RECORDED · TALK 01
Adam Chan
BattleBots Hack Night
Watch the talk→TALK 01 / 01
HACK NIGHTCompleted01 Talks
Battle Bots Hack Night & Watch Party (w/ OpenAI Codex)
// Featured Talks [1]
Builders on what they are actually shipping.
Recorded talks, technical context, and the workflow behind real product work.
RECORDED · TALK 01// Sponsors.log
Companies backing the room.
[DIAMOND] 1
[GOLD] 1
// LiveDemos.log [5]
Recorded work, ready to replay.
Watch the live demo presentations from this event.
Havok
DEMOEDHavok is an open-source intelligence tool to do bot matchups using OSINT data - Reddit AMAs, public site info, bot specs, and so on.
HackerSquadScout - an MCP server that turns public fight records into AI-queryable scouting intelligence, predicting the defect before the win.
DEMOED
Combat robots rarely lose because they're destroyed — they lose because they break. Batteries die mid-fight, weapons jam, bots get counted out. Those failure patterns are sitting in public fight archives, unmined. Scout mines them: Bright Data scrapes the records, we distill each bot into a dossier — weapon class, KO rate, documented self-failure incidents, and the exploit that follows — and serve it through an MCP server any AI agent can plug into. Ask Claude "what's Tombstone's weakness?" and it answers like a veteran pit crew: chronic recoil self-damage, three championship runs ended by his own systems — survive the first minute and he beats himself.
HackerSquadBotView AI
DEMOEDBotView AI transforms third-person combat-robot footage into an approximate synthetic first-person view from the selected robot. Users upload a short match clip, identify the robot they want to follow, and Runway Aleph 2 generates a reconstructed onboard-style perspective. The app presents the original and synthetic videos side by side with synchronized playback, seeking, speed controls, drift correction, and a tactical event timeline. BotView helps builders review opponent distance, movement direction, wall positioning, and collision timing from a perspective that normally requires an onboard camera. The bundled demo works without credentials. The hosted workflow securely uploads MP4 video through an Express server, keeps the Runway API key out of the browser, and clearly labels generated footage as a synthetic reconstruction rather than authentic onboard video.
HackerSquadARENA INTELLIGENCE
DEMOEDRead data analysis on what strategies are emerging for the 2026 Battle Bots Tournament, and explore counter strategies for what others are likely to bring.
BattleBets
DEMOED
N
This app uses computer vision to identify BattleBots competitors from any video stream, leveraging Bright Data to scrape historical stats, rankings, and fan sentiment to instantly calculate realistic pre-match win probabilities.
HackerSquad// Projects.log [10]
What builders shipped.
Check out the projects built during this event.
Fight IQ Copilot
Fight IQ Copilot helps BattleBots teams understand a fight while it is happening. Upload a fight replay and choose the two robots. The app watches the video in short chunks, identifies moments such as weapon contact, aggression, control, mobility loss, and danger, then shows who has the advantage and recommends what to do next. Bright Data gathers public information about each robot, including its weapon type, fight history, and known weaknesses. Vidarax analyzes the video as it plays, while Gemini turns the evidence into simple tactical advice. Vidarx is a real-time video intelligence engine build by me which will be open sourced soon, it runs both local AI on device including mobile as well as LLM powered backends. Unlike normal video analysis, results appear live instead of making the user wait for the entire video. Live demo: https://battlebot.cosminbararu.com
Scout
MCP server for battlebots w/ codex. Data scrapped using Bright data.
HackerSquadScout - an MCP server that turns public fight records into AI-queryable scouting intelligence, predicting the defect before the win.
Combat robots rarely lose because they're destroyed — they lose because they break. Batteries die mid-fight, weapons jam, bots get counted out. Those failure patterns are sitting in public fight archives, unmined. Scout mines them: Bright Data scrapes the records, we distill each bot into a dossier — weapon class, KO rate, documented self-failure incidents, and the exploit that follows — and serve it through an MCP server any AI agent can plug into. Ask Claude "what's Tombstone's weakness?" and it answers like a veteran pit crew: chronic recoil self-damage, three championship runs ended by his own systems — survive the first minute and he beats himself.
HackerSquadBotView AI
BotView AI transforms third-person combat-robot footage into an approximate synthetic first-person view from the selected robot. Users upload a short match clip, identify the robot they want to follow, and Runway Aleph 2 generates a reconstructed onboard-style perspective. The app presents the original and synthetic videos side by side with synchronized playback, seeking, speed controls, drift correction, and a tactical event timeline. BotView helps builders review opponent distance, movement direction, wall positioning, and collision timing from a perspective that normally requires an onboard camera. The bundled demo works without credentials. The hosted workflow securely uploads MP4 video through an Express server, keeps the Runway API key out of the browser, and clearly labels generated footage as a synthetic reconstruction rather than authentic onboard video.
HackerSquad
// Prizes.log [1]
Recognized work.
Best demo for BattleBots Project
WeAreDevelopers conference tickets (3max)You don't need to demo live to be considered for the best project demo. All you need to do is submit your video with your project on HackerSquad. Talk about how it uses Bright Data to do search and talk about what you built with respect to Battle Bots.
Join Our Next Event
Don't miss out on future events. Sign up to stay updated on upcoming hackathons and meetups.