AI Competition Radar: August 9–16, 2026

Three result sets cleared this week’s scoutability bar: each has a named winner or top placement, a dated source, and at least one concrete signal about what was built or who built it. The strongest evidence is not the prize money. It is the combination of a sharply defined workflow, a team that has shipped under constraints, and a credible next step beyond the stage.

Coverage window: August 9–16, 2026, Pacific time. A result announcement can fall inside the window even when the event itself finished earlier; the event date is shown separately. “Confirmed” means the public source names a winner or placing. “Pending” means the event or judging date is known but the public winner roster was not available in this check.

Fast scan

Event and status	What surfaced	Signal to test next
Aotearoa AI Hackathon Festival — confirmed	AUT’s South and City campus winners, plus the Waikato regional winner; more than 220 people attended the two AUT events.	Real-world problem framing, community and institutional partners, and a September national-judging path.
Physical AI Hackathon 2026 — confirmed	Team 23 won the Grand MVP and HK$100,000 at the Hong Kong final; more than 1,000 applications fed into a 160-plus-person final cohort.	Hardware execution and a possible investment / accelerator route, but the winning project itself is not described publicly.
UiPath AgentHack 2026 Global Finals — confirmed, secondary report	ADROSONIC entries InsureIQ and ADRO ACORD AI placed first and second in separate Maestro tracks; the competition reported 1,400 builders.	Enterprise insurance automation with a company-backed team; verify whether the entries are products, internal accelerators, or spinout candidates.
AgentHack 2026 by AI Season — pending result roster	The online event ran August 15–16 with a PKR 50,000 pool, but the public page still exposed the pre-result schedule rather than names.	Keep the page on the watch list; do not treat prize categories as a winner signal.

Confirmed results

1. Aotearoa AI Hackathon Festival: six named project signals across three regional legs

Event: Aotearoa AI Hackathon Festival, including AUT South Campus and City Campus, Auckland, July 7–8; Waikato regional leg, August 6–7. Organizer context: the festival is led by AI Forum New Zealand; the AUT events were delivered with Tū Ātea and SheSharp. Status: confirmed results, with national judging still ahead.

AUT’s August 13 recap says more than 220 people took part across two sold-out campus hackathons. The South Campus event had more than 100 participants and nine teams; the City Campus event registered 129 people. The program brought in students, whānau, alumni, industry professionals and more than 30 mentors, rather than limiting the field to experienced ML builders. 1

Five participants hold certificates at the AUT Aotearoa AI Hackathon Festival — *The AUT City Campus result page identifies Team Kailine as the campus winner and shows the event’s Aotearoa AI Hackathon Festival setting. 1*

The named results are unusually useful for first-pass screening:

Team Kailine — City Campus winner: Your Shopping Is One Call Away, an AI-enabled approach to making grocery shopping easier for older people.
Team Kaitiakidata — City Campus runner-up: Living Registry.
KPI-Kaitiaki-Positive-Impact — City Campus runner-up: EcoQuest.
Kai Loop — South Campus winner: a system intended to identify and redistribute food before it becomes waste.
Melting Pot and Just Enough — South Campus runners-up; Second Serve received the People’s Choice award, and Papatoetoe High School was Highly Commended.

The City Campus teams worked on sustainability, healthcare, accessibility and workforce development challenges from organizations including AI Forum New Zealand, Fisher & Paykel Healthcare and Woolworths New Zealand. The source does not provide public demos, repositories, founder profiles, or traction for the named projects, so the project titles should be treated as leads rather than validated products. 1

A separate Waikato result adds a stronger workflow-deployment signal. Wise Group names Ian Swainson, Emily Scerri-Davis, Piri Cash, Michael Tayler and Emma Campbell as the regional winners. Their prototype was designed with Workwise employment consultants to reduce administrative work and leave more time for supporting people with mental-health and addiction challenges into employment. Wise Group says the prototype used synthetic data and personas, not real client information. 2

Five Wise Group team members stand beside Aotearoa AI Hackathon Festival and University of Waikato signage — *Wise Group identifies the five-person team as the Waikato regional winner and describes the prototype’s employment-support workflow. 2*

Scout read: the best signal here is not that the projects won local events. It is that several teams started with a named operational problem and a partner close to the user. Kai Loop and the Workwise team are the first follow-ups to test for a working loop beyond the pitch. Ask for: a live or recorded demo; the data and evaluation protocol used after the event; who owns the code; and whether the team will continue to national judging at the Aotearoa AI Summit in Wellington in September. The public sources confirm that progression, but not a fundraise, customer contract, or standalone company. 1 2

2. Physical AI Hackathon 2026: Team 23 wins the hardware headline, but the build is still opaque

Event: Physical AI Hackathon 2026, Shenzhen preliminary round and Hong Kong final, August 2–5. Organizer context: launched by Brizan Ventures with SpacemiT and alloop.ai; co-hosted by HKSTP, OneRobotics, Direct Drive Tech, Zettlab, Z·Pilot and AgileX, with support from five Hong Kong universities. Status: confirmed result bulletin, updated August 12 by Shenzhen Government Online.

Team 23 won the Grand MVP and HK$100,000 at the August 5 final in Hong Kong Science Park. The event describes a 48-hour sprint across three tracks and five challenges. It received more than 1,000 applications worldwide, with more than 160 participants advancing; 78% of finalists were students. The total prize pool exceeded HK$300,000. 3

Six Team 23 members hold the Physical AI Hackathon 2026 Grand MVP award board — *The official recap identifies Team 23 as the Grand MVP winner and shows the HK$100,000 award; it does not describe the team’s winning prototype. 3*

The team profile is a useful, if incomplete, background signal: the six members included a business analyst, a master’s student in robotics at a Hong Kong university, a secondary-school student focused on full-stack hardware development, and three first-year undergraduates from Guangzhou and Shenzhen studying electronic information, electronic technology and automation. The recap names no individuals and gives no repository or demo link. 3

The organizer-side upside is clearer than the product detail. Winning teams will be considered for investment by Brizan Ventures, fast-track access to the HKSTP Ideation Program, and internship or full-time opportunities in the Brizan ecosystem. Those are pathways, not disclosed financing or acceptance. 3

Scout read: this is a strong hardware-talent sourcing lead, not yet a diligence-ready startup lead. The first question is simply, “What did Team 23 build?” Follow with the hardware bill of materials, autonomy boundary, test environment, repeatability outside the judging setup, and whether Brizan’s investment consideration has turned into a term sheet or program place. Until those answers appear, the prize and international applicant pool are the signal; the product is unverified.

3. UiPath AgentHack 2026: two enterprise entries from ADROSONIC reach the global podium

Event: UiPath AgentHack 2026 Global Finals. Status: confirmed by a Times of India report dated August 13; the public UiPath news listing did not yield a readable winner detail page in this check, so confidence is lower than for the first two items.

The report says InsureIQ won first place in the UiPath Maestro Case Track and also received Best Demo & Presentation. ADRO ACORD AI took second place in the UiPath Maestro BPMN Track. Both entries were developed by teams at ADROSONIC’s Lucknow Delivery Centre and apply agentic AI and automation to insurance workflows. 4

The same report gives useful funnel context: 1,400 builders entered, 333 projects were submitted, 203 progressed to evaluation, and 31 reached the live Global Finals. ADROSONIC’s source of strength is an existing enterprise engineering organization, not a disclosed standalone founding team. The report names Mayank, ADROSONIC’s founder, CEO and managing director, but gives no founder roster for either entry. 4

Scout read: this is the clearest enterprise signal of the week, but it may be a talent or product-partnership lead rather than a pre-seed company. Ask whether InsureIQ and ADRO ACORD AI are internal solutions, customer deployments, or ventures that ADROSONIC intends to spin out. Then request a demo, deployment metrics, model and workflow boundaries, governance controls, and a named customer reference. The public report supports placement and competition scale; it does not support a claim of revenue, funding, or independent company formation.

Unresolved or incomplete result bulletins

AgentHack 2026 by AI Season

AI Season’s official page says the online hackathon ran August 15–16, with winners scheduled for August 16 from 1:00 p.m. PKT. It lists a PKR 50,000 cash pool — PKR 25,000 for the Grand Champion, PKR 15,000 for second and PKR 10,000 for third — and judging dimensions including agentic depth, execution, impact, originality and demo clarity. At retrieval, however, the page still presented the pre-result schedule and did not name a winner, runner-up, project, or team. 5

That makes this a monitoring item, not a confirmed result. A prize page can establish the event’s scope and judging frame; it cannot substitute for a named podium. Recheck the organizer’s result or social channels in the next run before adding any team to the scouting table.

Next 7–14 days

Kipps.AI New Age India: results on August 21

The Kipps.AI event is in its build week through August 16, followed by demos and presentations August 17–19 and results on August 21. It is limited to 20 solo developers, requires every submission to use both Kipps Voice Agent and Chat Agent, and awards cash to the top five plus internships to the top two. The organizer frames the challenge around multilingual, multi-step workflows for sales, marketing and support. 6

For scouts, the result date matters more than the prize headline: the mandatory dual-channel requirement creates a clean comparison point for whether builders can orchestrate voice, chat, escalation and handoff rather than wrap a single prompt. The entry deadline has passed, so this is a follow-live and result-monitoring item.

AI Youth Festa 2026: application deadline August 21

This Korea–ASEAN AI startup competition is accepting applications through August 21 at 12:00 a.m. Pacific; the organizer lists the cutoff as August 21 at 4:00 p.m. Philippine time. NIPA organizes the program with support from the ASEAN–Korea Cooperation Fund. It is open to AI/ML startups or pre-founders linked to the Sustainable Development Goals across 11 ASEAN countries and Korea; 24 finalists will be selected, followed by an online acceleration program in September–October and a Philippines final competition and Demo Day on November 4–5. The top three prizes are listed as US$20,000, US$10,000 and US$5,000. 7

This is actionable for scouts who want early access to a cross-border pipeline, not a current result. The application requires a pitch deck, a 5–10 minute pitching video and evidence such as incorporation documents; the source also says that revenue, fundraising and award evidence may be attached optionally. 7

Scout priority

Request demos from Kai Loop, Team Kailine and the Waikato Workwise team. Their public descriptions start with specific users and workflows, but no public evidence yet shows repeatable deployment or company formation.
Clarify the corporate boundary around InsureIQ and ADRO ACORD AI. The podium is strong; the investment question is whether either entry is an independent venture, a product line, or a high-performing internal team.
Find Team 23’s missing build. The Physical AI result has the strongest hardware and ecosystem signal but the least product evidence.
Set two calendar checks: AI Youth Festa before the August 21 Pacific deadline, and Kipps.AI on August 21 for a named result. Keep AI Season’s AgentHack in the unresolved queue until its winner roster is public.

This week’s common pattern is practical problem selection under short build cycles. The next diligence step is the same for almost every lead: move from a stage photo and a prize to a runnable demo, an owner, and evidence that someone outside the judging panel wants the system.