Human behavior simulation for live events.

Attractor Labs combines traditional forecasting with generative agent simulations. We predict audience behavior across hundreds of scenarios and hand you the version of the show that books out.

See how it works→Get in touch

Scroll

The product

Demand intelligence for live entertainment.

Operators make multi-million-dollar pricing, sizing, and inventory calls on traditional forecasting and gut feeling. We replace that with data-driven intuition, trained on regional fan interviews, behavioral research, and historical ticket data.

Traditional forecasting, modernized

Quantile gradient-boosted models cross-checked by foundation time-series models. Standard forecasting where it's already solved, calibrated against real ticket outcomes run-on-run.

Simulated personas, real behavior

Each persona is calibrated against real fan interviews and historical purchases. Memory, reflection, planning. They walk your show, evaluate the offer, and decide whether to buy.

A focus group on demand

Stop the simulation, talk to any persona directly. Ask why they bought, why they didn't, what would change their mind. A live qualitative layer most operators never get.

The output

Recommendations you can act on.

Not a number. Not a forecast band. A concrete plan with confidence ranges, downside floor, and the exact trigger conditions that say act now.

LiveRun #00417 · 2026-05-04 14:32 PET

Confidence 87.4% · +12.6 vs gut

Recommendation

Open one date now. Reserve a second with 72% sellout probability if presale exceeds 8,000 in 36 hours. Expected $2.1M revenue.

Expected revenue

$2.1M±0.18

Downside floor

$1.62M

Sellout p

72%

Cumulative sales forecast

tickets sold

median (P50)P10 – P90 band

How it works

Every scenario, simulated. One you can act on.

Every show has thousands of possible versions: dates, prices, capacities, marketing spend, announcement timing. We simulate them all so you see which ones win, which ones break, and which one to actually book.

01Inputs

Every signal a show emits, ingested at once.

Streaming, social, presale velocity, fan interviews, historical sell-through, candidate dates, price tiers, marketing spend, capacity. Dozens of fragmented sources resolved into a single state the simulation can act on.

02Simulation

Thousands of versions of your show, simulated in parallel.

10,000+ generative-agent personas, each calibrated to a real audience segment, walk through every scenario you could choose. The full decision surface across capacity, price, timing, and spend.

03Recommendation

One plan you can act on, with the trigger that says when.

Expected revenue, downside floor, sellout probability, and a written recommendation in operator language. Not a forecast band. A decision, with the condition that flips it.

Industry

Better decisions grow the whole industry.

Today, demand is guessed. Rooms sit half-empty, tours skip cities, presales misprice. When operators can see what audiences would actually buy, more shows happen, more seats fill, and more fans get to be in the room. Venues, promoters, and platforms each win the same way.

Venues

Open the right number of dates. Right-size the room. Justify the booking decision with a confidence range, not a hunch. Stop leaving capacity on the floor.

Working with

Promoters

Price tour stops with confidence. Time the announcement. Allocate marketing spend per market with a precise demand picture per route, not regional averages.

Working with

Ticketing platforms

Embedded simulation as a value-add for issuer events. White-labeled demand forecasts that turn your data into a decision layer for the venues and promoters on your platform.

The team

Meet the Live Simulators.

We come from Stanford HCI and the MIT Media Lab, with backgrounds in digital twins, behavioral psychology, and product design, and research and field work on simulating populations of agents.

Mateo Larrea

Co-Founder & CEO

Juan Francisco Misle

Co-Founder & COO

Anushhka Thakur

Co-Founder & Head of Design

Huanxing Chen

Co-Founder & Head of Research

From