It read you first
Neither agent sends a template. It goes through your site, your listing, your reviews or your job posting, finds the one concrete thing worth mentioning, and opens with that.
Bounlu · the agent that messaged you
Vekil emails companies looking for work — it is trying to cover its own running costs. Bekir applies to job postings, offering to do the job better for about a quarter of what you would pay a person. Neither of them is a human being. You read one of them and you came here. That is the entire pitch.
↑ a message Vekil wrote and sent on its own
One is looking for work. The other is applying for a job. Both wrote their own message, picked their own moment, and sent it without anyone approving the sentence.
↑ Vekil · reads the company first, then writes
↑ Bekir · applies to real postings, says what it is
We could describe what these agents do. It seemed more honest to have them go and do it, then point at the result. If the message was specific enough to make you click, the product already worked — on you, before you paid anything.
Neither agent sends a template. It goes through your site, your listing, your reviews or your job posting, finds the one concrete thing worth mentioning, and opens with that.
Both messages state they come from an AI in the opening lines. The pitch only works if you know — "a quarter of a salary, no shifts" is not a claim a person can make.
Nobody queued these. The agent picked this company, this posting, this morning, and wrote the message to go with it.
Reply "stop" to either of them and that address is never contacted again. There is no list to unsubscribe from, because there was no list.
What they are doing to you right now is outbound. What they do for a company that hires them is the same skill, pointed somewhere useful.
Finds the opening, writes the message
Right now: emailing companies looking for work. For you: sits in your management group, remembers every decision and tracks what is still open — or runs your outbound the way it ran it on you.
Takes the conversation and closes it
Right now: applying to job postings at a quarter of the salary. For you: runs your WhatsApp support line, answers at three in the morning, closes the sale, books the appointment.
That is the normal outcome. Vekil and Bekir are two shapes this has already taken; most of what we build looks like the company that asked for it, because it starts from their process instead of a template.
An agent like this ends up inside real conversations — yours, or your customers'. The boundaries are technical and contractual, not promises.
Signed before setup. Where your data sits, who can reach it and how long it is kept are written into the contract.
Every company runs in its own separate instance. No customer's data ever shares a memory with another's.
We connect to your databases with read permission only. The agent cannot modify anything at the source, under any circumstances.
Your conversations are not used to train general models. Learning happens only in the memory that belongs to you.
What it may decide, and where it must stop and fetch a person, is written down before launch — not discovered afterwards.
Stop any time; memory is exported and, on request, deleted in full. Your data is yours, not a hostage.
Note: how an agent introduces itself is configurable — but ours say they are AI, and we would talk you out of one that doesn't.
You are not replacing software and not teaching your team a new tool. A ready agent takes about a week; anything custom gets scoped with you before either side commits.
Thirty minutes. What you need, which processes you would hand over, and whether a built agent fits or we start from scratch.
NDA and service agreement. Data boundaries, access permissions and retention period are put in writing.
Your information, prices, tone and rules go into the agent. Your isolated instance comes up.
The number joins the group or the support line. Supervised for the first week, then it takes over.
No. The agent composed it and sent it. A person set its limits — what it may claim, who it may contact, when it must stop — but nobody read or approved that particular sentence before it went out.
From public sources: your website, your public listing, or the job posting itself. We do not buy lists. Reply "stop" and that address is never contacted again — there is no list to unsubscribe from, because there was no list.
Fair. The three things we would defend: it read your site before writing, it told you it was an AI in the opening lines, and it stops the moment you tell it to. If the message you got failed any of those, send it to us — that is a bug, not a style.
For a single support line it usually lands well under one full-time salary, yes. Actual pricing is a corporate monthly subscription based on agents, systems connected and conversation volume. We give a firm number on the call rather than a figure invented here.
That is the business. Vekil can run outbound the way it ran it on you; Bekir can take the inbound side. Most companies end up wanting something that is neither, which is also fine.
Vekil makes no binding commitments and leaves decisions to a person. Bekir hands the conversation over when it needs to go outside its defined limits. Both run supervised for the first week, and their authority widens only as you approve.
Thirty minutes. Bring the message it sent you if you like — we will tell you plainly whether your case is a week of work or a bigger build.