BotBhai: lessons from automating social commerce

Social commerce has a powerful advantage: the customer can ask a question in the same place they discovered the product. It also has a painful weakness: the business has to keep answering every question by hand. BotBhai was built to protect the speed and personality of the conversation while taking the repetitive work off the owner's phone.

What changes when the inbox becomes a sales system

The first lesson was that language is product behaviour. Customers switch between Bangla, Banglish, English, images and voice notes in the same conversation. An agent that only understands perfect typed English is not useful for the real inbox it is meant to serve.

The second lesson was that answering is not the same as selling. A useful agent needs product facts, stock, variants, delivery rules, payment options and a way to capture the order as the conversation happens. Otherwise the owner still has to reconstruct the sale from screenshots.

The third lesson was knowing when to stop. Human handover, a visible stop control and clear boundaries are part of a trustworthy automation system. The goal is not to pretend the business has no people; it is to make sure people spend their time where judgement matters.

The short answer: three levels of build

BotBhai's automation model grows through three levels: Reply, Sell and Operate.

Level	Best for	What you get (short)
Reply	Pages drowning in repetitive questions	Product-aware answers to messages and comments in the page's preferred language
Sell	Shops taking orders through the inbox	Reply plus image and voice understanding, order capture, order management and analytics
Operate	Brands managing multiple channels or pages	Sell plus WhatsApp, higher message volume, human handover and custom integrations

What makes social-commerce automation useful

The useful parts are the ones that connect the first question to a complete, editable order.

Feature	What it does for the buyer
Bangla and Banglish	Understand the language customers actually use instead of forcing a formal script.
Product knowledge	Answer from approved product, stock, price, delivery and payment information.
Comment reply	Move public product questions into private conversations without losing the lead.
Image and voice	Handle screenshots, product photos, voice notes and spoken addresses.
Order capture	Collect product, variant, quantity, customer, address, total and payment details.
Human handover	Pass unusual, sensitive or high-value requests to the owner with context.
Controls and analytics	Let the team stop the agent, inspect conversations and understand what is happening.