Hire AI employees for your sales team
Scoped into a role like a colleague, supervised like one, and working inside the workspace your team already uses.
AI employees never read your inbox or a prospect's replies.
6 AI employees at work. Select one to follow it.
One memory of every contact.
Every lead, company and conversation your team captured sits at the center. An AI employee only ever works from what your team recorded. It never reads your inbox or a prospect's replies.
Hired into a role, like a colleague.
Ilse is a Research employee. She reads a record, looks up the company, fills only the empty fields and writes the result back into the memory, where a star lights for good. The role fixes which tools she may ever hold.
Nothing leaves without a person.
June drafts the next touch for a quiet lead, then stops. The email waits on Sara, your admin, who can edit it before she releases it. Her edit is what teaches June. Customer email waits for a person until an admin decides otherwise.
It brings you the ones that need you.
Nova finds a deal above the threshold you set and hands it to Mateo with a written brief: what it found, what it did, what it recommends. Open handoffs land on My Tasks.
Every new hire starts in shadow.
Tarek drafts a welcome note and would send it, but nothing leaves. Shadow drafts pile up on his timeline so you can judge his work before you release him. Every action is logged with the rule that allowed it, and can be undone.
Six seats. Yours to fill.
Hire into any of the six roles on a Business or Enterprise workspace. Each AI employee is $99 a month, including 1,000 credits, and the roster shows this same map of who is doing what.
A demonstration. The names, leads and workflows are examples; the rules under the map are the real guardrails.
What is a Lynqu AI employee?
A Lynqu AI employee is an add-on seat on a Business or Enterprise workspace that works your pipeline continuously, with the same tools and permissions a colleague would have. You hire it into one of six roles (Pipeline Ops, Research, Onboarding, Outbound follow-up, Scheduler or Campaign follow-through) and the role fixes which tools it may ever hold. Every new AI employee starts in shadow mode, writing nothing, so you can read what it would have done before you release it. Each action is scored into one of four risk tiers and routed the way you set it: run on its own, wait for a person, hand over, or refuse. Customer-facing email waits for a person until an admin lets that AI employee send unattended, and no AI employee ever reads your inbox or a prospect's replies. Everything it does is recorded with the rule that allowed it, and can be undone. It costs $99 per AI employee per month, including 1,000 credits.
You hire an AI employee into a role, not a task. The role fixes which tools it may ever hold and cannot be changed afterwards. Everything else you can tune at any time.
Pipeline Ops
Keeps the board moving: flags stale deals, fixes bad data and queues the next touch.
Never
Contacts a customer. The role holds no sending tool at all, so nothing within its reach reaches a prospect.
Nova on the map: Stale deal
Pipeline Ops watches for stale deals and missing next steps.
Research
Fills in who a lead is. Company, role and the duplicates hiding behind them.
Ilse on the map: Enrich new lead
Never
Overwrites what a person already entered. It fills empty fields only, and a duplicate merge waits for someone to approve the plan.
Onboarding
Watches how your workspace is set up and reports what is holding the team back.
Tarek on the map: Welcome note
Never
Writes to your CRM. This role is read-only: its findings land on its own results tab and nowhere else.
Outbound follow-up
Writes the next touch for leads that have gone quiet. A person releases every email until you say otherwise.
Never
Sends on its own until you allow it. Sending is off for every new hire, and sending unattended has to be earned against your own correction rate.
June on the map: Next touch
It never reads your inbox or a prospect's replies. Only what your team recorded.
Scheduler
Chases no-shows and offers new times, so a missed meeting doesn't become a lost one.
Bram on the map: Booking request
Never
Decides what a deal is worth. It proposes, reschedules and nudges; moving a stage or touching a number is not in its tool list.
Campaign follow-through
Picks up the people a campaign or event left behind, once it has closed.
Sol on the map: Expo follow-through
Never
Emails in the middle of your event. Follow-through waits for the campaign to close unless an admin switches that campaign to live.
How it works
Four steps, and you set the pace of every one of them.
Hire it into a role
Pick the role and the work it covers: a pipeline, a campaign, a set of leads. An admin hires; everyone on the team can see the roster and what it has done.
Read what it would have done
Every new AI employee starts in shadow mode and writes nothing. Its timeline shows the action it wanted to take and the reasoning behind it, before anything is written.
Set the guardrails
Decide what each risk tier does, which tools it may use, how much it may do in a day and when it has to be quiet. Anything you set to wait lands in an approvals queue you can edit before releasing.
Let it earn more
Sending starts off. An AI employee earns the right to send without waiting once enough of its messages have been approved and its work is being corrected rarely enough, and you can take that back in one click.
Bounded autonomy, not a black box. Everything below is set per AI employee, apart from the two floors nobody can configure away.
Four risk tiers, four routes. You set the route per tier, for each AI employee.
- Runs on its ownLow
- Waits for a personMedium
- Hands overHigh
- RefusesCritical
Two floors you cannot configure away
- External sends always need approval, whatever the tier says.
- Critical actions can never run unattended.
Outbound mail from an AI employee carries a machine-readable AI-content marking, and says so in the message itself whenever the AI employee is sending without a person releasing it. EU AI Act Article 50.
Four risk tiers, routed your way
Every action is scored (low, medium, high or critical) and you decide what each tier does: run on its own, wait for a person, hand over, or refuse outright.
It starts in shadow mode
A new AI employee watches and writes nothing. You release it when you have read enough of what it would have done.
A person releases every email
Customer-facing messages wait in an approvals queue, where you can edit the text before it goes or reject it and say why. The strongest correction signal it gets.
Every action is on the record
The timeline records what it did, which rule allowed it and why it thought it was right. Blocked and expired attempts are recorded too, so you can see what it wanted to do and did not.
Anything it changed can be undone
Undo one row, a selection, or a whole run. Anything that cannot be reversed is listed afterwards with the reason rather than failing quietly.
Ceilings, not targets
Actions per day, emails per lead per week, quiet hours, how much of the credit pool it may spend. An AI employee that hits one stops and waits.
Handoffs
A handoff is the headline, not the apology. When the work stops being the AI employee's to do, it stops and writes a brief for whoever picks it up: the subject, what it found and what it was about to do. Open handoffs land on My Tasks. You take it, resolve it, or hand it back with an instruction, and that instruction is what it works from next.
Six reasons it hands over
- Customer asked for a person
- Needs a human decision
- Out of scope
- Negative signal
- Budget exhausted
- Approval timed out
Handoff brief
Lands on My Tasks- Subject
- What it found
- What it did
- What it recommends
These are properties of how it is built, not switches somebody could flip.
It never reads your inbox
AI employees never read your inbox or a prospect's replies. They work from what the platform itself records: email opens, link clicks, document views, booking responses and capture-form submissions. A written reply lands in your mailbox, and the conversation is yours from there.
It cannot reach past its role
A role's tool list is a ceiling, not a default. A tool the role does not hold can never be granted, and anything you leave unticked is refused, so a misconfiguration and a prompt injection fail the same way.
It does not run sequences
No cadence, no enrolment, no day-three email. It looks at everything known about one subject, picks the single best next action, and schedules another look when the answer is “not yet”.
An add-on seat on a Business or Enterprise workspace. Nothing else about your plan changes.
Per AI employee
$99/month
One seat, one AI employee, and 1,000 credits a month to work with.
See all plansIf it needs more
Packs of 250, 1,000 or 5,000
Packs and the monthly allowance are one balance, so nobody has to track which credit came from where.
Business and Enterprise
Add-on seat
Admins add seats from Billing → Add-ons. Everyone on the team can see the roster and what the AI employees have done; hiring and guardrails are for admins.
A credit is one action: reading costs 1, an update inside your workspace 2, a customer-facing email 5.
See it run on your own pipeline
We will walk through the roles, the guardrails and what an AI employee would find in your workspace this week.