Dubai-Founded AHOY Labs Ranks 8th Globally on Enterprise AI Retrieval Benchmark

SISGAIN Technologies, an IT and software engineering company with a regional presence in the UAE, today shared its perspective on the reported EnterpriseRAG-Bench result of Dubai-based AHOY Labs, and what it signals for the UAE's enterprise and public-sector technology landscape.

According to a Khaleej Times report published on 6 October 2026, AHOY Labs, a UAE-founded AI company, ranked 8th globally among 26 systems evaluated on the benchmark. The benchmark tests how effectively enterprise AI systems retrieve, verify and use information from large organisational datasets. Reports also state that AHOY outperformed products from OpenAI, Amazon, Microsoft, Google and NVIDIA.

SISGAIN views the development as a practical example of retrieval-first enterprise AI: artificial intelligence judged not by how fluently it answers, but by how accurately it finds the right evidence inside an organisation's own data.

What Has Been Reported

According to published reports, AHOY Labs' results compare with OpenAI File Search as follows:

Key points:

• Overall score: AHOY Labs scored 74.95, ahead of OpenAI File Search (61.03).
• Other systems: Microsoft Azure AI Search scored 48.42 and Google Vertex AI Search scored 41.87.
• Answer accuracy: 79.6% of questions answered correctly, against 69.8%.
• Retrieval precision: an average of 1.03 irrelevant documents per response, roughly 15 times fewer than OpenAI File Search.
• Ranking: 8th among 26 systems evaluated, not first. SISGAIN does not describe this as the world's best AI.

From Fluent Answers to Verified Evidence

Enterprise users do not just need an answer. They need confidence that the answer is grounded in the right records, drawn from potentially millions of internal documents. When a system returns many irrelevant documents, staff must check sources themselves, noisy context can lead to wrong answers, and the time AI was meant to save is spent on verification.

That is why the precision figure is the most telling part of the result. It reflects a shift in the question being asked, from "can this AI write a convincing answer?" to "can it find, and show, the right evidence?"

SISGAIN notes that the result concerns an enterprise retrieval system, not a head-to-head comparison of underlying language models. It is best understood as evidence that retrieval quality, ranking and data pipelines now matter as much as the model itself.

Varun Arora, Director of Innovation & Growth, SISGAIN Technologies, said:

"What stands out is not the ranking but the 1.03 irrelevant documents per response. In enterprise AI, the real question is whether the system found the right evidence. Retrieval precision is what separates a dependable business tool from an impressive demo."

A Benchmark Built Around Real Enterprise Conditions

EnterpriseRAG-Bench uses a simulated enterprise archive of approximately 500,000 records, including:

• Emails and chat conversations
• Files and tickets
• Sales records
• Meeting transcripts

Systems are asked questions where the answer may be spread across several documents, contradicted by other records, or missing from the dataset entirely. That makes the test closer to a real enterprise knowledge problem than a conventional chatbot test.

The result also fits a wider industry movement. In July 2026, DevRev introduced Enterprise-Bench, an open benchmark that evaluates enterprise AI agents under production-like conditions involving fragmented data, siloed systems and permission boundaries. Together, these efforts show AI evaluation moving from generic model intelligence toward real enterprise performance.

What This Means for the UAE's Technology Ecosystem

AHOY Labs states that its platform is designed to run entirely on a customer's own infrastructure and can operate offline. For organisations in the UAE and GCC, the announcement carries several practical signals:

• Sovereign AI: government, defence, healthcare, energy and financial services organisations face stringent requirements around sensitive information. Control over where AI runs is becoming a purchasing criterion.
• Precision over fluency: buyers are starting to ask how accurately AI finds evidence, not only how natural its answers read.
• Benchmark-led buying: realistic, enterprise-grade evaluation gives organisations better questions to put to vendors and development partners.
• Local capability: a UAE-founded company reaching a globally competitive position supports the country's ambitions in AI adoption, secure enterprise technology and local technology development.
• From pilots to production: the next maturity step for many organisations is moving from generic GenAI experimentation toward secure, retrieval-first systems built on their own data.

SISGAIN Technologies provides AI development solutions in Dubai, including RAG Solutions, AI Agents / Copilot Development and AI Consulting, for organisations building accurate, secure and production-ready AI systems across sectors, including government, healthcare and financial services.

Data Sovereignty: The Question to Watch

AHOY Labs' offline and customer-hosted positioning places data control at the centre of any enterprise AI decision. Published reports confirm the benchmark results and the deployment claims, but the benchmark does not test an individual organisation's security, compliance or integration environment.

For organisations considering similar private or retrieval-based AI, SISGAIN recommends addressing several questions early:

1. Where is data processed and stored, and does this align with applicable UAE and sector-specific requirements?
2. Are access permissions in source systems respected when AI retrieves and presents information?
3. Can every answer be traced back to the records that support it?
4. Who is accountable for reviewing AI-generated answers before they inform decisions?

Any further detail on AHOY Labs' architecture and benchmark methodology should come from AHOY Labs and the EnterpriseRAG-Bench authors.

What Organisations Should Do Now

Businesses and public-sector teams in the UAE exploring enterprise AI can take the following steps:

1. Test retrieval quality, not just fluency, measuring how accurately and precisely a system finds supporting evidence.
2. Evaluate on realistic data, including fragmented, contradictory and incomplete records.
3. Decide the deployment model early, since cloud, private cloud, on-premises and offline options carry different governance implications.
4. Design governance first, with access permissions, data residency and audit trails defined before deployment.
5. Require source citations, so every answer points back to the records behind it.
6. Pilot before scaling, testing in one department or knowledge domain before wider rollout.
7. Choose a partner with engineering depth. Enterprise AI succeeds on integration, security and maintainability. An experienced software development company in Dubai can connect AI to existing ERP, CRM and data platforms.

The significance of this development extends beyond a single ranking. It reflects a direction of travel toward enterprise AI that is precise, evidence-based and under the customer's control.

These observations are forward-looking and should not be interpreted as predictions of specific outcomes. Benchmark results reflect performance under the stated test conditions and do not guarantee performance in any individual organisation's environment.

About SISGAIN Technologies

SISGAIN Technologies is an IT and software development company in Dubai delivering AI development solutions in Dubai and intelligent solutions, application development, healthcare technology, enterprise and product engineering, cybersecurity, cloud and infrastructure, digital transformation, UI/UX and QA, and consulting services.

The company works with organisations across the USA, UAE, UK, Canada and other international markets, with an office in Dubai, UAE.

Learn more at SISGAIN Technologies.

This press release is issued by SISGAIN Technologies. Details about AHOY Labs, EnterpriseRAG-Bench, the reported scores and rankings, and DevRev Enterprise-Bench are based on information published by third-party media and have not been independently verified by SISGAIN Technologies. Forward-looking statements represent observations about emerging technology trends and are not predictions of specific outcomes.