AI Competition Radar: July 19-26, 2026

Quick scan

The July 19-26 window produced one competition with an awards ceremony inside the window and one in-window result bulletin for an earlier Demo Day. The strongest immediate follow-up is AI-mant's personalized virtual physics lab in Vietnam. Cyber Valley adds three research-driven European ventures with more legible technical theses, but its underlying Demo Day happened on July 10; the result was published later.

Event	Result status	What is public	Scout action
Vietnamese Student HackAIthon 2026	Confirmed, awards ceremony July 23	Three category podiums; AI-mant won Category B with PhyLab, an AI-powered personalized virtual physics lab. 1	Request the PhyLab demo and test whether the personalization is more than content retrieval.
Cyber Valley AI Incubator Batch 8	Confirmed result bulletin published July 20; Demo Day listed as July 10	Seven ventures pitched in Tübingen. Sehnsora, CoGraph, and OptiGrocer won audience/presentation, business-model, and use-of-AI awards; each received $10,000 in AWS credits. 2 3	Start with Sehnsora for assistive hardware, CoGraph for developer infrastructure, and OptiGrocer for retail data and retention questions.
Global AI Hackathon Series with Qwen Cloud	Closed July 20; winners scheduled August 17	The schedule lists $70,000 across five tracks, with judging July 28-August 11. 4	Keep the submission gallery on the August 17 result date, not the July 20 close.
H0: Hack the Zero Stack	Judging runs through July 24; winners scheduled July 31	No winner names are on the current schedule page. 5	Recheck on July 31 at 2:00 p.m. PDT.
FIND EVIL!	Judging runs through July 31; winners scheduled August 7	No winner names are on the current schedule page. 6	Recheck the security-focused projects after judging closes.
UiPath AgentHack	Winners scheduled August 4	The current schedule page lists the result date but no public winner list. 7	Check for agent demos and deployment evidence when the list opens.

Confirmed results

Vietnamese Student HackAIthon: AI-mant leads the student pipeline

The Central Committee of the Vietnam Student Association held the closing ceremony and awards presentation on July 23. The competition drew students from Vietnam and 10 other countries, including participants from more than 20 universities and academies outside Vietnam. The organizers said the projects addressed education, healthcare, public administration, digital banking, and student-community work. 1

The result set is divided into three categories:

Category	First prize	Second prize	Third prize
A	To Thanh Tien, Ho Chi Minh City University of Agriculture and Forestry	Tran Long Vu, Ho Chi Minh City University of Technology and Education	Truong Minh Hoang Phuc, FPT School Hai Phong
B	AI-mant, with members from the University of Finance-Marketing, Hanoi University of Science and Technology, Ho Chi Minh City University of Technology, and Foreign Trade University	GovBridge, College of Information and Communication Technology, Can Tho University	AI Navigator SDK, with members from Hanoi National University of Education, Minerva University, and Hanoi University of Science and Technology
C	Nguyen Thi Hong Trang, University of Science and Technology, University of Danang	Neko Core, Vietnam Maritime University	Vo Quoc Linh, University of Science, Vietnam National University Ho Chi Minh City

The Category B projects are the most legible from a startup-scouting perspective. AI-mant built PhyLab, an AI-powered personalized virtual physics laboratory. GovBridge built an AI-agent platform for administrative procedures. AI Navigator SDK presented a navigation software development kit. The article gives no public demo, repository, funding, revenue, or customer information for these teams, so these are diligence leads rather than investable company profiles. 1

The best first question for AI-mant is whether PhyLab uses a structured model of physical concepts and experiments, or whether it is mainly a conversational layer over existing learning material. For GovBridge, ask for the government workflow, source-of-truth documents, and an audit trail for agent actions. The public result announcement does not answer those questions.

Cyber Valley Batch 8: three awards, seven venture theses

Cyber Valley's public news page is dated July 20 and says Batch 8 concluded with a Demo Day in Tübingen. The linked Max Planck Institute release is dated July 15, while the incubator's event page lists the Demo Day itself as July 10. That makes this an in-window result bulletin, not an in-window event date. The date difference is worth preserving because scouts using event dates for sourcing should not count it as a July 19-26 competition close. 2 3 8

Seven teams presented:

Comparity Labs is building a European frontier-AI evaluation and testing platform for human-preference alignment and human-AI collaboration, including model comparisons, oversight, and safety interventions.
TrustMath AI is an adaptive learning companion for high-school mathematics, with support tailored to how each student learns.
NexusSoft is developing re-attachable robotic end-effectors with standardized interfaces for embodied-AI systems.
Lutamichi turns combat-sports training into structured feedback, motion analytics, and gamified competition.
Sehnsora combines a wearable haptic band with AI smart glasses to help blind and visually impaired people navigate indoor and outdoor environments, launching with smart shopping assistance.
OptiGrocer helps shoppers find nearby prices, consolidate loyalty cards, and track grocery carbon footprints, while giving store owners retention insights.
CoGraph uses a dynamic knowledge graph to help developers understand and manage changing software codebases. 2

The awards were not a single ranked podium. Sehnsora won the Audience & Presentation Award, CoGraph won Best Business Model, and OptiGrocer won the Use of AI Award. The Max Planck release names the founders: Mohammed Raihan Soniwala, Shweta Kadam, Yograj Yadav, and Abhishek Vijay Potekar for Sehnsora; Magnus Hornstein and Bela Thrän for CoGraph; and Ifeoma Veronica Nwabufo and Pauline Ornela Megne Choudja for OptiGrocer. Each team received $10,000 in AWS credits. 3

The follow-up paths are distinct. Sehnsora needs hardware reliability, accessibility testing, and evidence that the haptic-plus-audio interface works outside a controlled demo. CoGraph needs a look at installation friction, permissions, and whether the graph improves software maintenance enough to change developer behavior. OptiGrocer needs evidence of retailer data access, shopper repeat use, and whether the carbon-footprint feature affects purchasing or is only a reporting layer. Those are diligence questions, not claims made by the organizer.

Results still pending

The current public pages show several events that have crossed a submission or judging milestone without exposing a winner list. They should stay out of the confirmed-results section.

Agentic AI Innovation Challenge 2026: the online student event ran July 1-15, lists 299 participants and three non-cash prizes, and still directs readers to official channels for winner announcements. No winner names or project list appear on the current Devpost page. 9
Qwen Cloud: submissions closed July 20. Judging is scheduled for July 28-August 11, with winners on August 17 at 2:00 p.m. PDT. The result page is therefore not late; it is still future-dated. 4
H0: judging was scheduled through July 24, with winners on July 31 at 2:00 p.m. PDT. 5
FIND EVIL!: judging runs through July 31, with winners scheduled for August 7 at 12:00 p.m. EDT. 6
UiPath AgentHack: the schedule gives August 4 at 6:00 p.m. EDT for winner publication, but the current page does not name winners. 7
OpenAI Build Week: the organizer says submissions are closed and puts the winner announcement on or around August 12 at 2:00 p.m. PT. The project gallery is not yet the same thing as a ranked result. 10

Scout calendar

The next result dates with the clearest action value are July 31 for H0, August 4 for UiPath AgentHack, and August 7 for FIND EVIL!. OpenAI Build Week follows around August 12, and Qwen Cloud reaches its winner announcement on August 17. Keep the source pages bookmarked and wait for named teams, project artifacts, and award context before adding any of them to a sourcing pipeline.

For the immediate outreach queue, start with AI-mant if education or simulation is in scope. Then compare Sehnsora, CoGraph, and OptiGrocer against the scout's thesis: assistive hardware, developer infrastructure, and retail intelligence. None of the public announcements discloses financing or customer traction for these teams, so the next useful artifact is a demo or technical deck, not a valuation estimate.