Use cases

Not features — jobs. Each one below is written the way the agent itself works: what triggers it, what the agent does, where you confirm, and what lands back in your chat. Every capability named here is live today; the ones still rolling out are labelled at the end, exactly as the agent labels them.

Online seller

Customer questions, answered while you pack

Your customers ask in chat — prices, delivery times, "is this in stock?" — in Bangla, English, or both in one voice note. The agent drafts the answer from live information; you stay on the parcels.

The work, defined

1. Trigger — a customer messages your chat, or you forward their question to the agent.
2. The agent — understands the question (voice notes included), checks what it needs on the live web, and drafts the reply in the customer's language.
3. You confirm — anything that commits you — a price promise, a complaint resolution — waits for your yes.
4. Outcome — the reply goes out in the same chat; anything unresolved becomes a support ticket with a reference.

Research the live web Open a support ticket

Support desk

Tickets with references, not lost chats

Complaints stop being screenshots in a group chat. The agent opens a native ticket right inside the conversation, hands the customer its reference, and tracks it in the unified inbox.

The work, defined

1. Trigger — a customer reports a problem, or you tell the agent to open a ticket.
2. The agent — shows its plan, then creates the ticket with the customer's own words and returns the reference (e.g. SUP-XXXXXXXX) in the chat.
3. You confirm — opening the ticket pauses for your yes; status checks never do.
4. Outcome — the ticket lives in the unified inbox; anyone can ask the agent "where is my ticket?" and get its real status.

Open a support ticket Check your support tickets

Agency

Research and drafts your clients can approve

A client brief lands; by the time your team sits down, the competitor check is briefed with sources and the first post is drafted — sitting in the approval queue, not posted.

The work, defined

1. Trigger — you describe the brief or the competitor in chat.
2. The agent — runs the deep research or competitor pass on the live web, then drafts the post and queues it.
3. You confirm — nothing ever publishes without the owner's approval; the queue shows what's waiting, approved, published or failed.
4. Outcome — a sourced brief in chat and a draft in the queue, ready for a human decision.

Check competitors Research a topic in depth Draft a post for approval

Solo founder

A second pair of hands at midnight

The question that can't wait for morning — a price, a rule, a rival's new offer. The agent answers from the live web, or routes the bigger job to the specialised bot built for it.

The work, defined

1. Trigger — you ask, in words or a voice note, whenever it comes up.
2. The agent — answers from the live web with sources; for a specialist job, Bot Finder matches the task to the right 1AI bot and drafts the prompt to send it.
3. You confirm — reads come back immediately; anything that writes, sends or opens pauses for your yes first.
4. Outcome — the answer with its sources, or the right bot named with a ready-to-send prompt. Where a bot is still being built, Bot Finder says so — two catalogue bots are live today, the rest are labelled building.

Research the live web Find the right bot for a task

Rolling out — labelled, not hidden

These skills are registered and fenced; the agent names them as "not yet" instead of bluffing: