People First Artificial Intelligence, S.L. · Knowledge-Based Company of the University of Seville
An AI chat answers one question at a time and forgets how you work. An agent system like AI Flock splits the work between specialized agents that keep their instructions, consult your documents, check each other and learn from every correction, with your approval at the critical steps.
How it works insideResearchers from the University of Seville and engineers with years of company projects behind them. When you tell us about a process, you are talking to the people who will then build the agents.
Founder & CEO
PhD in Computer Science from the University of Seville, with a thesis on multi-agent systems, and a professor there for more than 25 years. He researched agents for space missions with NASA Goddard. He created the FAST productivity method, which has trained more than 22,000 professionals in 54 countries.
CTO
Computer engineer with 25 years of experience in software development and technical training. He leads the development of the AI Flock agent engine and the applications that connect to it. He has spent years designing the context language models receive, since before it was called context engineering.
Software engineer
He started out automating business processes with custom development and data analysis. At AI Flock he works on the AI pipelines, from integrating customer data to deploying agents that cooperate with each other.
Software engineer
Software engineer specialised in multi-agent system architectures. At AI Flock she designs and builds automations meant to grow with the customer without having to be rebuilt.
CFO
Economist with more than 20 years in financial management, tax and business administration. He runs AI Flock’s financial strategy: SaaS metrics, model usage costs and public and private funding.
From the ISA research group and the SCORE Unit of Excellence at the University of Seville. Each line shows up in the AI Flock engine.
Autonomic Computing · Communications of the ACM, 2006
Agents that adjust their behaviour when conditions change, without anyone having to reprogram them.
In AI Flock: the agent takes on every correction you make.
PPINOT and RALph · Business Process Management
Languages to describe a business process: its steps, who does each one and how it is measured.
In AI Flock: a planner orders the steps and asks for your approval at the critical ones.
iPricings and CSOP · CAiSE 2025
Models to lower the cost of running a system while keeping its operating guarantees.
In AI Flock: each step loads only the context it needs, so the cost per query is predictable.
Process mining and conformance checking
Techniques to check, from the execution log, whether a process was carried out as planned.
In AI Flock: there is a record of what each agent did and in what order.
University of Seville research group working on BPM, software engineering and multi-agent systems.
Visit ISA GroupSmart Computer Systems Research and Engineering, a University of Seville unit of excellence that AI Flock collaborates with.
Visit SCOREJoaquín Peña researches multi-agent systems at the University of Seville and spends a research stay at NASA Goddard. The work leads to papers in Communications of the ACM.
The first company to come out of the university: four software products and a team of 22 people.
Productivity and digital transformation consultancy. We trained teams at Santander, Iberdrola, Thales, General Dynamics and Leroy Merlin.
A personal productivity method that has trained more than 22,000 professionals in 54 countries. It taught us how a team adopts a new tool.
Outreach on AI and work, guided by one principle: technology should make people more capable.
dighumi.orgTechnology transfer projects with companies kept surfacing the same need: agents that follow their processes and consult their documents. AI Flock comes from there.
People First Artificial Intelligence is set up as a Knowledge-Based Company of the University of Seville and collaborates with the SCORE Unit of Excellence.
Training clients during the Productivity Times years (2012-2024)
The company is called People First Artificial Intelligence for a reason. We have spent twenty years researching how to coordinate people and software, and every project leads to the same conclusion: technology pays off when people understand what the system does and keep the decision.
We call that technology that frees you up. Agents take on the searching, the cross-checking of documents and the first draft. People get back time and focus for the work that needs judgement: talking to the customer, reviewing the result and deciding.
Mission
Let any company work with AI agents without coding, without losing control of its data and within European regulation.
Vision
A humanist AI that serves people and companies: teams of agents that give people back hours to spend on what adds value.
Agents search, prepare and draft. People review and approve at the steps they choose themselves, and with every correction the agent does better next time.
Agents can show the source they consulted. If the agent cannot find the source, it says so. An incomplete answer can be completed; an invented one does damage.
FAST taught us that a tool gets adopted when it is used on everyday work. That is why agents are built with you, on your own documents and real processes.
What the agents learn belongs to your company and stays there even if the person who trained them changes roles. It is the part of the craft that gets lost with every handover today.
GDPR and the AI Act are part of the architecture from day one: there is a record of what each agent did, with which data and what the person approved, and data is stored on European servers.
The AI model can be swapped and what the agents store can be exported. If a better model appears tomorrow, or an open one running on your own machine, your work is still yours.
We develop these ideas in the open at the Digital Humanism Institute. Visit dighumi.org
How the platform handles itChoose how to start.
You create your account and start with trial credits to set up your agents with your own documents; we leave the first one running.
Request accessWe look at your most repetitive process, show you which agents we would set up and send you a summary the next day.
