---
title: About — Corbenic AI
description: A Delaware C-Corporation headquartered in Belgium, building the verified memory layer for artificial intelligence. Meet the team.
---
[Skip to content](https://www.corbenic.ai/en/about#main-content)
[ Corbenic](https://www.corbenic.ai/?hsLang=en)
- [Home](https://www.corbenic.ai/en/?hsLang=en)
- [Product](https://www.corbenic.ai/en/galahad?hsLang=en)
- [Pricing](https://www.corbenic.ai/en/pricing?hsLang=en)
- [Savings check](https://www.corbenic.ai/en/savings-check?hsLang=en)
- [About](https://www.corbenic.ai/en/about)
[Check your savings](https://www.corbenic.ai/savings-check?hsLang=en)
About
# Built on results, not promises.
We started by building Merlin, a deduplication tool that strips redundant data before it ever reaches a model. That work exposed a bigger problem: the same computation being paid for over and over. Galahad is a model-agnostic harness that sits between your workload and your silicon. Any model can run inside it; nothing about it is tied to us.
[Run the savings check](https://www.corbenic.ai/savings-check?hsLang=en)
Team
## Who is building Galahad
A small team with big energy. Combining our experience to go after something ambitious.
### Sietse Schelpe
CEO — Deep Systems & Science
Leads product vision, bare-metal C++ architecture, and formal verification (Z3, Lean). Author of the four arXiv papers behind the Galahad harness.
[sietse@corbenic.ai](mailto:sietse@corbenic.ai)
### Toon Colson
CSO — Sales & Partnerships
Seven years of B2B sales experience. Proven track record driving enterprise revenue growth and strategic partnerships.
[toon@corbenic.ai](mailto:toon@corbenic.ai)
### Marloes Decraemer
CPO — Product & Go-To-Market
Translates deep systems research into deployable enterprise products. Owns design partners and customer onboarding.
[marloes@corbenic.ai](mailto:marloes@corbenic.ai)
Advisors
- Gwen le Tiran, PhD — Head of R&D Group at Saint-Gobain
- Irene Balmes, PhD — Lead Data Scientist at Kynaps
Published research
- arXiv:2609.39358 [Working Around the Compute Ceiling: Byte-Exact Memory in Galahad Makes LLM Reading a One-Time Cost](https://arxiv.org/abs/2609.39358) October 2026
- arXiv:2607.23806 [A Frozen 12B Beats Frontier Models on Verified Work: 100% Accuracy, 0 Tokens, Bit-Exact, Forever](https://arxiv.org/abs/2607.23806) July 2026
- arXiv:2607.14431 [Smarter and Cheaper at Once: Byte-Exact KV-Cache Grafting Turns a Frozen Small Model into a Verified-Knowledge Flywheel](https://arxiv.org/abs/2607.14431) July 2026
- arXiv:2605.09990 [Merlin: Deterministic Byte-Exact Deduplication for Lossless Context Optimization in Large Language Model Inference](https://arxiv.org/abs/2605.09990) May 2026
- arXiv:2605.09611 [Byte-Exact Deduplication in Retrieval-Augmented Generation: A Three-Regime Empirical Analysis Across Public Benchmarks](https://arxiv.org/abs/2605.09611) May 2026
## Which of your workloads is bleeding budget?
We can probably safe you money. Test it for yourself with our one GPU for non-commercial use trough [GitHub.](https://github.com/corbenicai/galahad)
 Corbenic
[mailto:email@email.com](mailto:email@email.com)
Product
- [Galahad GitHub](https://github.com/corbenicai/galahad)
Company
- [Corbenic AI](https://www.corbenic.ai/en/about)
Contact
[sales@corbenic.ai](mailto:sales@corbenic.ai)
© 2026 Corbenic AI, Inc. · Delaware C-Corp
[Privacy](https://www.corbenic.ai/privacy?hsLang=en) [Terms](https://www.corbenic.ai/terms?hsLang=en)