TicketTrade AI

AI agents represent buyers and sellers, negotiating ticket deals automatically. Faster, fairer, and more dynamic than StubHub/TicketSwap — with real-time matching and transparent pricing.

Comment

Inspiration

The global secondary ticket market was valued at USD 2.85B in 2023 and is projected to reach USD 6.56B by 2032 (CAGR 9.7%). Yet, the major platforms — SeatGeek, TicketSwap, and StubHub — still operate on static listing models, where sellers post tickets and buyers manually search through pages of options.

This model breaks down when:

  • Buyers have group-size constraints
  • Sellers want partial liquidation of inventory
  • Buyers need adjacent seats or flexible bundles
  • Dynamic, real-time pricing matters
  • Negotiation is required but not supported

We were inspired to rethink ticket resale by replacing listings with preferences, constraints, and strategies, and allowing AI agents to negotiate optimal matches automatically. Instead of transaction friction, we wanted a system where both sides simply express what they want — and the agents do the work.

What it does

TicketTrade AI is an agent-driven ticket marketplace. Instead of browsing static listings, buyers and sellers submit:

  • price ranges
  • quantities
  • sensitivities
  • group constraints
  • willingness to split or bundle

Agents then perform real-time negotiation, discovering prices and creating matches automatically. The system supports:

  • dynamic price negotiation
  • partial liquidation for sellers
  • bundle enforcement for buyers
  • constraint-aware matching
  • transaction clearing across multiple agents

The result is a smart, fair, automated ticket-trading experience that eliminates the pain points of traditional platforms.

How we built it

We began by mapping the space:

  • defining market constraints
  • writing down edge cases
  • listing assumptions and negotiation rules

Next, we designed the agent-to-agent negotiation protocol, addressing questions like:

  • How do BuyerAgent and SellerAgent alternate turns?
  • How many sellers does a buyer negotiate with?
  • Which preferences are revealed, and when?
  • When do we finalize a transaction?
  • How do we resolve multiple matches simultaneously?

We then implemented the system in parallel:

  • Negotiation core logic — feasible intervals, agent prompts, turn-based negotiation, clearing
  • Frontend UI — real-time feedback, clean visualization of offers, and results

This parallel approach let us iterate quickly without losing conceptual coherence.

Challenges we ran into

Building autonomous negotiation was harder than expected:

  • Making LLM calls predictable and non-hallucinatory
  • Designing prompt structures that consistently reflect constraints
  • Handling edge cases like conflicting matches
  • Mapping seller and buyer preference models to meaningful negotiation behavior
  • Ensuring agents didn’t break rules or exceed allowed intervals

Additionally, integrating the UI and backend negotiation engine required careful coordination.

Accomplishments that we're proud of

  • Designing a fully dynamic negotiation protocol that replaces static listings
  • Building agents that follow constraints and negotiate reliably
  • Designing a system fundamentally different from StubHub, SeatGeek, or TicketSwap
  • Building a working prototype under hackathon time pressure

What we learned

We gained hands-on experience with:

  • Agent-driven market mechanism design
  • Structuring buyer–seller constraints into solvable negotiation flows
  • Using LLMs in a deterministic, controlled environment
  • Understanding the limitations and strengths of LLM agents in real economic interactions
  • Designing clearing models for multi-party negotiations

Most importantly, we learned how to blend AI reasoning with traditional marketplace logic to create new forms of automated commerce.

What's next for TicketTrade AI

Several exciting next steps:

  • Implement secure ticket escrow and real-time verification
  • Add multi-round negotiation with market heat modeling
  • Support for multi-agent auctions and batch clearing
  • Expand to venue APIs for direct inventory access
  • Allow event organizers and artists to use our negotiation engine directly
  • Deploy a public API so other platforms can use our agentic negotiation protocol

Ultimately, we see TicketTrade AI evolving into the first agent-native ticket exchange protocol, powering automated, fair, and transparent ticket trading at scale.

Built With

Share this project:

Updates