1 · HOW DO I CREATE ONE?
It all starts with the job.
Like any other hire.
Nobody hires by starting with the person. First you write down what it has to solve, toward what goal, on which channels and over which systems. Here it works the same.
- A name and a role
- Your team and your customers know who they are talking to.
- A goal
- What it has to achieve, not what tasks it performs.
- Scope and limits
- How far it decides on its own, and what goes to your team.
- A result it answers for
- Someone answers for the result, not just for the task.
They aren't the catalog. They're the six jobs where the difference shows up fastest.
Either way you leave with the same thing: the job written down and half an hour to schedule.
With the job written down, that half hour starts with your employee already running in a test environment.
Moving it to your operation is a matter of connecting your systems.
Half an hour with us and you leave with the job written down — whether or not you end up working with us.
2 · HOW DO I ADD IT TO MY TEAM?
It onboards the way anyone else does
When someone joins, you give them an email, add them to the group, tell them how the team works and show them where the information lives. Then you tell them what they own and how far they decide on their own. With an artificial employee it's the same.
Where you reach it
The channels your team and your customers already use. Nobody has to open a new tool.
- Web chat
- Gmail
- Slack
What you give it access to
The systems where your information already lives. With permissions per action, not per system.
- Tiendanube
- Shopify
- Mercado Libre
- HubSpot
- Google Workspace
Don't see yours? Tell us which one. Some are ready and others come in stages.
See every channel and systemWhat it knows how to do
Concrete skills, not generic capabilities. They get assigned by job: each employee has the ones its work needs, and none beyond that.
- Searches your knowledge base
- It answers from your documents, prices, and policies. If it cannot find something, it says so. It never makes it up.
- Queries your catalog
- It searches by what the customer asks for and shares the item link and photos. If nothing fits, it says so instead of forcing a recommendation.
- Escalates to a person
- It hands the case to your team with the reason and a summary of the conversation. The customer repeats nothing.
- It hands work to another artificial employee
- When part of the job belongs to another role, it delegates and keeps the conversation itself. The customer talks to one.
- Keeps the contact record
- It saves name, email, and phone when they come up. It confirms the data before saving and never overwrites without asking.
- Moves the lead through your funnel
- It changes stage when the conditions you defined are met, and leaves an internal note explaining why.
- Profiles the lead
- It records the answers to the questions that matter to you, across the conversation. No interrogation.
- It follows up on its own
- If the customer goes quiet, it writes again from where they left off. It respects the funnel stage and your business hours.
- It helps you coach yours
- An employee specialized in coaching the others. You explain what you want to change and it proposes the change. Applying it is your call.
And before it talks to a single customer, you test it yourself: real cases, an isolated environment, no risk to your operation.
And it speaks the way your company speaks: the tone, the manner, and the words you use with your customers.
With that it can already work. In two days it's running; in ten, tuned to your operation.
It depends on one thing we don't control: how much time your team can give it.
And when something changes in your operation, it learns it once and applies it every time: across every channel and every conversation.
3 · HOW DOES IT WORK?
An ordinary Thursday.
Nina, support specialist.
What a customer sees is a conversation. On the other side there is an order looked up, a carrier checked, and a delivery rescheduled. Hour by hour.
Change the industry and the names change. The question coming in, the system being touched, the limit and the record stay the same.
At the end of the day what's left isn't a feeling that it went well: it's the record of every conversation, every system lookup and every decision.
See Nina's full job4 · WHAT CONTROL DO I HAVE?
She works on her own.
She doesn't work blind.
Nina stopping at 11:20 was not luck. The margin was written down before her first day.
- Before · what it can do
You define what actions she can perform in each system and how far she decides on her own. What isn't allowed doesn't exist: it isn't an instruction she could read differently one Tuesday.
She starts narrow, with the minimum to do the work, and you widen it whenever you want. She doesn't talk to a customer until you enable her.
- During · where its scope ends
When a case falls outside that, or the customer asks for a person, she hands it to your team with the whole conversation and a summary of what was already discussed. Whoever picks it up has the context without reading it all.
And when she doesn't have the answer, she says so. That isn't advice we give her: it's written into how each of her skills works.
- After · what stays on the record
Every conversation, every system lookup and every decision are recorded and can be reviewed. You can see why she did what she did, not just what she did.
And it is measured by the same yardstick
Nina shows up on the same dashboard as the rest of your team, with the same columns: how much is open, what was answered on time, where things stall.
And when something goes wrong, the record says where: with that you adjust the scope, the limits, or who owns what.
And the data is yours
Your information is used to provide the service, and nothing else. It is not sold or rented. When an external AI provider is involved, it is under contract and limited to running your employees.
All of this was already in the job you wrote.
It's not a bot.
It's someone who owns the job.
| Artificial employee | Bot / automation | Hiring someone | |
|---|---|---|---|
| Scope | Owns a job | Completes a task | Owns a job |
| Memory | Learns and improves | Forgets after execution | Learns and improves |
| Channels | All at once | One | One or two at a time |
| Hours | 24/7, year-round | 24/7 | Business hours |
| Ramp-up | Two days | Hours | Months |
| Scaling | Instant | Instant | New search |
| Control | Permissions, boundaries, audit log | Rigid rules, no judgement | Trust and supervision |
Scope
- Artificial employee
- Owns a job
- Bot / automation
- Completes a task
- Hiring someone
- Owns a job
Memory
- Artificial employee
- Learns and improves
- Bot / automation
- Forgets after execution
- Hiring someone
- Learns and improves
Channels
- Artificial employee
- All at once
- Bot / automation
- One
- Hiring someone
- One or two at a time
Hours
- Artificial employee
- 24/7, year-round
- Bot / automation
- 24/7
- Hiring someone
- Business hours
Ramp-up
- Artificial employee
- Two days
- Bot / automation
- Hours
- Hiring someone
- Months
Scaling
- Artificial employee
- Instant
- Bot / automation
- Instant
- Hiring someone
- New search
Control
- Artificial employee
- Permissions, boundaries, audit log
- Bot / automation
- Rigid rules, no judgement
- Hiring someone
- Trust and supervision
The team you already have stays the same. What changes is what it works on.
Let's start with one job.
The rest can wait.
Tell us which part of your operation is overloaded and we'll tell you whether there's a model for it. If there isn't, we'll tell you that too.