Case study · AI Automation
An AI enquiry-to-quote agent for a B2B distributor
A multi-agent workflow that reads incoming enquiries, checks product data and pricing, drafts a quote and keeps the CRM updated — with a person approving every quote.
The challenge
Enquiries arrive through a shared inbox, WhatsApp and the website, with product details, quantities and delivery expectations written in different ways.
The sales team has to check stock and pricing across an ERP, a product spreadsheet and a CRM before a quote can be prepared. Routine requests slow down, and CRM records are often incomplete.
The goal was not to remove the sales team from the process. It was to give them a reliable first draft, the right context and a clear exception path.
What we built
- Intake agent that classifies the enquiry and extracts product, quantity and delivery details.
- Knowledge lookup across the product catalogue, stock levels and pricing rules.
- Quote-drafting agent that prepares line items, availability and a proposed response.
- Human review step with approval, editing and confidence thresholds.
- Action agent that sends the reply, updates the CRM and creates a follow-up task.
- Exception handling for missing products, unusual pricing and unclear requirements.
How it works
Receive and classify
Email, WhatsApp and form enquiries are normalised into a single request with the original message attached.
Gather context
The workflow retrieves product data, stock and pricing from approved sources and flags anything missing.
Draft the quote
A structured quote and reply are prepared, with the source data shown next to each line item.
Human approval
A salesperson reviews the draft, edits where needed and approves it before anything is sent.
Send and log
The approved response is sent through the original channel and the CRM is updated automatically.
Results
~70%
of routine enquiries drafted automatically
< 15 min
quote turnaround
100%
of handled enquiries logged in the CRM
Technology
- LLM agent workflow
- Email API
- WhatsApp API
- CRM integration
- ERP / product data
- Human-in-the-loop review
- Audit logging
Could your process work like this?
Share the workflow, system or machine you want to improve, and we’ll help you identify a practical next step.
Book a meeting