AHOY Labs Outperforms Major Global Players on EnterpriseRAG-Bench AI Ranking

The achievement reinforces AHOY Labs' growing position in the enterprise AI landscape

October 06, 2026 2:48 PM EDT | Source: B2Press BV

Montreal, Quebec--(Newsfile Corp. - October 6, 2026) - AHOY Labs (aka Trouvé), the AHOY technology group, announced standout results on EnterpriseRAG Bench, a public, open source benchmark for enterprise AI. AHOY Labs ranked 8th worldwide out of 26 systems, outperforming several major enterprise AI players evaluated in the benchmark. 

With a score of 74.95, AHOY Labs performed 18.67 points above the benchmark average and 24.71 points above the median score, placing its result well above the typical performance across the systems evaluated.

Cannot view this image? Visit: https://images.newsfilecorp.com/files/8552/317625_8e160f0ee48419ed_001.jpg

AHOY Labs (aka Trouvé) Announces EnterpriseRAG-Bench Results: 1.03 Irrelevant Documents Per Answer

To view an enhanced version of this graphic, please visit:
https://images.newsfilecorp.com/files/8552/317625_8e160f0ee48419ed_001full.jpg

The benchmark gives each AI system the full archive of a simulated company: more than 500,000 emails, chat threads, files, tickets, sales records, and meeting transcripts. It then asks 500 questions. Some answers are spread across several documents. Some documents contradict each other. Some questions have no answer at all. Systems are scored on whether their answers are correct and complete, and on whether they bring back only the documents that matter.

“We built AHOY around sovereignty from the start, so organisations can keep control of their data and still get accurate, well-sourced answers while running AI within their own infrastructure. This benchmark result is a meaningful milestone for the system and provides a useful measure of its current capabilities. Our focus now is to continue improving performance and expanding what sovereign AI can deliver,” said Jamil Shinawi, Founder and Group CEO of AHOY.

Precision with less noise:

The biggest gap is in precision. One major enterprise AI player evaluated in the benchmark returned an average of 15.7 irrelevant documents per answer, compared with just 1.03 for AHOY Labs. Irrelevant documents are noise: they bury the real evidence, add cost and slow people down. They also make it harder to check where an answer came from. With AHOY, the evidence behind an answer is the evidence that matters. For governments, regulators and organisations that must defend every decision they make with AI, that difference is decisive. 

Built for sovereignty

AHOY's platform is designed to run inside the customer's own premises, on the customer's own hardware. It can operate fully offline, so sensitive data never leaves the organisation, and AHOY's internal testing shows it delivers the same benchmark result when running fully offline. Sovereignty is built into the system itself, not added as a hosting option, so every AHOY deployment is sovereign. The platform is suited to sectors where data cannot go to a public cloud, such as government, defence, healthcare and energy.  

About EnterpriseRAG-Bench

EnterpriseRAG-Bench was published with an accompanying research paper. It is open source under the MIT licence.

Leaderboard: https://huggingface.co/spaces/onyx-dot-app/EnterpriseRAG-Bench-Leaderboard

About AHOY

AHOY is a deep-tech frontier lab and infrastructure company delivering Physical AI: infrastructure that perceives, decides, and operates the physical world. The company's modular stack runs on-chip, at the edge, on-prem, and in hybrid cloud configurations to provide auditable action for cities, transportation networks, utilities, sovereign sites, and other complex operational environments. Founded by Jamil Shinawi in 2018, AHOY builds the systems that move, sense, and secure the physical world, operating from offices in Canada, Saudi Arabia, the UAE, the U.K. and others.

About AHOY Machine Studio (AMS)

AHOY Machine Studio (AMS) is AHOY's unified platform for building, governing and deploying Real-World AI. AMS Core is the control plane for identity, policy, signed releases and audit. AMS Forge is where teams and partners build, test and approve models, agents and applications. AMS Anvil deploys and operates them in the field. AMS is chip-agnostic and runs anywhere from cloud-connected setups to fully offline, air-gapped installations on the customer's own infrastructure, so organisations keep full ownership of their data, models and IP.

Media contact: Ayax Bader, CMO AHOY - press@mailahoy.com

Disclaimer: Rankings and scores are as published on the EnterpriseRAG-Bench leaderboard on 3 October 2026 and may change. EnterpriseRAG-Bench is the property of its owner. No endorsement is implied.

To view the source version of this press release, please visit https://www.newsfilecorp.com/release/317625

info

Source: B2Press BV

Ready to Announce with Confidence?

Send us a message and a member of our TMX Newsfile team will contact you to discuss your needs.

Contact Us