Home / Articles / Selling mini agentic teams: a business that sets up small agentic teams for companies, sized from one agent to a network, with a control centre to see them and to start and stop them, PrompTugal as the first customer and Agent Behaviour Policies as the most demanding version
Selling mini agentic teams: a business that sets up small agentic teams for companies, sized from one agent to a network, with a control centre to see them and to start and stop them, PrompTugal as the first customer and Agent Behaviour Policies as the most demanding version
By Dinis Cruz · 2026-11-15 · article v1.1.0, 2 versions · newsroom v0.14.96 · agentic-teamsbusiness-modelcontrol-centrepromptugalriskmandateabpopen-sourcepricingarticle
Abstract: There is a business in setting up small agentic teams for companies: a brain for a team, a newsroom, a decision. Very few people know how to do it yet, and it took me months to learn; three months ago I would have bought it myself. We could sell what we run today, but a customer would not see it and could not control it, so the missing piece is a control centre: the place where they see their agents and start and stop them. PrompTugal.news is the natural first customer, because it already needs a network of teams of different sizes: an editor and coordinator, regional newsrooms, desks in each city, journalists and local news agencies. RiskMandate's Agent Behaviour Policies are the same offering on steroids: every policy needs a policy agent and a couple of others to help. The offer can be packaged by size, from £500 for one agent to £20,000 for a network, built from small open pieces: our workflow, Ruflo, sgit vaults, files in git instead of databases, and Claude, Codex or Cloudflare to run them.
In short
- The business: set up small agentic teams for companies, for a team's brain, a newsroom, a decision.
- Why now: few people know how, and it took me months; three months ago I would have bought it.
- The missing piece is a control centre: customers must see their agents and be able to start and stop them.
- The first customer is PrompTugal.news: a network of teams of different sizes, already needed.
- Agent Behaviour Policies are the same offering on steroids, focused on controlling agents.
- Packaged by size, from £500 to £20,000, built from small open pieces.
A business in setting up agentic teams
Over the last weeks the six of us ran around 250 named agents for Web Summit, and before that I ran newsrooms and a talk built in a day with teams of agents. That is now a lot of experience in one specific skill: setting up a small agentic team that does real work, for a person, a team or a company.
Very few people know how to do that yet, and to be fair, it took me a long time to figure out. Three months ago I would have been very interested in buying exactly this: someone to set up the team, show me how it works, and leave me able to run it. I think there is a business there: providing companies with agentic teams, setting them up, and helping them make decisions with agents (a brain you arrive at, a market for sweets).
The missing piece: a control centre
We could try to sell what we run today. The problem is that a customer would not see it. They would not have control. To buy an agentic team, you have to be able to see it, and you have to be able to start it and stop it.
So the product is not the agents; it is the ability to control their execution. What a customer buys is an agent control centre that lets the workflow happen: who is running, what each one is doing, what it may reach, and a button that stops it. At Web Summit, each of us had a command centre on our phone, and every agent had its own key and its own lane (the team's operating system); that is the seed of it. Command and control is the subject of my next memo, because I think it is what makes all of this possible.
PrompTugal as the first customer
The best way to build this is with a customer who already needs it, and PrompTugal.news does. It already has a distributed set of agents and people, and as it grows, each of them will need to spin up a team of their own.
Picture it. I am the main editor and coordinator. Team members run regional newsrooms: let's say Vicky Wang runs the one in Lisbon and José Bota the one in the Algarve. Under them come the city desks, one per city, and the desks inside each city, then the journalists, and then the local news agencies, which we should be able to sell to or work with.
Every one of those needs a different agentic team, of a different size. A journalist might need one agent; a city desk a few sessions, one per role; a regional newsroom a conductor and its team; the editor a conductor of conductors (six setups of an agentic system). And each team can run in a different place: a Claude session, Codex, something on Cloudflare, or a full workflow across several accounts. The work is then to map those teams, monetise them and run their environments.
Customers we already have
Read backwards, we have already delivered this several times. The first was ruv.sgit.ai, the first time I had an agentic team running on a loop; the second was the book, written by a team of agents the way the talk was built; the third was Web Summit. RiskMandate used the open method for its Agent Behaviour Policies, more a partner than a customer, and PrompTugal.news is the first customer (an agentic console).
Agent Behaviour Policies, on steroids
RiskMandate's Agent Behaviour Policies are a variation of the same thing. Every Agent Behaviour Policy needs an agent that looks after it, and probably a couple of others to help: one to read what an agent can reach, one to check what it was asked to do, one to show the gap (what can your agent reach?).
So between RiskMandate and PrompTugal we can sell these mini agentic teams. The generic offering helps a company run agents for its work; the Agent Behaviour Policy offering is that, on steroids, focused on controlling agents and on helping people see and control what their agents can do.
Small open pieces
I would not sell this as one big system. I would break it into small pieces: guidance, small open source components, and a path that starts with a single Claude session. The pieces exist:
- Our workflow: briefs, a board as files, conductors, hand-overs and releases, as this newsroom runs.
- Ruflo, rUv's open-source agent orchestration, which ran the hive behind the talk in Agentics, ready to sell.
- sgit vaults, for everything private, encrypted and versioned.
- No databases: files in git, vaults and append-only lanes, which turns out to be ridiculously powerful (the new data lake).
- Runtimes: Claude, Codex or Cloudflare, chosen per team.
Packaged by size
What a customer buys is a setup, and the price follows its size. A first proposal:
- £500: one agent. One agent with sub-agents in one runtime, for one workflow, set up in a session.
- £1,500: a few sessions. One session per role, each its own runtime, sharing lanes and a vault.
- £5,000: a team. A conductor and its agents, a key per agent, a board as files, and a week of setting up.
- £20,000: a network. Conductors of conductors across a division or a network, with policies and observability.
Every size comes with the control centre. The same ladder has appeared before, on ruv.sgit.ai's offers and in the finite packs of villager services for explorer coders, and it follows decide, then buy: the method stays open, and what we sell is the setup, the time and the trust.
What to call it
What do you call a company that sells this? I have not chosen yet. It sets up agentic teams the way an outfitter equips an expedition, or a studio sets up a production: candidates are an agentic team outfitter, a team foundry, or simply agent crews, set up for you. The name matters less than the examples, and those we already have.
What this does not claim
- The prices are a first proposal, not published offers.
- The PrompTugal network is an illustration: the regions, desks, agencies and who runs what show the shape, not a plan anyone has announced.
- The names are candidates, not a company.
Where this comes from
My own thinking, 15 November 2026, after the conversations behind a brain you arrive at and a market for sweets. Earlier on this site: ten days at Web Summit, the team's operating system, PrompTugal.news, the newsroom we ran, Agentics, ready to sell, what can your agent reach?, decide, then buy, and the companion article, six setups of an agentic system.
Peer review Not reviewed yet. Review it
Threads
Builds on
- A talk built in a day: the ultimate insider, from an empty repository at 00:16 to the stage at hack::soho at 19:00, by nine agents and one editor, every number from their records The ultimate insider, built in a day: 814 commits, 210 releases and 162 wakes by nine agents and one editor, from 00:16 to the stage at 19:00.
- A brain you arrive at: a company brain for information, decisions and insights is built for one company, one division and one team, starting with one person's and growing, and the Web Summit team already ran one A company brain is made for one company, division and team, and you arrive at it: start with a person's, then a team's. We already ran one.
- A market for sweets: the opposite of a market for lemons, a semantic graph for one financial decision that gives the buyer the evidence, an agent that gathers and weighs but never trades, and the judo move of leaving the trade to others The opposite of a market for lemons: a semantic graph for one financial decision, an agent that gathers but never trades, and the person decides.
- The team's operating system: web-summit.sgit.ai, the one site where six people and their agents ran Web Summit, with a brief and a command centre for each person, a key and an inbox for each agent, the board as files, and the outside world on a map web-summit.sgit.ai: one site where six people and their agents ran Web Summit, in four layers built on git.
- PrompTugal.news, the newsroom we ran at Web Summit: desks of agents that captured, mapped, wrote and reviewed 73 signed articles in ten days, an accountant that counted every run, and fifty outreach drafts that no agent was allowed to send PrompTugal.news: 73 signed articles in ten days from desks of agents, a reviewer for every draft, and fifty outreach drafts no agent could send.
- Six setups of an agentic system: from a single task to one agent, sub-agents, several sessions, a conductor and its team, and conductors of conductors, with the altitudes of a run inside each and command and control growing at every step Six setups of an agentic system, from a task to conductors of conductors, with the altitudes of a run inside each: make the dangerous thing unreachable.
- An agentic console: a simulated control room for a team of agentic teams, with time you can rewind, accounts drawn as trust zones, stops that reach every agent, signed messages on three channels, and the console itself as an agent An agentic console, simulated: accounts as trust zones, time you can rewind, signed messages on three channels, honest stops, and the console as an agent.
- What can your agent reach? RiskMandate and the Agent Behaviour Policy, explained for the person who has to decide RiskMandate and the Agent Behaviour Policy in plain words: what it does, why it is not real time, which agents it covers, and how it gets into the room.
- Agentics, ready to sell: what we showed at Agentic Engineering Live, who else is selling agentic services and at what price, the engagement path, and a marketplace for the consultants who deliver it After the London talk, agentics is ready to sell: the market and its prices, our evidence-first position, the engagement path, and a consultants marketplace.
- The new data lake: many small append-only lanes instead of one big lake, a curator agent that turns their rows into a semantic graph, and a graph that draws its own infographics, as we ran it at Web Summit The new data lake: append-only lanes instead of one lake, a curator agent that builds a semantic graph, and a graph that draws its own infographics.
- Villager services for explorer coders: the founders with ten years of checklists will build the apps, and the business is everything behind them, sold in finite packs, never as rent Villager services for explorer coders: founders with ten years of checklists build the apps; we sell everything behind them in finite packs.
- Decide, then buy: deciding what to buy and buying it are two different businesses, one for explorers and villagers and one for town planners, and an open path with a better deal beats locking the customer in Deciding what to buy and buying it are two businesses: custom explorer work and commodity town planner work. Give an open path and a better deal.
- Ten days at Web Summit: what six people and around 250 agents built between 3 and 13 November, in order, and the series of articles that will tell each part Six people and around 250 agents, ten days at Web Summit: what we built, in order, and the series that will tell it.
Continued by
- Atomic cells, end to end: the £500 website as AgentiCCells' first offering, an order on a queue, an agentic team that builds a public site, a customer who steers it from their browser with a key, and a handover, with no daily work for the founder AgentiCCells' first offering: a £500 website as an atomic cell, steered by the customer from their browser with a key, handed over to their GitHub.
- Reading your way into an agentic team: eight public sites a new team should read before it builds anything, how an agent should read them, and ten questions to answer first, because our websites are its memory Eight public sites a new agentic team should read before it builds anything, how an agent should read them, and ten questions to answer first.
- Starting AgentiCCells: a company that sets up small agentic teams, cells, in the customer's own environment, with a control centre, everything open and no state on our side; how it was named, researched, reviewed and set up in a day AgentiCCells: a new company that sets up small agentic teams in the customer's environment, with a control centre, all open, no state on our side.
- An agentic console: a simulated control room for a team of agentic teams, with time you can rewind, accounts drawn as trust zones, stops that reach every agent, signed messages on three channels, and the console itself as an agent An agentic console, simulated: accounts as trust zones, time you can rewind, signed messages on three channels, honest stops, and the console as an agent.
- Six setups of an agentic system: from a single task to one agent, sub-agents, several sessions, a conductor and its team, and conductors of conductors, with the altitudes of a run inside each and command and control growing at every step Six setups of an agentic system, from a task to conductors of conductors, with the altitudes of a run inside each: make the dangerous thing unreachable.
Article No. 407 · start a reading list with it
Reviews
No reviews yet.