An Indian pharmaceutical contract manufacturer we work with uses an AI voice agent to call every website enquiry within minutes and sort it before a salesperson touches it. The agent first works out what the person actually wants (to buy, a job, or something else), then asks only the questions that matter for that path. Buyers are asked product segment, business model, quantity and market, and tagged hot, warm or cold. Job seekers are routed to HR. The sales team now calls qualified buyers, not a mixed inbox. This is the AI sales agent applied to B2B manufacturing.
Build a triage agent for your enquiries: Start free on Edesy — Rs 50 free credit, no demo needed.
The problem: the enquiry form is a mixed bag
A manufacturer's "Contact us" form collects everything: distributors, pharmacy owners, hospital buyers, export agents, job seekers, students and vendors. The sales team was spending the first part of each day calling people who were never going to buy. Genuine buyers waited, and some went elsewhere.
Step 1: find out why they enquired
The first question decides the whole call:
Namaste {{name}} ji, main {{company}} ki taraf se ek AI assistant bol rahi hoon. Aapne humari website par enquiry ki thi. Kya aap humare products kharidne ke baare mein jaanna chahte hain, ya job ke liye?
Step 2: the buyer path
For buyers, the agent asks one question at a time:
- Product segment: tablets, capsules, syrups, injectables, ointments or other.
- Business model: contract (third-party) manufacturing, or a franchise/distribution model.
- Quantity: expected monthly or first-order volume, in the buyer's own terms.
- Market: domestic or export, and which states or countries.
- Contact details: name, company, and a WhatsApp number for the quotation.
Then the agent tags the lead hot (clear product, volume and timeline), warm (interested, details unclear) or cold (just exploring), and the sales team is notified on WhatsApp with the summary.
Step 3: the job-seeker path
For job enquiries, the agent captures qualification, years of experience and the role they want, and passes it to HR. The call stays short and polite. These people might be your future employees.
Rules that made it work
- Never quote prices, MOQs or delivery dates. They depend on the product and change. The agent says the sales team will send exact details on WhatsApp.
- Say abbreviations the way the trade does. Industry short forms need spelled-out pronunciation rules in the prompt, so the agent reads them letter by letter.
- Say quantities the Indian way. "Ek lakh tablets", not "one hundred thousand". See pronunciation of numbers and acronyms.
- Check the phone number has 10 digits, repeat it back, and ask again at most twice.
- Stop talking after the closing line. See why AI agents hang up too early, or too late.
Language
The default is Hindi-English, which suits most Indian B2B callers. The agent switches to pure English or pure Hindi when the caller clearly prefers it. Export enquiries usually get English.
What sales receives for each lead
| Field | Example |
|---|---|
| intent | buyer |
| segment | syrups |
| model | contract manufacturing |
| quantity | about 1 lakh bottles/month |
| market | domestic, 3 states |
| lead_temperature | hot |
| captured |
Results can also go to HubSpot, Salesforce, Zoho CRM, GoHighLevel or a Google Sheet.
Why speed matters in B2B
B2B buyers usually enquire with several manufacturers at once. The first supplier to call with the right questions gets the first real conversation. See speed to lead with AI callbacks, and if your enquiries come from marketplaces, call every IndiaMART lead within a minute.
Start with last week's enquiries
Upload last week's form enquiries and see how many are real buyers. Create a free account, start from the lead qualification template, or talk to us.