DMs and comments arrive at all hours, and a person can only answer a slice of them. The Conversation agent answers each one using the persona and knowledge base you configured, builds it into a contact profile, and stops when your handoff conditions are met — passing that contact to sales.
A customer asks about pricing at 2am, I see it when I get in, and they're long gone
The Conversation agent is online during activity hours for the account's region and answers the settled questions from your knowledge base. Anything it can't answer is parked and flagged rather than bluffed.
Five platforms, a dozen accounts, switching between inboxes — I always miss a few
Conversations from 6 platforms land in one inbox, grouped by contact rather than by account. Someone who reached you on two platforms shows up as a single profile.
The AI replies read like a template and customers spot the bot immediately
Personas set the role and tone, phrasing libraries give consistent verbal habits, and scripts govern how a conversation advances. The same question gets a different answer from a different persona.
A salesperson leaves and three months of customer conversations walk out on their personal messenger
Conversations, profiles and tags live on the platform and the accounts belong to the company. During a handover you reassign the accounts and contact groups to someone else, and nothing is lost.
The handoff conditions and the closing stage are both yours to define. AI catches and organises; anything involving quotes, commitments or the final push goes to a person.
DMs and comments from every platform account land in one inbox. A contact's conversations across platforms sit in a single profile, so there's no switching.
Pick a persona and activity hours, import your own knowledge base, and spell out which questions must go to a human. This configuration is what decides what the AI can and cannot say on your behalf.
It only answers what the knowledge base supports, advancing along the script. It replies in whatever language the customer uses, translating when needed.
Requirements, source and stage are pulled from the conversation into profile fields and tags. Long-term memory across sessions means the next conversation picks up where the last one stopped.
Contacts that meet your conditions enter the handoff queue and the AI stops replying. Sales continues in the same inbox with the full history in front of them.
A price, a stock commitment or an after-sales condition becomes a promise the moment it's sent, and walking one back costs far more than answering a few minutes later. So the last step goes to a person: the AI asks what needs asking and records what needs recording, and sales just carries the conversation forward.
Conversations from 6 platforms in one place, grouped by contact, filterable by account, tag and status.
Answers from your own knowledge base and nothing beyond what it supports. Switch it on or off per account.
You define the triggers. Once matched, the AI stops and the contact joins the queue for sales.
Personas, activity hours, phrasing and scripts. Keep one account on one set over time and its voice stays consistent.
When a contact returns weeks later it remembers where the conversation stopped and what they asked for, so they don't have to repeat themselves.
Profile fields and the tag scheme are yours to define, with bulk tagging, tag filters and profile export.
Save your common lines as snippets and send them in one click after taking over, instead of retyping.
Read incoming messages in your own language and send in the customer's, with both the original and the translation kept on record.
The Conversation agent is software. The accounts, environments and knowledge it works from are yours to provide. We spell that out so you don't plan against the wrong cost.
Brings accounts to a normally usable state before handing them over. Persona and activity hours are shared between both roles.
Leads collected across 6 platforms enter the contact list, becoming the people the Conversation agent reaches out to.
When breaking down the goal it decides which accounts handle follow-up and which persona they use, then sends the open items to you for approval.
The same account both posts and converses. Comments and DMs from its content land straight in the inbox.
It will get things wrong. The AI answers only from your knowledge base, so if that's vague or self-contradictory the answers will be too. The real safety net is your handoff conditions: put pricing, stock and after-sales — anything that can't afford an error — on the must-escalate list, and the AI stops when it hits them. Before going live, run a batch against real past conversations, read the answers, then scale up.
There's no standard figure — it depends on how much your knowledge base covers and how tight your handoff conditions are. The dashboard shows replies sent by AI and contacts escalated; read them together. A high AI share alongside a low close rate on escalated contacts means things that should have been escalated weren't.
No. Once a contact is handed off, automatic replies for them are switched off and only people speak. Sales takes over in the same inbox with the full history and profile there. When they're done, you can manually hand the contact back to the AI for ongoing follow-up.
Structured information pulled from the conversation — requirements, preferences, the stage reached — rather than a full copy of the transcript. When the same contact returns weeks later, the AI can pick up where it left off. What's remembered is visible in the contact profile, and you can correct or delete it.
Yes. Filter profiles by tag, source or stage and export them into your own CRM. Conversation transcripts can be exported by date range too.
Technically yes — filter contacts by tag and send in bulk. But platform limits on unsolicited marketing messages vary widely, and WhatsApp is especially strict. Start small, keep the frequency down, and treat broadcasting as a supplement rather than your main channel. The account risk here comes down to your sending strategy.
A consultant will put together a concrete configuration recommendation and resource quote around your platforms, contact volume and existing messaging.