Chief of Staff
Hires the other agents by conversation, with manager approval; sends the daily brief; watches costs and licences.
AI coworkers for SMBs on Microsoft 365
We deploy named AI coworkers into your Microsoft Teams — on your own servers, approved by your managers, for a fixed monthly cost. Capacity for teams that can't hire. The first one in four weeks.
No new tool · Nothing leaves your infrastructure · No surprise token bills
Built for our own company before it was sold to anyone.
0 cloud credentials in the agents' hands.
Every consequential action approved by a person.
Audit log of every command, approval and cost.
Meet the team
Six of the eight launch roles. Each company picks the naming theme — people, sea animals, mythology, or simply the job title.
Hires the other agents by conversation, with manager approval; sends the daily brief; watches costs and licences.
Tracks tasks in Jira, Planner, Asana or monday.com, chases owners on stalled items and closes the loop with whoever decides.
Prepares proposals and quotes from your templates and price lists; a salesperson reviews and sends.
First-line internal support in Teams: answers what it knows, opens the ticket and escalates to a person when it doesn't.
Operates your AT-certified invoicing software through its API, prepares reconciliations and payment reminders. It never issues or alters fiscal records; the certified accountant keeps responsibility.
Onboarding, leave and internal policies — prepares, reminds and answers; the decision always stays with the responsible person.
Three guarantees
No new tool. Nothing leaves your infrastructure. No surprise token bills.
Agents are coworkers in Teams: they answer DMs, take part in channels, get @mentioned.
They run on your servers, one container per agent. Secrets in your company's 1Password, injected into the containers by a service account the agents never see.
Seat-based subscription. Hiring, configuring and retiring agents happens in chat — never in a terminal.
How it works
A fixed-price pilot, an agreed KPI, and governance handed over at the end.
We map processes with your managers, pick the first role and prepare the VM on your infrastructure.
Who: you: a VM and one M365 consent · us: the rest
The first coworker joins: manages access, licences and announcements, and sends the daily brief.
Who: you: 1 h of interviews · us: install and configure
Hired by conversation, with the approval card in the manager's hands. It starts doing real work.
Who: you: approve with a tap · us: integrations and tuning
We measure against the baseline, train your managers (2 h) and hand over the adopted governance.
Who: you: review · us: training, policy, hand-over
The rest — install, integrations, training, governance, operations — is on us.
See the pilot in detail →Why now
The gap isn't technology: it's deployment and governance. That is what we deliver, done for you, at a cost that fits an SMB.
70%
of Portuguese adults already use generative AI
Deloitte, 2025
9.4%
of small Portuguese companies have it at work
INE, November 2025 (49.1% of large firms)
> 40%
of agentic projects will be cancelled by 2027 over cost, unclear value or weak risk controls
Gartner, 2025
Why not…
Each is good at what it was built for. None puts named coworkers inside your Teams, on your servers, with human approval and a fixed cost.
| Enlogical | Copilot Studio | SaaS “AI employee” platforms | Block Buzz | |
|---|---|---|---|---|
| Where it lives | In your Teams, as named coworkers | In M365, as individual bots | In a US SaaS app | In a new chat app (pre-1.0) |
| Where data lives | On your servers | Microsoft cloud | Vendor cloud | Self-hosted |
| Model | Any (Claude, OpenAI, on-prem) | Microsoft only | The vendor's | Several |
| Cost | Fixed per seat | Credits ($200 / 25k) | Per token or credit | Free + infra |
| Human approval | On every consequential action | Partial | Varies | Unfinished |
| Cross-tool work | Yes — agents build the integrations | Limited | Large catalogue | In development |
| Who owns it | You: code, agents, data | Microsoft | You rent | You |
Only Teams messages cross a zero-trust tunnel. Nothing is used to train models.
Lawful basis per function, DPIA support, erasure that reaches agent memory.
Every consequential action goes through an Approve / Reject card in a manager's hands.
Commands, approvals, costs and the model actually used. Retention ≥ 6 months.
FAQ
No. These are custom agents inside Teams, running any model, on your servers. They coexist with Copilot and can be registered in Agent 365 / Entra Agent ID so IT keeps one control plane.
It stays on your VM or private cloud. Only Teams messages cross a zero-trust tunnel. Model calls run under enterprise DPAs and are never used for training.
Agents propose; a manager approves with a tap. They hold no cloud credentials, have per-agent permissions and everything is in the audit log. It is the platform's rule, not a promise.
Your managers, by chat — hire, retire, grant access, change schedules. We run operations: updates, backups, watchdogs, monthly review.
A fixed monthly cost: managed operations plus seats (model subscriptions). No per-message bills, no token surprises. Figures come in a proposal after a 20-minute call. See how to deploy.
Chief of Staff working in week 2; first role agent in week 3; KPI measured and governance handed over in week 4.
You keep the code, the agents, the data and the governance policy. No lock-in: the hand-over is documented.
Yes. They build and document their own integrations, reviewed by us — including Primavera, PHC, Moloni and InvoiceXpress through their APIs.
No. They give capacity to teams that can't hire and take repetitive work off people. Decisions stay with people — it is the platform's rule. We deliver the worker information required by Portuguese law (Lei 13/2023) before go-live.
Agents operate your AT-certified invoicing software through its API; they never issue or alter fiscal records. The certified accountant keeps the responsibility they always had.
Pilot in 4 weeks
A 20-minute call, a real agent at work, no slides.