Weekly meetups, ship sprints, demo nights.
Personify by TAG x Anthropic x Tulip
The Personify Challenge
AI got smart before it got good at talking.
Personify is a one-day build challenge in Amsterdam for 30 selected participants designing AI that feels more natural to interact with.
Today's AI can reason, summarize, and generate, but it often misses the human layer: tone, timing, memory, emotion, context, subtext, and knowing when to push back instead of agree.
As AI moves from prompt boxes into voice interfaces, ambient agents, healthcare intake, tutoring, customer support, sales, and financial advice, these interaction defaults start to matter.
The challenge
Every AI ships with the same personality: helpful, agreeable, bland. Trained to find the most logical answer, not to feel like anyone. Your job: build an experience where AI isn't bland.
That can be anything. An infuriating character. A strange, memorable experience. An AI that actually tries to get to know you and builds a relationship over time. Something weird. This is a creative exploration: how far can you get from the default assistant personality toward something with real character?
Angles to steal from:
Voice & tone: adaptive personality that shifts with context
Conversational memory: AI that remembers what matters and forgets what doesn't
Emotional calibration: knowing when to be concise vs. elaborate, when to push back vs. agree
Multi-turn coherence: maintaining identity, vibe, and intent across long conversations
Cultural nuance: AI that doesn't just translate but communicates
Teams & deliverables
Teams of 3–5. Form your team in advance; anyone without one gets matched on the day.
Every team ships three things:
A working demo
A 1 minute video
The repo
The day
09:30 — Kickoff and briefing
10:00 — Build begins
17:00 — Hard deadline
17:30 — Showcase and winners announced
Judging
Creativity: how far did you get from the default assistant?
Feasibility: is this something that could keep working and keep going on multiple interaction?
Code quality: is the repo built well?
Prize
€2,000 in credits for the winning team.
Who should apply
AI/ML engineers and NLP researchers, designers and conversational UX specialists, graduate students and PhDs (UvA and beyond), and startup founders building in the interaction layer.
Partners
Anthropic · ElevenLabs · Tulip · University of Amsterdam · TAG
Why now
The next wave of AI adoption depends on trust, and trust depends on interaction quality. The technology is ready. The experience layer is not.
Weekly meetups, ship sprints, demo nights.