| |
|
| |
Learning was never text-only. Kids read out loud, scribble on worksheets, point at diagrams, & think with their hands. Now AI can hear, see, & watch too.
SVAI Arrival Guide .pdf
Multi-Modal AI for Education
A text box only catches what a student types. A great teacher notices much more: the pause before a hard word, the crossed-out step in a math problem, the moment a class goes quiet. Multi-modal AI can finally pick up those signals. Voice, handwriting, images, & video can all become part of how we teach.
The hard part isn't the demo. It's building tools that teachers & families actually trust, & showing that students really learned more. That takes clear learning outcomes, not just a clever model.
Join Beta Fund & co-host Epic at Snowflake's Silicon Valley AI Hub in Menlo Park for a full day of building multi-modal learning tools. The day includes a panel on where AI education is headed & demos in front of the room.
Build tutors that listen. Build tools that see the work. Build learning you can measure.
Approval required. About 120 builders. You need an approved registration to get in: no walk-ins, no +1s. Bring your laptop.
Join our Discord for credits, team formation, & support:
Join Discord Server
2-4 builders per team
Format
One-day, build-first hackathon in Menlo Park
2-4 builders per team
Every project uses at least two modalities: text plus voice/audio, images/handwriting, or video
Working demo required (3-minute demos)
Use public, synthetic, or consented data only. No real student records.
Hosts & Co-host
Beta FundSilicon Valley fund backing AI founders from inception to seed. Invests $200K-$3M & is built on the Beta community of 500+ startups & 300,000+ founders.
Beta UniversityRuns an 8-week pre-acceleration program that helps early-stage founders build VC-ready companies with proven know-how from Silicon Valley. Apply now!
EpicCo-host. The leading digital library for kids 12 & under, with 40,000+ books, audiobooks, & learning videos from 250+ publishers, used by teachers & families. Epic also co-hosted our AI + Education Hackathon at Stanford in September.
Venue Sponsor
Snowflake Silicon Valley AI HubOur venue for the day, Snowflake's AI hub in Menlo Park. Thank you to Snowflake for hosting us.
Agenda
9:00 AM Check-in & Breakfast
10:00 - 10:20 AM Opening remarks & kickoff
10:20 - 11:00 AM Panel: What Multi-Modal AI Changes in Learning
11:00 AM Build
12:00 PM Lunch
1:00 PM Build
4:00 PM Hard submission deadline
4:00 - 5:00 PM Demos (3 min limit)
5:00 - 5:30 PM Audience voting & award ceremony
Focus Tracks
Track 1 Voice & Audio Tutors
Tutors that listen as well as talk. Hear a student read or think out loud, catch exactly where they get stuck, & respond like a patient coach.
Ideas: a read-aloud fluency coach that hears which words a kid stumbles on a think-aloud math tutor that spots the misconception in a spoken explanation a speaking partner for language learners voice-first learning for students with dyslexia or low vision
Track 2 Vision, Handwriting & Documents
Learning happens on paper. Build AI that reads handwritten work, worksheets, diagrams, & textbook pages, & gives feedback on the steps, not just the final answer.
Ideas: snap a photo of handwritten math & get step-by-step feedback grade a stack of scanned worksheets & group the common mistakes turn a picture-book or textbook page into an interactive lesson understand a student's science diagram or lab sketch
Track 3 Video & Classroom Understanding
Video holds a lot of what happens in teaching & learning. Turn lectures, demonstrations, & classroom moments into feedback for students & teachers.
Ideas: turn a lecture video into chapters, a study guide, & a quiz give teachers feedback on talk time & questioning from a classroom recording watch a student's experiment or presentation & coach them on it a video explainer that adapts to what the learner already knows
Track 4 Real-Time Multi-Modal Tutors & Agents
Tutors that work live, like sitting next to a great teacher. Combine live video, voice, & screen or camera input so the AI sees what the student is doing & helps in the moment.
Ideas: a tutor on a live video call that watches a student work a problem & nudges at the right time a screen-aware coding or homework coach a phone-camera lab assistant for hands-on science an AR or camera guide that teaches by pointing at real objects
Track 5 Multi-Modal Storytelling
Stories are how kids learn to read, imagine, & remember. Build AI that blends voice, images, & video into stories that teach, react to the reader, & let kids create their own.
Ideas: turn a child's drawing & voice recording into an illustrated, narrated story an interactive read-along that reacts to how the reader is doing turn a history or science lesson into a narrated illustrated or video story kids co-creating a story by voice, with images generated as they talk multilingual narration that reads the same story in a child's home language
Track 6 Multi-Modal Learning Assessment
Build quick, fair, automated ways to understand where a learner is, using voice, handwriting, & interaction instead of long one-on-one tests.
Ideas: a reading-fluency screener that listens to a child read aloud early dyslexia signals from speech & spelling a math-fluency check from handwritten work teacher-ready reports that turn results into next steps
Wildcard (Open Category)
Any multi-modal idea that doesn't fit the tracks above. If it solves a real problem in learning & shows a measurable learning outcome, build it.
Approved registration required. No walk-ins. No +1s. Address shown to approved guests.
Hosted by Beta Fund Co-hosted with Epic Venue sponsor: Snowflake Silicon Valley AI Hub Beta University Events
|
|
|
|
|
|
|
|
