c0mpiled_
← back to articles
hack·march 30, 2026·17 min read

c0mpiled-7: San Fransokyo, pt2

by james tan2.4k reads

Overview

It was great to be back in Tokyo for our second San Fransokyo event.
We met a new group of students and got to catch up with a number of returning participants. In less than four months since our last trip, the quality of what teams shipped stood out immediately. The technical depth and attention to detail continue to define the submissions we see from Japan. We especially liked the projects that connected YC’s Request for Startups with needs that are specific to the Japanese market.

By the numbers, we had 230 attendees and guests, with 65 project submissions built in under four hours. Duncan Ross gave a strong talk on building, experimenting, and being comfortable with failure. The University of Tokyo IPC team ran a networking session that worked really well. It helped people form teams quickly and brought together groups that had never met before in a format that felt genuinely new.
And a big thank you to Mori Building for making the event possible and supporting us with the space, food, and drinks.

Reflections

  • Good: Pre hackathon team matching worked extremely well. We saw a lot of new teams form that would not have come together otherwise. People were genuinely excited to meet, learn, and build together, and some teams had very strong chemistry right away.
  • Good: Engagement was meaningfully higher this time. Submissions were tightly aligned with the YC RFS and teams took a more consultative approach. The live translation support from Jia and Jian at Shisa AI played a big role here.
  • Good: Having Duncan Ross spend time directly with teams was a major unlock. You could see the impact in the projects. Submissions were more focused, more refined, and clearer in what they were trying to do.
  • Good: Judging ran much more smoothly than previous events.
  • Good: Submission quality improved across the board. Projects were more complete and more engaging.
  • Good: Running pitches in parallel with judging worked well. It filled downtime and gave every team a chance to present to the full audience.
  • Good: Jia and Jian Shen built a project analyzer during the hackathon itself. It reviewed submissions for completeness and evaluated whether teams used agents or multi agent systems. Standout moment.
  • Bad: Food. We under ordered for the first time across our events. This should not happen again.
  • Bad: Presentation quality still needs work. Strong technical work often fell short because it was not communicated clearly to judges or the audience. We tried to coach teams live, but this needs to be systematized. Likely a simple site with clear guidance.
  • Observation: Many participants asked whether their idea was “good,” fundable, or how to pitch it. A common pattern was focusing on a narrow instance of a problem. What we want to push is tying that instance back to a larger structural issue. It is not just that a form is broken, it is that the system around it is broken. This is a muscle we want to keep building.
  • Observation: Fewer teams than expected used multi agent systems. This was surprising. It reinforces that the US, and SF in particular, operates in a different reality when it comes to AI tooling. It also highlights how large the opportunity still is for builders in this space. We are watching this closely over the years.

Judges

In-person:

Remote + Sponsors:

Prizes & Recognition

1st Place ($5,000): Engram

2nd Place ($2000): Noob

3rd Place ($1000): Yaorozu

Most Venture-Backable: Mercy

Best Design: PM Studio

  • Queenie Wu
  • Tomoyasu Hirano
  • Tsubasa Kawakami
  • Gaku Noguchi

Most Impressive Engineering Lift: Origin

All Submissions

3 Sketches
Description: PdM 3 Sketches is a tool that takes a product goal and meeting transcript as input, then automatically generates and live-previews three distinct UI sketches (React + Tailwind) from an expert PdM perspective.
Team: Satoru HANADA, Yanagisawa Kyosuke
GitHub: https://github.com/2222-42/PdM-three-sketches
Video: https://drive.google.com/file/d/1RxiCnbZdFqkzdIQn5c0SPn2Pwh8u-0du/view#

42Tokyo
Description: Maisen is an AI-powered dashboard that automatically clusters scattered user feedback into actionable problem groups using LLM and VLM, generating summaries, reproduction steps, and severity scores to help product teams prioritize fixes faster. The system captures session recordings alongside feedback, enabling teams to see exactly what users experienced while reducing hours of manual feedback analysis into minutes.
Team: Fumiya Endo, Sakurai Moulay Ali, Taisuke Fujise, Takuma Kono
GitHub: https://github.com/sakemyali/c0mpiled-Tokyo2026
Video: https://www.youtube.com/watch?v=FaoSwetcCgQ#

Air Shift
Description: Air Shift is an AI-native guidance platform that supercharges physical workers by providing real-time, multimodal instructions through smart glasses and an iOS app. By instantly downloading master-level expertise into the hands of rookies, it solves the global skilled labor shortage and enables anyone to perform complex industrial tasks from day one.
Team: Shohei Yukawa, Yabe Tatsuki
GitHub: https://github.com/test-aitest/AirShift_new
Video: https://youtu.be/JtbNGVPfZsU#

airflo
Description: airflo is a voice-first AI product manager that turns spoken thoughts into organized tasks, auto-schedules priorities, and notifies your engineers, designers, and GitHub, all hands-free. It closes the loop between ideation and execution by drafting mocks, generating cycle trailers, and syncing with your meeting notes so nothing falls through the cracks.
Team: Marii Komiya, Yuki Obata
GitHub: https://github.com/mariikomiya/airflo
Video: https://drive.google.com/file/d/1q8g1sphwIoZwZZjWBg4LLFl-VX0Q_GMK/view?usp=sharing#

AuthentiFlow
Description: We're building AuthentiFlow — an AI agent that turns a day-one hire into an expert authenticator by auto-detecting inspection checkpoints, interactively highlighting where to look, and guiding users step-by-step through professional-grade sneaker authentication with transparent verdicts. We're replacing the $2B+ manual authentication bottleneck at StockX, GOAT, and secondhand marketplaces.
Team: Ryotaro Shimada, Sho Fukada, Haruki Yajima
GitHub: https://github.com/ShoFukada/yc-c0mpiled-hackathon
Video: https://drive.google.com/file/d/141RvxMKgFhluZCECKr7WB5HiPQ9-Ihwm/view?usp=sharing#

Brand Buddy
Description: Brand Buddy is a two-sided marketplace that matches brands with the right creators using verified public data and AI scoring — no agencies, no guesswork. Brands find creators whose audiences actually fit their customers, and creators get discovered by brands that are genuinely relevant to what they make.
Team: Shawn, Jack, Ashley, Paddy
GitHub: https://github.com/juminoz/ai-agency
Video: https://youtu.be/UrJU0-AwF68#

BREAK x MAKE
Description: HarborClaw: AI that notices when you're not okay — and reaches out before you have to ask. Built on VIVIDAS, an AI for Government Platform — so when disaster strikes, it already knows your context and speaks to you as your neighbor.
Team: Shinsuke Kaneko, Yusuke Endo, MASAHIRO TAKECHI, MASAHIRO OKAMURA
GitHub: https://github.com/shintaku81/c0mpile_VIVIDAS
Video: https://drive.google.com/drive/folders/1IhxGQsM-4ZvGhkVDKg-oGyVaCrtk5eq2?usp=sharing#

carecall
Description: An AI that calls patients to remind them of appointments, medications, and prescribed exercises through natural phone conversation — no apps needed.
Team: liwgou
GitHub: https://github.com/liwengggou/hospital
Video: https://www.loom.com/share/8b2bd886489c43af9a8e924dae1020b3#

CAREGLASS
Description: AI-powered smart glasses for caregivers that provide real-time guidance to perform safer, lower-burden care while automatically generating care records. By analyzing video and audio during care tasks, the system offers context-aware support that reduces caregiver stress and improves the quality of care.
Team: Shunta Asano, Kotaro Shobayashi, Amon Suzuki, Ayumu Matsui
GitHub: https://github.com/ayuuum/visual-care-coach
Video: https://drive.google.com/drive/folders/1Rwh4RWoy-RUi7x18saWJ8uvqmt6t1aqG?usp=drive_link#

ClearClaim (SSA-3373-BK)
Description: When people go blind from physical trauma, they have 10 days to prove it to Social Security or lose emergency benefits. 69% are denied on the first try — ClearClaim cross-checks their answers against legal frameworks in real time and prompts them to say what examiners need to hear.
Team: Van Nguyen
GitHub: https://civic-ai-beta.vercel.app/narrative
Video: https://www.veed.io/view/b668ed63-362a-4d57-a4d5-1edde98be26b?source=editor&panel=share#

CountFaith
Description: CountFaith is an AI-Native Agency platform that secures drug integrity by identifying high-risk supply chain anomalies and dispatching latent pharmacists for physical verification. It leverages real-time professional data to bridge the trust gap in the global pharmaceutical market and save lives.
Team: Yu Iwase
GitHub: https://github.com/Yu220284/CountFaith
Video: https://youtu.be/Uq2LtEgy2HM#

dairii
Description: dairii is the AI-native P&C insurance agency for Japan. We don't sell software to agencies, we run the work and make the agency model scale like software.
Team: Yuichi Sakano, Yutaro Kido, Yuma Katsuno, Joseph Ju, Koki Ueno
GitHub: https://github.com/dairii-ops/ai-native-insurance-agency-hackathon-202603
Video: https://www.loom.com/share/8153b390876945b790d2c65e00438915#

Datusuka
Description: Companies embed our SDK to let AI run scalable user interviews and user testing, collecting real feedback and improvement suggestions. With one click ("FixwithAI"), the AI automatically implements the proposed fixes, accelerating the product PDCA cycle.
Team: Takaya Okamoto, ShogoTakada
GitHub: https://github.com/takaya-okamoto/sf-tokyo
Video: https://drive.google.com/file/d/1DifC5hDrU-NMaTw2R08HJCRFxUyfCkl5/view#

ear-coach
Description: your on-demand personal trainer
Team: Javier Mendoza, Toshi Ichikawa
GitHub: https://github.com/pr3stigous/ear-coach
Video: https://www.youtube.com/watch?v=wHFyVru4QkU#

Echo-Cat
Description: A cat watches your screen and detects repetitive work patterns. Approve with one click, and it auto-generates an agent skill to handle it for you.
Team: Kota Shimizu, @r1cA18
GitHub: https://github.com/Olemi-llm-apprentice/echo-cat
Video: https://drive.google.com/drive/folders/1xegSIkARel15OqOu9R8hENC6NY6_i23M?usp=sharing#

Engram
Description: Networking Memory Copilot — An AI-powered memory assistant that turns conversation notes from meetings, events, and networking into a person-centric knowledge graph (Neo4j + Graphiti). It automatically extracts people, topics, interests, and follow-up actions from your notes, so you never forget who you met and what you discussed.
Team: Yuri Kurashima, Naoto Nishida
GitHub: https://github.com/rim0o8/c0mpiled-7
Video: https://youtube.com/shorts/vY24W-l8nhY#

FactoryOS
Description: Factory OS is an AI control tower for steel mill operations that brings together orders, stock, furnace and line status, schedules, and energy data into one decision layer. It helps mill managers replace fragmented manual coordination with faster, smarter production recommendations for urgent and day-to-day planning.
Team: Ganga, Samantha, Yoshihiro, Mongoo
GitHub: https://github.com/yoshi-bbb/factory-os
Video: https://drive.google.com/drive/folders/1JZ59nPrndbzc8sWHVwcjrfQqDX2k2FMi?usp=sharing#

GuideSight
Description: GuideSight is a real-time AI coach for physical work. A worker points a camera at the task, and GuideSight watches what they do and talks them through it live, like an expert standing beside them.
Team: Ishaq Ibrahim Khalifa, Altair Adilkhan
GitHub: https://github.com/Gentle-mann/guidesight
Video: https://drive.google.com/drive/folders/1zFMEBfu4M_yTIlg3UFLtOqioqm-BJK4c?usp=sharing#

Helper helpers
Description: We are building "Helper Helper," an AI coach for home caregiving.
Team: Subaru Kawami, Sumina Koiso, Katsunori Takahashi
GitHub: https://github.com/Newond/2026-03-08-hackathon
Video: https://drive.google.com/file/d/11P7bRWjH5xmEWhQNNU7QisPS7JKYJcI0/view?usp=sharing#

house deep search -- real estate ai agent
Description: my real estate agent was slow to reply, so we replaced him with ai
Team: Przemek, Mario, Abdel Fahmy
GitHub: https://github.com/mariomeissner/ai-broker
Video: https://drive.google.com/file/d/1rm0-lwpHLBGin--KEq06Ed0Jh8XxkKuj/view?usp=sharing#

Iterate
Description: Iterate is an AI product manager that automates the build-measure-learn loop by ingesting product analytics, error logs, and user feedback to generate actionable insights. It connects analysis directly to feature development and A/B testing, enabling teams to iterate faster with data-driven decisions.
Team: Takumi Hara, kazu, Daiki Gunji, Yudai Sasaki
GitHub: https://github.com/iterateapp/iterate
Video: https://drive.google.com/drive/folders/1Azsk2vjQu2ugvvWkR2dBiBEsGZ7xSs9Z?usp=sharing#

JamSaver
Description: Self-driving cars cause traffic deadlocks because each vehicle's AI prioritizes its own safety and can't coordinate with others. JamSaver is a cloud-based control tower that detects and resolves these jams in real time, sold to governments as public infrastructure.
Team: Daisuke Kuriyama, Natsuki Fukano, Steve Won
GitHub: https://github.com/ekusiadadus/jamsaver
Video: https://www.loom.com/share/25e22ea92062477c98ea1e5642957531#

Jun
Description: Decentralized Decision-Making in AI Agents
Team: Jun Kawai
GitHub: https://github.com/Jun0908/Meerkast
Video: https://youtu.be/y3EMbQIh9_Y#

Kangying Li
Description: Native AI Agency for investigating house
Team: Haruka Abe, Kangying Li
GitHub: https://github.com/hrkskov/2026-c0mpiled-hackathon
Video: https://drive.google.com/file/d/1U5uELCnxSzBb0Mmojpg1BHbVNdluoos8/view?usp=drive_link#

Laplace Lattice
Description: LaplaceLattice is an AI-native hedge fund platform that captures early market alpha by analyzing local-language news sources hours before they reach global English-speaking markets. It utilizes a fine-tuned AI model to extract sentiment from these native reports and dynamically adjusts Monte Carlo simulations to generate superior commodity price predictions.
Team: Yuki Murakami, Taiki Nakamura, Rafael Mori, Hammon Dutra
GitHub: https://github.com/0xtkey256/laplace-lattice
Video: https://www.loom.com/share/eb684aed4f7c4e0596eec3d7122482df#

Mercy
Description: Mercy is secure AI for whistleblowers, journalists, and lawyers handling highly sensitive evidence. It analyzes, summarizes, and redacts documents inside hardware-protected confidential computing environments, so even Mercy cannot access the raw data.
Team: Brilliant Aksan, Yoshi Sato, Diego Limo
GitHub: https://github.com/brilliantaksan/mercy
Video: https://share.icloud.com/photos/0218vcv1xGJBu5zTmyFiG1Vcw#

NextBridge
Description: Protecting Medical Data Sovereignty with ZK
Team: AK
GitHub: https://github.com/NextMed-main/Main
Video: https://main-virid-omega.vercel.app/#

Noob
Description: Noob, the embedded AI designer that thinks and acts like your Head of Design would, directly inside Figma, Slack, Linear, ...
Team: Giacomo Bottoli, Lilit Grigorian
GitHub: https://github.com/gbottoli/design-modifier-plugin.git
Video: https://www.loom.com/share/3f03bb43ba84462b8307e7d20094e6c8#

Nozue
Description: PDM with policy
Team: kaoru.nozue
GitHub: https://github.com/Nozium/c0mpile
Video: https://drive.google.com/file/d/1XafTxdBjv34vETXfs_4WRHoBdFXFjx1c/view?usp=sharing#

Nullifier
Description: Nullifier is a zero-knowledge proof strategy marketplace where anyone's AI agent can monetize trading edge without revealing it.
Team: Guoliang Chen, Yuanfan Wang, Asuma Yamada
GitHub: https://github.com/yuanfanwang/ai-fund-platform.git
Video: https://drive.google.com/file/d/1P8f2u7dzLS5A7UI6D-fZ6pKCWqxd8BlY/view?usp=sharing#

onepercent
Description: one percent is an intelligence layer for prediction markets to identify short term market biases to make repeated +1% returns
Team: ibrahim yildiz
GitHub: https://github.com/ibrhmyi/one_percent
Video: https://www.loom.com/share/4a6d12513f7f4e77ab90523c95b025ba#

Origin
Description: Gemini said Origin is the first AI-native hedge fund that solves the multi-billion dollar "Synthesis Gap" by replacing the 72-hour human research loop with a Darwinian swarm of autonomous agents. While traditional fund managers are bottlenecked by manual 10-K analysis and compliance-heavy workflows, Origin collapses institutional-grade conviction into a 12-second autonomous execution cycle.
Team: Natalia Mykhaylova, Takuya Tominaga, Phat Nguyen, Azamat Erkinov
GitHub: https://github.com/kikman-git/origin
Video: https://www.youtube.com/watch?v=pV6Tua9APh8#

PM Studio
Description: Product decisions, co-created with agents.
Team: Queenie Wu, Tomoyasu Hirano, Tsubasa Kawakami, Gaku Noguchi
GitHub: https://github.com/Telepotch/yc-pm-studio
Video: https://drive.google.com/drive/folders/1cZ1cb-zJAQxkd1qgDRQfhzoy-8PzCJyj?usp=sharing#

PrimitiveCredit
Description: Credit & Enforcement using MRTD
Team: Sam Lo
GitHub: https://www.canva.com/design/DAHDV5isHhY/ShPfXYant2dKJbk3TYf2Ng/view
Video: https://www.canva.com/design/DAHDV5isHhY/ShPfXYant2dKJbk3TYf2Ng/view#

Project Alice
Description: This product is an AR-based worker monitoring and instruction system that uses AR glasses to capture images of a worker's viewpoint and analyze them with AI to understand the current task. The analyzed results are shown to a project manager in real time, who can monitor progress and send instructions back to the worker through the AR display.
Team: Shinnosuke Nakano, Kotaro Kurita, Ryunosuke Kurita, Yuu Suzuki
GitHub: https://github.com/HINOKAGUZUTI/GembaEye.git
Video: https://drive.google.com/drive/u/0/folders/1d5fEVwWIkbIqtE6U60qZuwR_Ty_vlsTz#

PrompTrip
Description: Problem: Planning and managing travel is fragmented across dozens of apps, emails, and calls — leaving travelers overwhelmed and on their own when things go wrong. Solution: An AI travel assistant that takes your plan and constraints once, then handles every booking, change, cancellation, and question for your entire journey — end to end.
Team: Tuba Ali, Yuki, Yoshi, Jack
GitHub: https://github.com/ReadingNought/20260308_Hackaton_demo1
Video: https://drive.google.com/file/d/1pzkOrmsR1azcKCrcG_OlYGuBX4sTEdIW/view?usp=sharing#

PROZEN
Description: The AI-native OS for product bets. You didn't start this to write specs. You started this to make the right bet.
Team: Toru Tano
GitHub: https://github.com/Torutesu/PROZEN
Video: https://drive.google.com/file/d/18k0YYZQUGQ2ewHUEoxWvddZamIiKMjQz/view?usp=sharing#

Primitive Credit
Description: ID tagged
Team: Sam Lo
GitHub: https://github.com/sam16777213/idcheck
Video: https://www.youtube.com/shorts/-HEpyOqpEhY?si=gOXIa9oLhi9TbssD#

RepAIr
Description: Repair is an AI-powered platform that diagnoses vehicle problems from user input, images, and sound, then generates step-by-step repair guidance through AR smart glasses. It also recommends and delivers the necessary parts, enabling users to perform professional-level repairs themselves while building a data-driven marketplace for automotive maintenance.
Team: Kotao Ueda, 上田光太朗, 石橋亨祐, 植竹空 Sora Uetake, 藍 光祐
GitHub: https://github.com/kosuke-eth/RepAIr
Video: https://www.youtube.com/watch?v=V37NJuf2Po8#

Rojin
Description: Social OS – match making platform
Team: Shuji, Ivan Abundus, Nivya, Shoichi Aizawa
GitHub: https://github.com/shinji-Yama77/rojinmatch
Video: https://drive.google.com/drive/folders/18CmWC_Val3lWLWqvVGOHRwrNH1nP_7dF?usp=sharing#

Sekai Kanamori
Description: A portfolio intelligence app that aims to bridge the gap between retail and institutional investors, incorporating tailored AI insights, quantitative metric calculations, and statistical prediction.
Team: Sekai Kanamori
GitHub: https://github.com/SekaiKana/c0mpiled-hackathon
Video: https://drive.google.com/file/d/15BIzh5Zn3FoXSDxgiBdF_nZR3A-ujuxa/view?usp=sharing#

SenjinUT
Description: PostAd.ai is an AI-driven platform that transforms product URLs into high-converting UGC video ads by automatically analyzing market data and competitor trends to generate strategic, ready-to-use creative assets. It eliminates the friction of manual influencer management, enabling agencies to scale with a high-margin, software-powered production model.
Team: Akihiko Shimoyama, Sayaka Yamashita, Shuji Kato, Takemichi Seki
GitHub: https://github.com/Takemichi0/PostAd.ai/tree/main
Video: https://drive.google.com/drive/folders/10fUz-6tJeMtdxTBdSV169vqcV18uyi_y?usp=sharing#

Ship Skip
Description: ShipSkip is a Feature QA platform where AI users, trained on your real customer data, evaluate new features before release. It helps you stop noisy updates and ship only what truly matters.
Team: Kiichi Watanabe
GitHub: https://github.com/qiichos/ship-skip
Video: https://www.loom.com/share/0f1cc36a6e0d418f954639b6751c8e94#

Silver Assist
Description: An AI-powered voice assistant that guides elderly users through everyday computer tasks — like banking and video calls — with real-time screen guidance, so their families never have to worry again.
Team: kmander, otakeshi63, ryankaminsky, shimpei.kawakami68, kohteddyhayashida
GitHub: https://github.com/Teddy0527/yc-hackathon
Video: https://www.loom.com/share/f1d1cbbcfa414087b09b702e7994519e#

SUSHI SAMOSA
Description: Turn business cards into real-time, evidence-backed contact briefs with a live research pipeline and compatibility scoring. See who to follow up with, why they matter, and what to do next — all in one decision-ready interface.
Team: Hiroyuki Matsushima, Animesh Harsh
GitHub: https://github.com/Aisometry/1day_-hackathon
Video: https://drive.google.com/file/d/1hlmQMvab4yUOGW0UItA2RWNGAAHOnGRq/view?usp=drivesdk#

SuzuPay
Description: SuzuPay is a next-generation QR payment service that allows travelers to spend stablecoins like USDC and USDT at local merchants with a low 1% conversion fee. We utilize a proprietary AI engine to automate merchant credit scoring, enabling shops to bypass weeks of bank bureaucracy and start accepting digital assets in minutes.
Team: Shivam Singh Chauhan, Sourabh Kapure, Shubham Kumar, Kosuke
GitHub: https://github.com/SuzuPay/SuzuPay-XRPL
Video: https://www.youtube.com/watch?v=ehgKsY6culM#

SynapseAI
Description: Pain: Current AI tools create a "wall of text" that causes cognitive overload and provides generic, one-dimensional answers. Solution: We replace dialogue with a "Living Mind Map" where five expert personas engage in a visual, spatial debate to spark strategic thinking.
Team: Notef Kawamoto, Satoshi Suzuki
GitHub: https://github.com/satoshi-szk/c0mpiled
Video: https://drive.google.com/file/d/1ArI38xyzgDIKc0N_zzBwmljXwRxvYyLS/view?usp=sharing#

Takeshi Kanamori
Description: war impact quantify analyzer
Team: Takeshi Kanamori
GitHub: https://github.com/TakeshiKanamori20/post-conflict-analyzer
Video: https://youtu.be/7c7YHQzVmOw#

tarot1234
Description: QB Auto-Agent is an AI-powered full-lifecycle vehicle diagnostic system that monitors cars from the production line to daily driving, using edge AI to detect issues, diagnose root causes, and estimate repair costs in real time. By automating inspections for manufacturers and providing drivers with a 24/7 intelligent "car antivirus," it reduces factory inspection costs while enabling proactive vehicle maintenance with full data privacy.
Team: jiashankmg
GitHub: https://github.com/zzzgjj49/QB-Auto-Agent
Video: https://github.com/zzzgjj49/QB-Auto-Agent#

Team Cognisor
Description: Aura is an AI-powered CFO assistant for startups that helps founders understand their finances and make smarter growth decisions. It connects revenue, expenses, and financial data into one intelligent system, analyzes burn rate and runway, and simulates scenarios like hiring or increasing marketing spend. By turning complex financial data into clear insights, Aura helps startups increase revenue, manage cash flow, and make confident strategic decisions.
Team: Farhan Morshed, Irfan Faisal
GitHub: https://github.com/Morshedwork/cfo
Video: https://youtu.be/_lyVool1sfM?si=QD7BdB7Ywv0g6kb5#

TeamAligner
Description: AI-powered alignment engine that turns cross-functional chaos into decisions — cutting Time to Product Outcome drastically.
Team: Charles Shimokura, Eitaro Yamatsuta
GitHub: https://github.com/EitaroY/team-aligner
Video: https://drive.google.com/file/d/1LOV52CpQ3-HkHy-R3PZhsYdXJ18BfUVD/view?usp=sharing#

Tetra Polar Bear
Description: AI that turns product truth into culturally resonant taglines. It aligns brand claims with customer reviews and adapts the message for different markets.
Team: Toshi Ichikawa
GitHub: https://github.com/toshi-tetra/tetra-x-cultural-conversion-engine
Video: https://www.dropbox.com/scl/fi/ufw80c48dzerz4w0t0jt4/Tetra-Polar-Bear.mp4?rlkey=qk8seykpm0vbkq56h6ua1st6n&st=vr7472e7&dl=0#

Tomoki
Description: We use AI to generate blog content that helps the right users discover new products. Our platform turns blog content into a scalable engine for product discovery and customer acquisition.
Team: Hirotsune, Tomoki Tanzawa, Yujun Mori
GitHub: https://github.com/tontiruu/signal-boost
Video: https://drive.google.com/file/d/1NtoNDGaZfqE61jfKj5I6M6xIT61ZSPf_/view?usp=drive_link#

Treasury Gateway
Description: We build
Team: Kazuki, Arthur, Daito Yamagishi
GitHub: https://sepolia.etherscan.io/address/0xe69606565988e8a2bc9048ec22645670164fab96
Video: https://drive.google.com/drive/folders/1BmAHwJ1ELUyVdtOibjjtn4gJS01kFi4h#

TrustBridge
Description: Returning medical data from hospital silos to the patient's hands
Team: Kosuke
GitHub: https://github.com/kosuke-eth/TrustBridgeDEMO
Video: https://photos.app.goo.gl/vPADCUZfovZPAP1u8#

Unvibe
Description: Unvibe is an AI-native agency that turns broken vibe-coded apps into production-ready software, automatically detecting and fixing the security vulnerabilities, exposed secrets, and code quality issues that AI coding tools leave behind. Static analysis finds the problems, AI fixes them, and the user gets back secure, shippable code in seconds.
Team: Batyrkhan Kanash
GitHub: https://github.com/batyrkhan9/unvibe
Video: https://www.loom.com/share/781cf998e0bb423195e909095ba8bb99#

Userinterview2
Description: Validate Your Ideas Faster with AI-Driven User Interviews. Say goodbye to manual interview scheduling and tedious transcript analysis. Our platform lets you define a learning goal, spin up an interactive AI avatar to interview your users, and instantly generates a comprehensive report on whether your product hypothesis is validated. Perfect for product managers and UX researchers who want to scale their qualitative research.
Team: Alex, Sasha, Koki
GitHub: https://github.com/alexweberk/userinterview2
Video: https://drive.google.com/file/d/1eV3QGhb0IKMwoDmhdoqMdSpRW31QRt05/view?usp=sharing#

VIVESAFE
Description: Our project is VIBE SAFE, an AI agent coding sandbox service. We believe AI coding can push humanity forward.
Team: Ryu, Kota, Kohi
GitHub: https://github.com/Keyhole-Koro/safeVibeBuilder
Video: https://youtu.be/yJ_dwi4Gdqc#

YAOROZU
Description: Just as YAOROZU gods are believed to dwell in all things, AI agents will keep a watchful eye on every site, issuing real-time work instructions (quests). Workers can accept the nearest quest from a map on their smartphone and complete on-site operations by following oracle-like voice guidance and AR.
Team: Koji Itano, Haruto Matanabe, Shoto Morisaki, Sota Kobayashi, Keita Yasui
GitHub: https://github.com/koji-itano/yc2026
Video: https://drive.google.com/file/d/1hKJSPRuCPoEWMqaHy8MkhwacZEStFCqc/view?usp=sharing#

YS & KO
Description: Medical data structuring support and clinical trial patient search
Team: kento okawa, shogo yoshizaki
GitHub: https://github.com/swkima-dev/ai-medical-record-agent
Video: https://youtu.be/0PbkvOJnVrw#

Zac in Amon
Description: AI agent-native payment system taking advantage of Stable coins.
Team: Haruto Miyazaki
GitHub: https://github.com/Zac-eng/agi-wallet
Video: https://drive.google.com/drive/u/0/home#

Zero to one
Description: AI native design agency for Building information model
Team: Sho
GitHub: https://github.com/shousami7/ycrealty2#

All Photos
















* * *