Skip to content
All projects
Case study · engineering

Revisa Express

A six-hour expert review, delivered in fifteen minutes — in production, every day.

80% Less Analysis Time

Context & Constraints

Academic consulting teams spend a disproportionate amount of time on first-level readings: understanding a client’s paper, diagnosing structural/methodological issues, and assessing quality before deep revision begins.

This initial diagnostic phase was time-consuming, costly, and highly dependent on senior reviewers’ availability. The problem wasn’t writing alone — it was the lack of a fast, structured, and repeatable way to apply academic judgment at scale.

Architecture

A production multi-agent analysis system structured around the 5 pillars of academic writing (Clarity, Methodology, Relevance, Technical Quality, Norms). Each pillar is implemented as a specialized AI agent that produces structured evaluations. A Redactor agent then consolidates all analyses into a single, standardized diagnostic report exported as a branded PDF. Runs in production with Langfuse tracing, Sentry monitoring, and automated health checks.

Revisa Express — multi-agent review pipeline

My Role

Co-founder and sole engineer. Translated tacit academic evaluation criteria into explicit, modular AI agents. Designed, implemented and operate the multi-agent orchestration and backend architecture (Python/Django/CrewAI/PostgreSQL/pgvector/Celery/Docker) — a system my own business depends on daily.

Results & Validation

5 specialized agents mirroring academic evaluation

Automated PDF diagnostic report generation

In production with Langfuse tracing and Sentry monitoring

Reduced analysis time by ~80%. High-quality academic evaluation is not intuition—it is structure. By codifying tacit expert criteria into 5 distinct agents, we proved that complex intellectual critique can be scaled without sacrificing rigor.

Stack

PythonDjangoCrewAIPostgres

Building something where reliability matters?

I design, ship and operate systems like this one.

Let's talk

Building GenAI where reliability matters? Let's talk.

leonardo@leonardosa.pro

Madrid, Spain · from August 2026 · English / French / Portuguese / Spanish

© 2026 Leonardo Costa de Sá