An inbound agent that answers in seconds, at 3am, every time.
Most inbound leads are lost to response time, not to price. We build the agent that replies immediately over WhatsApp and email, qualifies against your own criteria, and books the meeting straight into a calendar.
You are not losing leads on price. You're losing them on latency.
Almost every business we look at has the same leak, and almost none of them have measured it. Enquiries arrive at 9pm, on a Friday, or while the two people who could answer are already on calls. The message sits. By the time somebody replies the next morning, the prospect has messaged two competitors and is deep in conversation with whichever one answered first.
The uncomfortable part is that this has nothing to do with how good your service is, what you charge, or how strong your reputation is. It is purely a function of who replied first. You can be the best brokerage in Dubai and lose a serious buyer to a mediocre one with a faster thumb.
The instinctive fix is more staff, and it doesn't work. Adding headcount extends your coverage window; it doesn't eliminate it. Nobody staffs 2am. Nobody staffs the fifteen minutes when the whole team is in a meeting. And every additional person raises your fixed cost against a problem that is fundamentally about latency, not capacity.
The correct fix is an agent that never queues. Not a chatbot with buttons, and not an autoresponder that says 'thanks, we'll be in touch', but an agent that reads what was actually written, works from your qualification criteria, answers the questions a prospect asks before they'll agree to a call, and books that call while the intent is still warm.
What happens between the message and the meeting.
Message arrives
WhatsApp, email, web chat or Instagram DM. The agent is event-driven rather than polling on a schedule, so it sees the message the instant it lands, not on the next cron cycle.
Intent and context read
It reads the actual message rather than matching keywords. A message saying 'is the 2BR in Marina still available, and what's the service charge' is understood as three things at once: a specific unit enquiry, an availability question, and a cost objection forming.
Qualified against your criteria
Budget, timeline, location, financing status, decision authority, whatever your criteria actually are. Not a generic lead score somebody else's product designed for a different market.
Objections handled from your playbook
The questions that stall a conversation, price, service charges, handover risk, contract terms, get answered from material you wrote and approved, in your positioning, not improvised.
Meeting booked
Straight into your existing calendar, with availability respected and a confirmation sent. No back-and-forth about times, which is where a surprising share of warm leads quietly die.
Escalation when it's needed
Anything outside the playbook, anything high-value, anything where the prospect asks for a person, routed to the right human immediately with the full thread attached, so nobody starts the conversation over.
Four things people assume this is.
A decision-tree chatbot
Button-menu bots fail the moment a prospect writes a real sentence. This reads free text, including a message with three questions in it and a spelling mistake.
An autoresponder
'Thanks for your message, we'll get back to you' buys nothing. The prospect still leaves. The point is a substantive first reply, not an acknowledgement.
A replacement for your sales team
It handles the first ten minutes, the part that's the same every time. Your closers get a qualified prospect who already has their basic questions answered, at the moment they're most engaged.
A generic AI wrapper
It runs on Usetta, which we built and operate on our own pipeline. When it needs to behave differently for your business, we change it. There's no vendor roadmap in the way.
Concretely, what's delivered.
- Instant first response on WhatsApp, email, web chat and Instagram, measured in seconds, not hours
- Qualification against your own criteria, not a generic scoring model designed for a different market
- Objection handling from your playbook, in English and Arabic, including mixed-language threads
- Calendar booking straight into your existing scheduling tool, with availability respected
- Human escalation rules you control: when to hand over, to whom, and with what context attached
- Full transcript logging, so you can audit every conversation the agent ever had
- A dashboard showing what came in, what qualified, what booked and what was escalated
- Ongoing tuning, because the playbook is a living document and we adjust it as you learn what converts
We run it on our own inbound first.
We run this on Anqor's own inbound before selling it. That isn't a positioning line, it's the reason we can make changes quickly, because we feel the same problems our clients do, in the same week.
The product is Usetta, a multi-tenant system in production. It handles WhatsApp Business API, email and web chat, stores conversations in Postgres, and runs its classification and reply generation through frontier models with routing to cheaper models for the simpler turns.
What this won't do.
- It will not close a complex deal. It qualifies and books; the human closes.
- It needs a real playbook from you. If your qualification criteria only exist in one person's head, the first week of work is getting them written down, and that's genuinely useful whether or not you automate afterwards.
- It is not free to run. Model inference has a per-conversation cost. For high-value leads that's trivially worth it; for very low-value, very high-volume enquiries we'll tell you if the arithmetic doesn't work.
- It shouldn't commit inventory without a confirmation step. An overbooking costs more than a slower reply.
Straight answers.
How fast does it actually respond?
Seconds. The agent is event-driven rather than polling on a schedule, so it replies as soon as the message lands. There's no queue and no business-hours gate unless you deliberately set one. Some businesses do want an out-of-hours message that sets expectations differently, and that's a configuration choice.
Will it say something wrong to a customer?
It works from a playbook you approve, and it escalates rather than guesses when a question falls outside it. Every conversation is logged in full, so if it does get something wrong you can see exactly what happened and tighten the playbook. You're correcting a document, not debugging a black box. In practice the first two weeks after launch involve real tuning, and we plan for that rather than pretending it launches perfect.
Does it work in Arabic?
Yes, Arabic and English, including threads that switch between the two mid-conversation, which is how a large share of UAE enquiries actually arrive. This is handled at the model layer rather than by translating an English response, which is a visible quality difference to a native reader.
What happens to leads it can't qualify?
They escalate to a human with the full thread attached, so your team doesn't restart the conversation. We'd rather the agent escalate too often at first and get tuned down than have it guess confidently and lose a serious prospect.
Can it work with our existing CRM?
Yes. We build around what you already run rather than asking you to migrate first, because migration is a separate decision and usually a bad one to bundle into an automation project. If you don't have a CRM, the dashboard covers the basics until you need more.
How long does it take to set up?
Two to three weeks for a working first version, most of which is defining your qualification criteria and objection playbook rather than engineering. If those already exist in written form it's faster.
What does it cost to run?
There's a per-conversation inference cost, which varies with how long conversations run and which models the routing selects. We'll model it against your actual enquiry volume before you commit, and tell you if the numbers don't justify it at your volume and deal size.
Do customers know they're talking to AI?
We recommend being straightforward about it. In practice it doesn't reduce conversion, because response speed matters far more to a prospect than who typed the reply, and being caught pretending is a worse outcome than disclosing.
Anqor is a small studio, so you deal with the person who builds it. If this isn't the right fit for your business, I'd rather say so on the first call than sell you something you don't need.
LinkedIn →See it answering, in your own words.
The fastest way to judge this is to watch it handle a real enquiry from your business. Bring one to the call.