AN ONGOING STUDY OF APPLIED INTELLIGENCE

The hard part is not getting AI to answer. It is getting it to understand the work.

FREIRE // OS is an applied artificial intelligence ecosystem where specialized agents, knowledge and systems are integrated to tackle real problems. The goal is not to impress with answers. It is to understand the work, its exceptions and the limits of automation.

UNDERSTAND THE SYSTEM EXPLORE GUIDES

REAL EXPERIENCE · METHOD · CLEAR LIMITS

freire-os — agent-runtime SIMULATED

~/freire-os / sessions / 001UTF-8 · zsh

ROUTING + CONTEXT PROCESSING

Example of how it works. It does not show the VM in real time.

FIG 01 — LAYERED WORKFLOWCONCEPTUAL / DEMONSTRATION

01 / CONTEXT

02 / EXECUTION

03 / EVIDENCE

04 / GOVERNANCE

01 — THE SYSTEM

An answer is easy.
An operation needs structure.

Agents do not work in a vacuum. They depend on the right context, the right tools and criteria to recognize when they should not act.

FREIRE // OS / REASONING FABRIC

ARCHITECTURE SIMULATION // 01

FROM PROMPTS TO OPERATIONAL SYSTEMS_

Intelligence in motion.

A task moves through context, specialties, tools and validation. The route changes when the situation calls for a different decision.

001SIMULATION / TRACE

Conceptual simulation: routes, events and pulses illustrate functional relationships, not real telemetry. MOVE / CLICK NODES / SWITCH SCENARIOS

01 / CONTEXT

Understand before acting.

Know the vocabulary, the criteria and the exceptions of the environment. Not just receive a longer prompt.

context.load(knowledge, rules)

02 / TOOLS

Execute within limits.

Interact with real systems, with least-privilege access, traceability and reversible actions whenever possible.

permissions.check(action)

03 / GOVERNANCE

Know when to stop.

Recognize uncertainty, record evidence and hand sensitive decisions back to the people accountable for them.

review.request(human)

// DESIGN PRINCIPLE

Autonomy is not the absence of control. It is trust earned through evidence.

02 — MY TEAM / DIGITAL SPECIALISTS

This is my team.
Specialists who work together.

I built FREIRE // OS by bringing together digital agents with different responsibilities. Each one contributes capabilities, context and tools — and knows when a decision needs to come back to me.

06DIGITAL AGENTS
WITH DEFINED ROLES

IN FOCUS / 01 OF 06Joe NakamuraCOORDINATION

Turns requests into coordinated work. Identifies the right specialist, tracks dependencies and gathers what needs a human decision.

SKILLS (CAPABILITIES AND ABILITIES)

classify_requestroute_to_agenttrack_execution

ILLUSTRATIVE FLOW

> request.classify()
> agents.route()
> evidence.review()

A DIGITAL TEAM, NOT A CATALOG OF BOTS

The skills shown are technical examples of capabilities, not confirmation that each integration is already installed or approved. Sensitive actions require permissions, checks and approval.

SEE THE METHOD ↗

03 — THE METHOD

Less spectacle.
More work understood.

Consulting and hands-on building start with the same question: which problem is worth solving, and what needs to be true for the solution to work?

Before automating, understand who does it, why, what changes and where exceptions show up. Interviews, real examples and documentation are part of the work.

INPUT → people + processes + friction

Organize reliable sources, decision rules, domain knowledge, versions and criteria for asking for clarification when information is missing.

INPUT → rules + sources + boundaries

Use the right tools, limit permissions, log actions and set up paths for recovery and review.

INPUT → tools + least privilege

Test with hard cases, record decisions, compare behaviors and learn from mistakes without inventing results.

INPUT → cases + criteria + evidence

Expand responsibility once the operation shows maturity. Some decisions remain human — by design, not by failure.

INPUT → trust + review + limits

// AUTOMATION WITHOUT CONTEXT CAN ALSO SPEED UP MISTAKES.

04 — PUBLICATIONS

AI that fits real life.

Free guides, written for people — not for algorithms. Learn to use artificial intelligence at work and in everyday life, with examples, limits and plain language. The guides are currently available in Portuguese only.

ALL GUIDES (PORTUGUESE) ↗

{ }GUIA / 01

START HERE · PT

5 real tasks to start using AI

Examples for cutting repetitive work without outsourcing your decisions.

BEGINNER · 6 MIN↗

>_GUIA / 02

MORE USEFUL PROMPTS · PT

How to ask AI better, without jargon

A simple method to get better answers and know what needs checking.

PRACTICAL · 5 MIN↗

//GUIA / 03

UNDERSTAND AGENTS · PT

An AI agent is not just a chatbot with a name

What changes when AI has to query systems, take action and follow rules.

CONCEPTS · 7 MIN↗

FREE CONTENT · NO SIGN-UP REQUIRED TO READ

06 — ABOUT / CONSULTING

Practice before promise.

FREIRE // OS is an independent project of applied experimentation, agent engineering and consulting. It was born from the need to make artificial intelligence useful in real routines, not just impressive in a demo.

I am interested in the part that does not fit in a thirty-second video: imperfect information, processes with exceptions, conflicting rules, integrations, auditing and decisions that should not be automated.

I do not believe in agents that magically “learn your business” after filling in a form. I believe in discovery, context, testing, feedback and continuous improvement. That is the work where I focus my experience and practice.

AN EVOLVING SYSTEM. A PERSPECTIVE UNDER CONSTRUCTION.

07 — CONVERSATIONS ON APPLIED AI

OPEN FOR DISCUSSION_

Facing a real challenge involving AI?

Not every process needs an agent. And not every agent is ready to understand an operation.

If you are exploring possibilities, running into limitations or trying to evolve a solution, share the context. A good conversation starts with the problem, not with a ready-made package.

An initial discovery conversation, with no automatic commercial commitment. Do not send passwords, customer data or confidential information.