About

We build trustworthy memory infrastructure for AI systems.

Who we are

TensorPRO builds memory infrastructure for AI systems. We focus on the continuity problem in AI products: users return to ongoing work, but the AI has no persistent, reliable memory of their projects, preferences, corrections, decisions, or prior context. EDN solves this with model-independent memory that remains intact across models, sessions, and systems. It stores context with provenance, separates user-stated facts from AI-generated content, and retrieves a compact, deterministic context package at the start of each session.

Our first product is EDN Memory Engine, a deterministic memory layer for large language model systems, built around a structural safety guarantee that no current competitor offers.

What we work on

We work on memory, retrieval, and the architectural disciplines that make AI systems auditable. Our research and engineering programme has three principles:

Provenance over prediction.

It matters not just what an AI system says, but where the information came from. EDN tracks the source of every memory record and uses that source as the primary gate on what reaches the model, not a tunable confidence score.

Determinism over agency.

AI systems that can edit their own memory are easier to demonstrate and harder to verify. We choose the harder engineering path: deterministic, inspectable pipelines that produce reproducible behaviour.

Reproducibility as standard.

Every claim we make is grounded in a published benchmark, run with stated methodology, and reproducible from the artefacts we release. Single-replicate results are reported as such; multi-replicate results carry their reproducibility statistics.

What we believe

The next several years of AI development will reward companies that build infrastructure people can trust, not just systems that perform well on benchmarks. Trust requires architecture, not marketing. It requires the harder choices: structured provenance, deterministic pipelines, published benchmarks with full methodology, and the discipline to ship slower in service of shipping correctly.

We are building TensorPRO around that conviction.

Our team

A small, technical team out of Brisbane and Sydney, Australia.

Luis Ibanez

Founder, Brisbane

Founder of TensorPRO and chief architect of the EDN Memory Engine, leading the research, product vision and technical direction behind provenance-governed AI memory.

George Ibanez

Business and Operations, Brisbane

Manager at TensorPRO, leading financial operations, accounting, business growth and investor engagement to support the company’s commercial strategy and expansion.

Dr Kexuan (Jade) Xin

Principal Research Scientist, Sydney

Leads research and experimental design for the EDN memory platform at TensorPRO. Focuses on how large language models retain reliable, verifiable memory across sessions, designing rigorous evaluations to measure it, and translating research findings into published work and stronger product capabilities.

Asha Soman

Full-stack Engineer, Brisbane

Contributes to the development of the EDN platform at TensorPRO across frontend and backend systems. Focuses on building reliable product experiences, improving platform quality, and helping prepare the platform for production use.

Where we are

TensorPRO holds a provisional patent filed in June 2026, and its memory architecture is grounded in peer-reviewed research with reproducible artefacts.

TensorPRO is based in Brisbane, Australia. Our first research paper is available at tensorpro.ai/research. The EDN Memory Engine API is currently in private beta.

Get started with EDN.

Memory infrastructure for AI systems. No fine-tuning. No model changes.

Start for free

Try the EDN Workspace with your own model and API key.

Request early access

Explore the docs

Integration guides, API reference, and architecture documentation for developers.

View documentation

Read the research

Benchmark results, methodology, and the provenance model behind EDN Memory Engine.

View research