What we build03 of 12Chats14 things we build
Customers wait hours for a simple answer.
Every WhatsApp, chat and mail is answered from your own prices and stock, and your team takes over on the same thread.
Tested on twenty of your own past chats before it goes live, and kept working as AI changes.
Eashan Gupta, founder. You meet him on the first call.
All 14
Everything we build for chats and messages.
Each one opens with what we build, three examples and what we don't promise.
A sample chat on a phone: a weekend enquiry answered inside the firm's rules, and a viewing booked.
-
Customer chats answered, day and night
An AI agent answers customers on WhatsApp and Instagram, day and night, from your real prices, stock and bookings.
What we build
- An agent on your official business number and Instagram account, in your tone.
- Answers drawn from your live price list, stock, bookings and policies.
- Voice notes and photos understood, in Hindi, English and regional languages.
- A hand-off to your staff in a shared inbox, with the whole chat and a summary.
For example, in your trade
- Store chains. A furniture store's AI agent answers 'is this sofa available in grey and when can it come to Wakad' from live stock and delivery slots.
- Diagnostic labs. A clinic's AI agent tells a patient the next open slot with the orthodontist and books it from the chat.
- Education. A coaching institute answers batch timings and fees in Hinglish at midnight and books a demo class.
What we don't promise. We use only the official business platform and message only people who opted in; the personal app is never automated. It books and answers from your own information and never gives medical, legal or financial advice; those questions go to a qualified person.
-
A website assistant that answers and books
Your site's assistant answers from your own pages with the source, finds the right product and books a slot.
What we build
- An assistant that reads your pages, brochures and policies and shows the source for each answer.
- A product and service finder that asks two or three questions and shows the match.
- Lead capture and booking straight into your CRM and calendar.
- A weekly list of questions it couldn't answer, so you can fix the page.
For example, in your trade
- Manufacturing. A pump maker's assistant asks the flow rate and head a buyer needs and points to the right model and its data sheet.
- Law firms. An architecture firm's site answers what it designs, where and for what size of project, and books a first call.
What we don't promise. It doesn't answer beyond your content. Outside it, it says it doesn't know and offers a person.
-
An AI support agent that fixes the problem
Refunds, cancellations, address changes and order questions are handled inside your systems, within your rules.
What we build
- Procedures written in plain words: what to check, what it may do and when to ask a person.
- Actions in your order, billing and booking systems, with a limit on each action.
- Identity checks before anything changes on an account.
- Practice runs on hundreds of past tickets before it talks to a customer.
For example, in your trade
- Store chains. A D2C brand's AI agent checks the courier status, offers a replacement for a damaged parcel and issues the return label with no ticket raised.
- Software companies. A software company's AI agent downgrades a plan, works out the refund within the policy limit and sends the confirmation.
What we don't promise. It won't resolve every ticket. Anything outside the written procedure or above the limit goes to a person, with the case summed up.
-
Every email read, sorted and drafted
Every email to your shared inbox is read, sorted and drafted in your tone for a person to send.
What we build
- Triage that tags each mail: enquiry, quote, complaint, invoice or junk.
- Drafts in your tone from your files and past replies, scored before anyone sees them.
- Attachments read and filed: purchase orders, invoices, documents.
- A rule that nothing sends without a person's tap.
For example, in your trade
- Moving. A freight forwarder's rate requests are sorted by lane and drafted with the current rate card, waiting for a person to check and send.
- Relocation. Move enquiries that arrive overnight from three time zones have a draft reply waiting when the office opens, with the missing details listed.
What we don't promise. Every reply waits for a person's tap, until you choose otherwise for a kind of mail.
-
Every channel in one inbox
WhatsApp, mail, website chat, calls and social messages sit in one place, with each customer's full history.
What we build
- One shared inbox for every channel, open source and in your own name.
- One customer record that follows them across channels.
- AI drafts and summaries inside each conversation, with your team deciding.
- Assignment, reply timers and reports by person and channel.
For example, in your trade
- Hospitality. A hotel group sees a guest's booking mail, WhatsApp check-in questions and the complaint call in one thread.
- Brokerages. A property manager sees every tenant's mails, WhatsApp photos of the leak and the call log in one place, with the next reply drafted.
What we don't promise. Not every platform's old history comes across. Some only allow recent messages, and we say which.
-
Quotes from your rates, on the chat they asked in
An enquiry becomes a checked quote from your rate card, sent on WhatsApp or mail once a person approves.
What we build
- An agent that reads the enquiry and asks for whatever's missing.
- Pricing from your rate card and rules, with every line explained.
- A branded quote document, ready to send.
- Approval on a phone, then sending and follow-up on the same chat.
For example, in your trade
- Moving. A transporter's AI agent quotes a Pune to Chennai part load from the rate card, and the sales head approves on his phone.
- Associations. A caterer's AI agent turns 'wedding, 300 guests, veg, Jaipur, March' into a quote with three menu options.
What we don't promise. It never quotes a price without your rules behind it. Anything outside the rate card goes to a person to price.
-
Orders, bookings and payments inside the chat
Customers browse, book and pay by UPI inside the chat, on WhatsApp or your website.
What we build
- Your catalogue or open slots inside the chat, with live stock and prices.
- A cart or booking held while the customer decides.
- A payment link through your payment provider, confirmed back in the chat.
- The order written into your system and the confirmation sent.
For example, in your trade
- Agriculture. A bakery takes custom cake orders on WhatsApp, collects the advance by UPI and sends the kitchen the order card.
- Hospitality. A homestay books two nights and takes the deposit in the chat, and the room shows as blocked on its calendar.
What we don't promise. Payments go only through a licensed payment provider, and we never hold or move customer money ourselves.
-
Customers told before they ask
Order, shipment and case updates go out on WhatsApp the moment things change, including delays and why.
What we build
- Triggers from your systems when a status changes or a date slips.
- Messages in the customer's language with the reason and the new date.
- Replies understood, so 'can you deliver Saturday instead' becomes a request.
- A count of the 'where is it' calls each update saves.
For example, in your trade
- Moving. When a truck is held at a state border, the consignee gets a WhatsApp with the reason and the new arrival time before they think to call.
- Diagnostic labs. A lab tells each patient when the sample reaches the lab and sends the report as soon as the doctor signs it.
What we don't promise. We send only the message types your customers agreed to, at sensible hours, and never marketing dressed up as an update.
-
Every complaint put in front of the right person
Angry messages, VIP customers and late replies are spotted and put in front of the right person.
What we build
- Tone and urgency read on every message, call and review.
- Rules for who handles what: VIP, legal threat, safety, refund above a limit.
- Timers that escalate when a reply is late.
- A daily list for the owner of what's still open and how long it's waited.
For example, in your trade
- Hospitality. A guest's message mentioning food poisoning goes straight to the general manager's phone with the booking details.
- Finance. Any customer mail that mentions the ombudsman or a lawyer goes straight to the compliance head.
What we don't promise. It won't catch every unhappy customer. Quiet complaints slip through sometimes, so the owner's list shows everything still open.
-
Every review answered, every bad one raised
Every online review gets a reply in your voice, and the bad ones reach a person first.
What we build
- Reviews pulled from your listings and marketplaces into one list.
- Replies drafted in your voice, specific to what the customer said.
- Low ratings sent to the branch manager with the customer's details.
- A monthly view of what customers praise and complain about, by branch.
For example, in your trade
- Agriculture. A restaurant chain replies to every review across its twelve outlets, and each outlet manager starts the day with its complaints.
- Diagnostic labs. A clinic answers reviews without ever mentioning a patient's treatment, and a doctor reads any that raise a clinical concern.
What we don't promise. We never post fake reviews or reply as a customer, and a health business's replies never mention treatment.
-
A helpdesk for your own staff
HR, IT and admin questions answered in the team chat, with leave, tickets and claims done there too.
What we build
- Answers from your HR, IT and admin policies, with the clause quoted.
- Actions: apply for leave, raise an IT ticket, book a room, claim a reimbursement.
- Private answers about pay and leave balances, only to the person asking.
- A list of questions it couldn't answer, so the policies get fixed.
For example, in your trade
- Manufacturing. Shop floor staff ask in Hindi how many leaves they have left and apply for two days from the same chat.
- IT firms. An IT firm's helpdesk resets passwords and grants tool access after the manager's tap, with no ticket queue.
What we don't promise. On pay or legal matters it answers only from the written policy. Anything beyond that goes to HR with the question attached.
-
What your customers are really saying, every week
Every chat, mail and call is read for themes, lost-sale reasons and gaps, in one weekly page with quotes.
What we build
- Every conversation grouped by topic, product and outcome.
- Reasons for lost sales and complaints counted, with real quotes.
- Questions customers ask that your website or team can't answer.
- A one-page weekly read for the owner and each team head.
For example, in your trade
- Education. An institute learns that most parents who didn't enrol asked about hostel safety, and adds a page and a visit slot for it.
- Software companies. A SaaS company finds the invoice line behind most of its billing chats and rewrites it.
What we don't promise. It never claims a reason it can't show. Every theme links back to the conversations it came from.
-
Ready for your customers' AI assistantsNew
When a customer's own AI assistant calls or messages to book or ask, your business answers it safely.
What we build
- A door for other people's agents: what they may ask, see and do.
- Checks on assistants that identify themselves under the new open standard, with proof of what the customer allowed; a caller who won't say whether it's an AI goes to a person.
- Bookings, quotes and status through that door, with a record of each.
- Your website and phone line kept working for people exactly as they are.
For example, in your trade
- Hospitality. A traveller's assistant asks a hotel for a quiet room with a late checkout, and the hotel's door confirms and books it within the guest's limit.
- Law firms. A buyer's assistant books a strategy session with a consulting firm through its open door, without anyone filling in a form.
What we don't promise. The standards for this are brand new, so we don't promise every assistant will work with it; we build to the open ones and keep up.
-
Your business inside the AI apps people useNew
Customers find, ask and book you from inside the AI assistants they already use, with your prices and rules.
What we build
- An app for the big AI assistants, built on the open standard they share.
- Your live catalogue, availability and booking inside that app.
- Sign-in and payment handed to your own systems.
- Listing, review and updates handled for each assistant's app directory.
For example, in your trade
- Brokerages. A buyer asks an AI assistant for two bedroom flats near a Noida metro station under 80 lakh, and a developer's app shows its live inventory.
- Hospitality. A traveller planning a Goa trip books a boutique hotel's room inside the assistant without opening its website.
What we don't promise. Each AI app decides who gets listed and how answers are ranked, so we don't promise placement or traffic.
Most owners don't know
What changed in how customers reach you.
A customer's AI assistant asks for a table for four, and the slot is taken in your diary.
It works today. We keep it working tomorrow.
How it works- Work it out, most consultants stop here
- Build it, most software firms stop here
- Run it
- Answer for it
- We stay
Tell us where your customers wait for a reply.
In 45 minutes with Eashan you leave knowing what we'd build first, and why. Nothing to sign.
What stays yours
- Your accountsin your company's name
- Your codeevery line of it
- Personal dataabout your customers and staff, on your own machine unless you approve otherwise in writing
- Plain files and a how-to guideif you stop
Eashan Gupta, founder. You meet him on the first call.