The market: who has the problem, and why now
This page is an analysis, not a forecast. It says who has the problem and what they do about it today, with sources. It gives no market size and no growth rate: the first six months to a year are for finding out, by selling (the investment).
Four groups with the problem
1. Experts whose time is the product. A teacher, a consultant, an advisor sells hours. Around each hour sits work that is not the expertise: the website, the bookings, the preparation, the follow-up. Dinis's argument is that agents now make that execution cheap enough for the expert to have a team of their own (The expert founders, 15 October 2026). Today they do it themselves, hire an assistant, or leave it undone.
2. Small teams and startups that want to work with agents. Anyone can open a chat with an AI. Few have a team of agents that runs on a schedule, with keys, budgets, a record of what each did, and a person who decides. Our own setup is the example: the team, the agents, the board.
3. Organisations and ecosystems that need to be seen. Associations, regions, delegations and events hold a lot of public information that nobody has put together. In days, PrompTugal.news made a sourced portrait of Algarve Evolution, with its programmes, partners, numbers and events, and invited the organisation to correct it. Regional bodies are already paying attention to AI for small companies: Algarve Evolution runs a pilot of free AI and robotics consultancy for tourism SMEs (INOVA ALGARVE 3.0, on the same page).
4. Companies that run AI agents and must answer for them. Once an agent can act (send, buy, change, publish), someone has to say what it may do and check the gap between that and what it can do. That is RiskMandate's ground (Agent Behaviour Policy). Dinis has written that analysts expect a large share of agentic projects to be cancelled (Agentics ready to sell, 16 October 2026); projects without a mandate are the ones at risk.
What they do today instead
| Alternative | What it does well | Where it falls short for this buyer |
|---|---|---|
| Do it themselves, with a chat AI | cheap, immediate | no schedule, no memory, no rules, no record; it stops when they stop |
| A consultancy or an agency | experienced people | priced by the day; slow to start; the buyer rarely keeps the method |
| An automation or agent platform | ready-made pieces | the buyer moves into someone's platform; mandates and review are the buyer's problem |
| Hire | a person who learns the business | the cost and the time to hire; the work still does not scale |
Our answer sits between the consultancy and the platform: a team that does the work, in the open, which the buyer can see before paying and keep afterwards (how it works).
Why now
- The cost of execution has collapsed. That is Dinis's central claim, and this week is the test of it: a working operation for a team of people, in days (the story).
- Trust is the bottleneck. The question has moved from "can an agent do it?" to "may it, and who checks?". Open code and written mandates answer it in a way a closed product cannot (open source and the moat).
- European buyers want to see it working first. Dinis's reading of Europe's investors and buyers is that they "want to see it working, value lower, and ask for more revenue" (The expert founders): a team that shows the work first fits that.
What we do not know yet
- Which of the four groups buys first, and at what price. The newsroom's customers' agent has proposals, all untested but one (how we make money).
- How much of a customised team a buyer can run without us, and therefore how much is service and how much is product.
- How the competition responds. Anyone can fork the code; the bet is that they cannot fork the pace (open source).
We will learn these by selling, and these pages will say what we learned.