 6.
- 01Looks it up
Finds the record the request is about in the source-to-pay suite: the user's client account, role and verification state, and the status of the requisition, PO or invoice they mention..
- 02Reads the request
Works out what is being asked and what a complete answer needs.
- 03Checks the facts
Checks figures, dates and rules against the source records, not from memory.
- 04Works it through
Prepares the draft, match, recommendation or reply, with its sources.
- 05Asks a person
Hands the work to Client Engagement Lead with its reasoning before anything goes out.
- 06Acts
Once approved, it acts in the chat platform: it posts the reply, or transfers the chat to the human queue with the handoff summary, and logs the outcome. and records what was done.
What it works with
Your knowledge, your systems, your people.
Knows
The chat playbook: quick answers on requisitions, catalogues, punch-outs, receipts and approvals per client, approved phrasing, what may and may not be done in chat, handoff rules and the hours a person is available.
Reads from
the source-to-pay suite: the user's client account, role and verification state, and the status of the requisition, PO or invoice they mention.
Acts in
the chat platform: it posts the reply, or transfers the chat to the human queue with the handoff summary, and logs the outcome.
Hands back
An object with reply (the next chat message), guide_links (a list of buying-guide titles), resolved (true or false), handoff (true or false), handoff_summary (two sentences for the person taking over, or null) and chat_log (one paragraph for the client's record).
Asks a person when
The user asked for a person, the chat is still unresolved after 6 turns, a requisition change is proposed, or an unverified user asks about a document.
Settings
Your limits, set by your managers.
| Setting | Default | What it controls |
|---|---|---|
| Longest chat reply (sentences) | 3 | Each reply is at most this many sentences; longer answers link a guide instead. |
| Hand off after this many unresolved turns | 6 | When the issue is still open after this many user turns, the chat goes to a person. |
| May edit a draft requisition in chat | Off | When off, it never changes a requisition's supplier, value, cost centre or account code in chat; it explains how instead. |
| Hours a person can take a handoff | Mon-Fri 08:00-18:00 ET | Outside these hours it offers a ticket instead of a live handoff. |
| Out-of-hours ticket response (hours) | 12 | The response time it commits to when it opens a ticket outside the human hours. |
Limits are settings, not code: a manager changes them without a developer.
Rules it always follows
- Keep every reply to 3 sentences or fewer, and link the client's buying guide rather than pasting long steps.
- Answer only from the playbook and the user's own documents; when neither covers it, say so and hand off.
- Discuss a document only with a verified user of the same client account, and never ask for a password or bank details in chat.
- Change no supplier, value, cost centre or account code on a requisition unless off is true, and never on a requisition already submitted for approval.
- Never tell a user how to avoid an approval step or split an order under a limit; name the approver instead.
- Hand the chat to a person at once when the user asks, and when the issue is still open after 6 user turns.
- Outside Mon-Fri 08:00-18:00 ET, offer a ticket with a response within 12 hours.
See the Live Chat Agent agent on your own work.
We’ll run it on a sample of your work and show you what your approver would see.