Solution · Custom agents
Agents built for your process.
Kentor understands how your company works, gives the agent the context and the house rules, and builds, tests and maintains the agent in your routine. It does the work in the tools the team already uses and asks first on what matters.
Kentor Bot
Used 2 tools
Draft for Carlos at Mirante
Hi, Carlos! The renewal proposal has been stalled since the 12th. I set aside the payment term you asked for. Shall we close this week?
Send to Carlos now?
OK to send
Message sent · 10:42
Sample agents
Agents that already run at companies like yours.
Each one was born from a real process. Yours starts the same way: from your process.
Sales
Opportunity score per account
Gives each account a score with the rule explained. Delivers the list in order and the why behind each position.
Didn't find yours? It doesn't exist yet. We build it.
The problem
A good agent starts with context.
A generic AI writes well, but doesn't know who your customer is, what your policy is or who needs to approve. Without that, it gets things wrong with confidence, and someone on the team has to redo it.
Doesn't know who the customer is.
The same customer is in the CRM, the order system and email, and nobody put it together.
Kentor Brain brings it all into a single record, and every piece of information says where it came from.
Doesn't know the house rule.
The maximum discount, the brand's tone and the exception only one person knows.
The rules are written down, with an owner and a history, and the agent follows today's version.
Sees more than it should.
An agent with access to everything is a risk.
Permissions apply by team, to people and to agents. The sales agent doesn't read HR's payroll.
That's why we start with the process, not with the agent.
How we work
You tell us what jams. We build, test and maintain.
Nobody has to build a flow, write a prompt or have a technical team. Our team builds the agent together with whoever knows the work.
- 01
We build.
The agent comes out of the mapped process, with the company's context and rules, in the tools the team already uses.
- 02
We test.
Before acting on its own, it runs in shadow mode: the team operates, Kentor watches and adjusts until the agent gets it right your way.
- 03
We maintain.
The process changed, the rule changed or the tool changed, and the agent keeps up. If the company prefers, Kentor operates and evolves the agents day to day.
What's built stays with your company: the mapped process, the knowledge in Brain and the agents' configuration.
Day to day
You ask. It does. You see what was done.
01 · Order
Ask as you would ask someone on the team.
A message or a ready-made skill from the bar, like "Follow up on a stalled lead" or "Day summary by area."
02 · Execution
It works in your tools, and shows every step.
Every tool used becomes a line in the conversation. Before sending, publishing or changing anything important, it asks.
03 · Report
Every action, logged.
Time, tool and who approved. What a rule blocked shows up with the reason.
Do mapa à produção
Do mapa à produção.
O mesmo caminho para todo agente: primeiro entender, depois automatizar.
Kentor OS
Entender onde o agente vale.
O OS entrevista quem faz o trabalho e desenha o processo como ele acontece, com o atalho que ninguém escreveu. A oportunidade vem com a evidência e a conta.
Included
What comes with each agent.
- The map of the process the agent will take over (Kentor OS)
- The company's context and rules, with permissions by team (Kentor Brain)
- The agent built into the tools you already use (Kentor Bot)
- Approval before what matters: you define what it does on its own
- Log of every action, with time, tool and who approved
- Test in shadow mode before release
- Training for whoever will operate, on real work
- Follow-up and adjustments after it goes into production
- Everything that's built stays with your company
Case
Agency routines, built from the team's real process.
Traffic agency · 12 accounts
Power Publi
A manager agent looking after 12 accounts and hours given back to the team.
View caseFrequently asked questions.
Do I need a technical team?
No. Kentor builds, tests and maintains the agent. On your side, whoever knows the work gets involved.
Does the agent act on its own?
Only in what you enable. By default, it asks before sending, publishing or changing anything important, and every action is logged.
What if my process is different from all this?
That's the most common. The sample agents show what's already been done, but yours starts from the map of your process, not from a ready-made template.
Which tools does it work in?
In the ones you already use: messages, email, CRM, spreadsheets and systems, each with the permission you grant.
What about company data?
The agent only sees what its team can see. Permissions apply to people and to agents, and data handling follows LGPD.
How long until the first agent is running?
It depends on the process and the tools. We work in short cycles, each ending with something running, and give you the timeline in the first conversation.
Which routine do you want off your hands?
Tell us the process that stalls. We'll show how it becomes an agent and what it needs to run.