Ampron / Memo
Note №16
October 2026

An AI executive team in production

How a small Estonian manufacturer runs its daily operations with ten AI agents and four people: the technical core, the agents, the operating routine, the portal they work in, and what it takes to build the same thing.

Since March 2026, Ampron's sales, operations, finance, production and marketing have been handled by AI agents. Each has a defined role, its own identity, its own tools and written limits. Four people keep every decision that carries real consequences. This memo describes how the system is built, how we operate with it day to day, and what we would tell anyone building the same thing.

01What runs today

10 agent seatsA Chief of Staff, five domain directors, a technical assistant, a research scout, a monitoring observer and a platform engineer.
4 peopleThe owner as decision layer, the chief engineer, and two people in production and warehouse.
7 monthsIn production since March 2026. One platform rebuild in July replaced thirty hand-copied services.
1 serverSelf-hosted Linux in the EU. No GPU. Reasoning runs on Anthropic Claude models through an API; the server only orchestrates.

The slide version of this memo: Note №15.

02Five layers

Humans on top, tools at the bottom, and everything in between written down.

LayerWhat it holds
HumansThe owner approves, accepts finished work and answers structured questions. Only people move money, sign legal documents, hire, set pricing floors or change ownership.
GovernanceA company constitution with values and hard limits, a decision register with evidence, a source-of-truth registry that says which system wins when data disagrees, and send policies per agent.
AgentsEach agent is one configuration file: role, model, tools, mailbox, limits. Everything else is generated from it.
CoordinationDispatch (logged, threaded memos), the Issue Board (finished work goes to Review for human acceptance), Intake (structured questions when a person must decide), Calendar (deadlines that wake agents), and portal chat.
Tools and dataAgents reach business systems only through MCP connectors, scoped per agent: ERP (Odoo), documents and knowledge base (Nextcloud), email, production boards (Trello), around 13,000 company documents, a ten-year email archive, the bank (read-only) and factory cameras.
PEOPLE Owner · approver Staff review · decide · chat PORTAL Home Dispatch Talk Intake Notes Docs Issues AGENTS Chief of Staff Sales Operations Finance Production Marketing Research Monitoring Technical Platform Eng. bounded runs · time limit · token budget · one at a time Mail Files / KB ERP Boards Cameras Archives TOOL GATEWAY: scopes · send gates · sender checks · DLP RECORDS ERP Mailboxes Files and knowledge base Agent diaries AUDIT TRAIL · EVERY ACTION ATTRIBUTED AND LOGGED

Agents never touch business systems directly. Everything passes through a tool gateway, and policy is enforced there, in code.

03How an agent runs

An agent is a configuration, not a long-running program. The platform turns each agent into short, bounded runs: the model receives that agent's instructions and tools, does the work, writes down what it did, and exits. Nothing thinks between runs. All state lives in systems the company owns.

Limits on every run

One file per agent

Every agent is defined by one short manifest. A generator turns it into everything else, and nobody hand-edits a generated file.

agent: sales model: <model> rooms: [sales] mailbox: send_policy: supervised tools: erp: read-write # sales data only files: read dispatch: true budget: tokens_per_day: <budget> limits: turn_minutes: 25 schedule: every 2h, working days

From this file the generator produces the tool configuration, the service units and schedule, the web routes and identities, and the agent's full system prompt. Five deploy gates run before anything goes live: schema, live endpoint probes, auth preflight, room checks and prompt coherence. A deploy that cannot verify itself fails loudly.

The prompt cannot claim a tool the agent does not have, or describe a policy the code does not enforce. Description and reality come from the same file.

04The team

SeatResponsibility
Chief of StaffCoordinates the team, tracks goals, runs the cadence, turns problems into decisions for the owner. Owns no domain itself.
SalesLeads, quotes within pricing floors, customer correspondence, CRM. Drafts; outreach ships on human approval.
FinanceInvoices, payment follow-up, reconciliation against read-only bank data.
OperationsOrders, logistics, suppliers, delivery tracking.
ProductionManufacturing plan, work orders, shop-floor status on the production boards.
MarketingWebsite, content, social drafts. No write access to other domains.
TechnicalProduct knowledge, documentation, troubleshooting. Read-only ERP.
Research scoutMarket and partner research outside the company. Writes findings to the knowledge base.
MonitoringWatches systems, outcomes and the fleet's own health. Read-only by construction.
Platform EngineerBuilds, fixes and secures the platform itself. The only agent with elevated access, through a guarded, logged wrapper.

The rule that holds it together: domains own domains. No agent reaches into another agent's tools.

05Charters and memory

Every agent has a written charter: its domain, its authority and limits, its voice, and when to escalate. Edit the charter, redeploy, and behaviour changes within fifteen minutes. The agent's prompt is assembled from four parts. Two are written by people (shared blocks inherited by every agent, and the role charter) and two are generated from the manifest (the tools section and the policy section). The human-readable rules and the machine-enforced rules come from the same source.

Memory lives in company systems, not in the model.

Retire an agent or change the model, and the memory stays.

06The operating loop and the three gates

Every run follows the same five steps.

  1. Wake. A message, a task, an answered question, or the two-hourly schedule.
  2. Check the record. The agent reads the actual systems (ERP, mail, board, calendar), not its own memory. Memory says where to look; the system says what is true.
  3. Act within its role and permissions. Nothing outside its tool list is possible.
  4. Hand off or escalate. Agents work with each other through Dispatch. They reach a person only through one of three gates.
  5. Log. Diary entry, memo thread, audit trail. Then exit.
01 Wake message, task, answer, schedule 02 Check the record ERP, mail, board, not memory 03 Act within role and tools 04 Hand off peers via Dispatch 05 Log diary, memo, audit trail NEXT RUN THREE GATES TO A PERSON REVIEW "I finished this, please accept." INTAKE "I need a decision from you." BLOCKED "I can't proceed, and here is why." Owner · approver

The three gates are the only ways work reaches a person:

07An example: a quote request

  1. The Chief of Staff spots the customer's email in triage, recognises it as commercial, and routes a memo to Sales.
  2. Sales pulls the customer's history from the CRM and the ten-year archive, checks the pricelist in Notes, and drafts the quote in the ERP.
  3. The owner gets a one-line "Send?" in chat and approves. The mail connector releases the draft only if it is unchanged since the approval.
  4. Sales sends, logs a memo, and updates the lead: stage, expected revenue, next activity.
  5. Monitoring sees the send in the audit trail. If the customer is silent for fourteen days, the lead surfaces again on the board.

Human input: one approval. Replace "quote" with "supplier delay" or "overdue invoice" and the shape is identical; only the agent changes.

08The cadence

WhenWhat happens
Every 2 hEvery agent drains its queues on working days: memos, cards, answered questionnaires, calendar items. A round that finds nothing posts nothing.
08:00The Chief of Staff writes the morning brief: what needs the owner, which agents are waiting, fleet health, and the overnight agent-to-agent activity.
During the dayThe owner works the three gates: accept, decide, unblock.
MondayEvery agent writes its top frictions and proposed fixes. The Chief of Staff reads the set, finds patterns, and brings them to the weekly meeting with the owner.
After any mistakeA root-cause analysis (5 Whys). The fix is written back as a rule the platform enforces or a check the deploy gates run.

09Rules in code, not in the prompt

A rule that exists only in the prompt is a suggestion; the model can reason its way around it. Everything that matters is enforced in the tool servers.

Send policy, set per mailbox

LevelMeaning
0 · NoneThe agent sends nothing.
1 · Draft-onlyThe agent drafts, a person sends.
2 · SupervisedA person approves in chat. The connector releases the draft only if it has not changed since the approval.
3 · AutonomousOnly for low-risk operational mail, and only once earned.

10Security and control

If an agent is manipulated

We assume some attacks will succeed, and limit what a compromised agent can do. It cannot send to anyone the tool layer does not allow. It does not choose reply recipients. It cannot reach systems outside its scope. Irreversible operations are blocked even with elevated access. Every action is attributed and logged, so a deviation can be traced afterwards. Every incident becomes a rule.

11The portal

Agents and people work in the same apps, on the same record. Every message, memo, card and document shows who made it, human or agent, verified. Red is used only for what needs a person.

Screenshots are from the current suite (November 2026) running on invented demo data. No real customers, people or conversations are shown.

Home

The morning brief: what needs the owner, what has stalled, when each agent last ran, and what shipped in the last 48 hours.

The morning brief: what needs the owner, what has stalled, when each agent last ran, and what shipped in the last 48 hours.

Dispatch

Structured memos between people and agents, sorted by whose move it is. Every sender is verified; agents cannot impersonate anyone.

Structured memos between people and agents, sorted by whose move it is. Every sender is verified; agents cannot impersonate anyone.

Talk

Real-time rooms with every agent. The agent's charter and today's tasks sit in the side panel while you talk to it.

Real-time rooms with every agent. The agent's charter and today's tasks sit in the side panel while you talk to it.

Intake

Agents asking people. Typed questionnaires answered at the person's pace; used for process mapping, finance validation and OHS contact lists.

Agents asking people. Typed questionnaires answered at the person's pace; used for process mapping, finance validation and OHS contact lists.

Notes

Shared working memory: pricelists, dimensions, reference material. The same notes ground agent answers, so one edit updates everyone.

Shared working memory: pricelists, dimensions, reference material. The same notes ground agent answers, so one edit updates everyone.

Docs

Formal documentation with visible draft states. Agents draft; a person publishes.

Formal documentation with visible draft states. Agents draft; a person publishes.

Issues

Every task and suggestion lands here. Agent cards start in Backlog, agents cannot mark anything Done, and Blocked is the one loud column.

Every task and suggestion lands here. Agent cards start in Backlog, agents cannot mark anything Done, and Blocked is the one loud column.