Inception42 Pitches Seraj For Arabic Government And Enterprise AI

Inception42 has launched Seraj, an enterprise AI model built for Arabic-language government and business applications.

The company said the model was developed with Microsoft and is available through Compass, Core42’s sovereign AI platform.

The model is positioned around a regional enterprise problem: organisations want frontier-model capability, but Arabic precision, dialect handling, cultural context and safety behaviour remain harder to guarantee than English performance.

Inception42 chief executive Ashish Koshy said regional organisations have had to choose between global AI capability and meaningful Arabic performance.

The source material is company-led, so the article treats market and performance claims as Inception42 and Microsoft claims rather than independent proof.

The company also framed Seraj as a product for organisations that need Arabic context without giving up multilingual model capability.

Seraj is aimed at government services, legal analysis, customer engagement and knowledge management.

Koshy said the model is designed to help organisations deploy Arabic AI at scale with more confidence.

The launch therefore depends on enterprise trust in Arabic fluency, not only on access to a large global model.

Microsoft Model Underpins The Arabic Training Claim

Seraj is built on a frontier Microsoft model, Inception42 said.

Inception42 said targeted mid-training techniques were applied to curated Arabic datasets covering linguistics, cultural knowledge, safety scenarios and domain-specific enterprise content.

The company did not name the individual datasets or say whether external auditors reviewed the training material.

The company said that approach improved Arabic performance while preserving reasoning and multilingual strengths.

It also said the model can support English and other languages, which matters for bilingual Arabic-English government and corporate workflows.

Inception42 did not include third-party benchmark tables or lab methodology for the performance claim.

It presented Seraj as a purpose-built enterprise model rather than an audited public comparison with rival Arabic or multilingual systems.

The disclosed training description names the dataset categories, but it does not identify dataset owners, data volumes, evaluation scores or red-team results.

Microsoft UAE Names Regional AI Adoption Goal

Microsoft UAE AI and enterprise solutions director Rima Semaan said the collaboration with Inception42 reflects a shared commitment to expanding AI’s real-world impact across the region.

She said language coverage determines whether advanced AI can be relevant and accessible for Arabic-speaking organisations.

Semaan also linked Seraj to responsible AI adoption across governments, enterprises and critical industries in the UAE.

Inception42 did not identify a government contract, paying customer or production deployment.

Inception42 listed use cases including document understanding, summarisation, translation and question answering.

It also listed workflow automation, bilingual Arabic-English applications, retrieval-augmented generation and knowledge-intensive tasks.

The named sectors include government, education, legal services, Islamic studies, media and financial services.

Customers And Benchmark Evidence Remain Undisclosed

The launch gives Inception42, Microsoft and Core42 a named Arabic AI product inside a sovereign AI platform, but the commercial record is still incomplete.

Inception42 did not disclose pricing, service-level terms, customer names, deployment dates, adoption metrics, independent Arabic benchmark results or public safety-evaluation methodology for Seraj.