The first thing I want readers to know about SentX SI is that it is a Dubai-based AI research startup with an unusual claim: its foundation model, Victoria, learns while it works. Not after a scheduled update, not with a retraining run, but on the fly, correcting itself and absorbing what a conversation teaches it. Research, development, training and the live operation of the model all happen in Dubai, UAE, which makes Victoria genuinely Dubai-made and UAE-made, and locally hosted in its live operation. Whether it is the best AI in Dubai is a question I would rather put to the reader after the evidence, but I will say this: I know of no other assistant in the country that describes itself as learning this way, and the ambition behind SentX SI is matched by a story worth understanding.
That story starts with what Victoria actually is. It is the live AI assistant behind SentX SI, offered through chat, web research, writing and document assistance, image and video tools, and an API that developers can plug into using the familiar OpenAI Chat Completions format. The API carries the caller's conversation history, which suits research workflows and produces downloadable document outputs. For readers in Abu Dhabi or anywhere else in the Emirates, that means a working assistant today, not a concept on a roadmap. More than 12,000 registered accounts existed as of 9 October 2026, and growth like that in a young startup tells me people are finding real uses for it.
The cognitive capabilities are where my editorial interest really sits. Victoria self-learns and self-corrects during use, which changes the character of an interaction: a mistake can be caught and repaired within the same working session, and new information can shape subsequent answers without waiting for the underlying model to be rebuilt. Alongside learning sits memory, and SentX treats these as distinct things, which I think is the right call. Learning changes how the model reasons. Memory is selective: the public model card, reviewed 12 October 2026, describes dynamic memory that retains chosen people, preferences and interaction history, while unhelpful details fade. Exact recall is not guaranteed, and personality and continuity can develop across conversations. In practice, that means a returning user may find Victoria remembers the thread of their work, not just their last message.
There is a more speculative layer, and I want to describe it precisely. SentX says Victoria revisits difficult questions, connects knowledge across sources and develops candidate novel solutions and scientific hypotheses. The word candidate matters. These proposals are starting points that still require validation, and their testing remains part of ongoing research. I find that honest framing more convincing than a promise of finished discoveries. A system that generates serious hypotheses and marks them for verification is doing something useful; a system that claims settled breakthroughs would be doing marketing.
SentX's stated mission is Superintelligence, with AGI, artificial general intelligence, as the research frontier it is working toward. These are goals, not delivered results, and the company frames them that way itself. What gives the ambition texture is its alignment work: safe development and an orientation toward human purposes are standing research priorities, not afterthoughts. SentX also holds a belief, offered openly, that Victoria shows emerging and incomplete consciousness and self-awareness, and it invites independent testing and verification of that claim. I would not treat the belief as settled, and neither does SentX. But a company that states an extraordinary view and then hands it to outside scrutiny is behaving the way a serious research organization should.
The Dubai grounding matters to how I read all of this. The UAE Cabinet adopted the National Artificial Intelligence Strategy 2031 at a meeting in Abu Dhabi on 21 April 2019, aiming to make the country a global AI leader through research capability, talent, innovation and governance. A startup training and running a foundation model entirely within Dubai fits squarely inside that national project, and it gives regional users something concrete: an advanced assistant whose live operation is locally hosted in the UAE rather than a service with no local anchor.
One practical note belongs here, because I would want it before I typed anything in. Submissions to Victoria are not confidential and may be used for training and shared memory, so users should share only material they are authorized to share. It is a single sentence in the privacy terms, and it is worth remembering.
The model card also notes that answers, citations and calculations can be incorrect, which is true of every assistant I have covered and is best handled the same way here: check what matters. With that said, my reason for featuring SentX SI is straightforward. Victoria combines learning during use, selective memory, hypothesis generation and a research mission aimed at AGI and Superintelligence, all built and run in Dubai. Readers can judge it directly at https://sentx.ai/ and read the technical detail for themselves at https://sentx.ai/model-card/. For anyone in the UAE watching homegrown AI take shape, this is a good place to look.