---
okf_version: "0.1"
---

# Guides — Rota Nacional

* [How to compare models without exposing real documents](https://ai.exponencialadm.net/okf/en/guias/como-comparar-modelos-sem-expor-documentos-5e896a.md) - Build a small task set, measure quality and cost, and evaluate the effect of your privacy policy.
* [How to check transport and credentials before integrating AI](https://ai.exponencialadm.net/okf/en/guias/como-conferir-transporte-e-credenciais-antes-4f6f00.md) - Verify the destination, keep TLS checks enabled and separate keys by application and environment.
* [How to decide what an agent may remember](https://ai.exponencialadm.net/okf/en/guias/como-definir-o-que-um-agente-03545a.md) - Build useful memory with allowed fields, retention limits and a test that detects accumulated personal data.
* [How to send only the necessary passage in a RAG workflow](https://ai.exponencialadm.net/okf/en/guias/como-enviar-so-o-trecho-necessario-2b5ec4.md) - Enforce access before retrieval and process retrieved passages before adding them to prompts.
* [How to constrain an AI agent's tools](https://ai.exponencialadm.net/okf/en/guias/como-limitar-as-ferramentas-de-um-0ba347.md) - Separate lookup, proposal and execution. Give each action limits that prompts cannot expand.
* [How to prepare transcripts for AI analysis](https://ai.exponencialadm.net/okf/en/guias/como-preparar-transcricoes-para-analise-com-48c1b6.md) - Review identity and retention at every stage, from recording to summary, using fictional samples first.
* [How to review context before every AI call](https://ai.exponencialadm.net/okf/en/guias/como-revisar-o-contexto-antes-de-60b68b.md) - Inventory questions, history, documents and tools. Every stage should send only the information its task requires.
* [Using AI to extract TTPs from threat reports](https://ai.exponencialadm.net/okf/en/guias/ia-pode-revelar-lacunas-na-deteccao-c25d85.md) - A January 31, 2026 summary of a Microsoft-described workflow reports using AI to extract TTPs, map detection coverage, and flag possible gaps, with specialists reviewing and validating the results.
* [SimpleQA evaluates model factuality](https://ai.exponencialadm.net/okf/en/guias/simpleqa-como-avaliar-factualidade-e-alucinacoes-f8e34a.md) - Published on October 30, 2024, SimpleQA contains 4,000 human-written factual questions, each with a single indisputable answer. An autograder labels answers correct, incorrect, or not attempted; two annotators verified the reference answers.

# Follow

* [RSS](https://ai.exponencialadm.net/en/guides/feed.xml) - Guides
* [HTML](https://ai.exponencialadm.net/en/guides/) - Guides