Featured · Aug 3, 2026
Research
Avra · Dispatch
Dispatch
Research notes, engineering work, and platform writing from the team. Relational models, temporal graphs, and the systems we put behind enterprise decisions.
- Sep 4, 2026Platform
LLMs vs. RFM: why ChatGPT can't predict your customer's next move
LLMs read, summarize, and reason over language, but they can't tell you which customer will churn or default. Why prediction over structured business data needs a different kind of foundation model.
Júlia Jordão·7 min read - Aug 21, 2026Research
Graph foundation models in production
Part 3 of a three-part series on Avra's Graph Foundation Models: the measured performance gains from a live client deployment — where relational embeddings were the single largest driver of improvement — and what we're building next.
Jonas Rodrigues·6 min read - Aug 11, 2026Research
Inside the Graph Foundation Model
Part 2 of a three-part series on Avra's Graph Foundation Models: why heterogeneous graph neural networks, what the proprietary Large Knowledge Graph is, and the engineering decisions behind the model — Matryoshka embeddings, sampling at scale, and hard constraints.
Jonas Rodrigues·16 min read - May 26, 2026Company
Forward Deployed Engineers: the future of enterprise AI is built inside the customer
What Avra's FDEs actually do, and why enterprise AI in Brazil doesn't work without consultative deployment
Júlia Jordão·9 min read - Sep 24, 2025Research
Decoding the DNA of the Brazilian economy
Why we built a Graph Foundation Model to understand the Brazilian SMB ecosystem — and how graph-based AI can decode the risk and potential of millions of companies traditional systems treat as a black box.
Bruno Alano·6 min read - Sep 24, 2025Company
The AI revolution in credit will be hyper-personalization
Reflections from the CEO on Avra's starting point — and on why the right question in credit is no longer whether to lend, but under what conditions the operation makes sense for the company at that moment.
Viviane Meister·2 min read