Skip to content

Applied AI lab

IAna

AI assistants that answer with the real documentation in front of them.

Manuals, syllabi, regulations: IAna builds RAG systems and agents that cite the source instead of making it up. Two in production, open to try.

A project by Cristian Pérez · AI Engineer

OposIA in actionLive
Question
Retrieved · tema 12 · 0.87

…el plazo para la interposición del recurso de alzada será de un mes, si el acto fuera expreso…

Retrieved · tema 12 · 0.81

…transcurrido dicho plazo sin interponerse el recurso, la resolución será firme a todos los efectos…

Answer

One month if the act is express (Law 39/2015, art. 122.1). After that, the decision becomes final.

Illustrative example of the OposIA flow

SourcesAnswers from the documentWith a citation to the exact passage, not from memory.
AgentsFlows that plan and verifyOrchestrated with LangGraph, not a loose chat.
ProductionDeployed on its own domainYou can open it and use it today.

Projects

Deployed systems, open to try.

OposIA dashboard with its modes: chat with the syllabus, study a topic, practice questions, mock exam and progress

OposIA

Live · beta

Prepare your civil-service exam with tests generated from the official syllabus.

  • Exam-style questions with the official penalty (1/3 per miss)
  • Timed 100-question mock exams with instant scoring
  • Progress per topic and review sessions on your weak points

RAG · ChromaDB · Cohere rerank · FastAPI · PostgreSQL

99.93% available over 24 h · Updated last month · History

MasterScroll onboarding: chapter 1 of 7, asking about D&D experience

MasterScroll

Live · beta

Create your D&D 5.5e character with an assistant that knows the rulebooks.

RAG · LlamaIndex · ChromaDB · LangGraph · FastAPI · WeasyPrint

99.79% available over 24 h · Updated last month · History

In the lab
MarketMind

Multi-agent market analyst: price, news, technical and fundamental analysis synthesised by a supervisor.

Planned

How it works

What IAna can do with your documents.

01

An assistant over your documentation

Manuals, procedures, regulations or syllabi: the assistant answers with the source in front of it and cites the passage.

02

Agents that do tasks, not just chat

Flows that search, compare and verify before answering. With state, memory and clear limits.

03

From prototype to production

API, frontend, database and deployment on its own domain. What you see here is running, not sitting in a notebook.

Let's talk →

About

Cristian Pérez · AI Engineer

Computer Engineer with a Master's in Artificial Intelligence. I build RAG systems and LLM agents that work over real documentation and actually ship: API, frontend, data and infrastructure.

IAna is the lab where those systems get built, measured and published. Each project solves a concrete problem and can be tried without signing up for anything.

What I work with

LLM & retrieval
RAG · LangGraph · LangChain · LlamaIndex · ChromaDB · Cohere Rerank · Groq · Ollama
Backend
Python · FastAPI · PostgreSQL · Redis · SSE
Frontend
TypeScript · React · Next.js · Tailwind
Infrastructure
Docker · Coolify · Hetzner · Cloudflare · GitHub Actions

Contact

An assistant over your documentation, an agents project or a job offer? Write to me.

iana.soporte@gmail.com