Multi-agent platform for live-event coordination — discovery, scheduling, on-site logistics in real time.
Summary
Aletheia is an event tech engagement running 2025–present. Multi-agent platform for live-event coordination — discovery, scheduling, on-site logistics in real time. The engagement is ongoing; multi-agent orchestration · real-time decisioning is the current headline result.
Facts
Client
Aletheia
Sector
Event Tech
Dates
2025–Present
Key result
Multi-agent orchestration · real-time decisioning
Site
aletheia.events
Engagement
Discovery sprint → build alongside → hand-off
The problem
Live events are coordination-dense, latency-sensitive, and multi-stakeholder. Static rules-based scheduling cannot adapt to real-time signal. Aletheia replaces the rule engine with a network of autonomous agents that negotiate decisions in milliseconds.
The architecture
Multi-agent runtime with explicit role boundaries (attendee · organiser · venue). LangGraph orchestration, streaming inference, evals on every agent action.
01Agent layerattendee, organiser, and venue agents with explicit role boundaries, negotiating decisions rather than following a rule tree.
02OrchestrationLangGraph coordinates the handoffs; every transition is traced and replayable.
03Inferencestreaming responses inside a real-time budget, so coordination lands within the window a live floor allows.
04Evalsevery agent action scored in CI and again on production traffic.
Aletheia architecture, top to bottom — agent layer, orchestration, inference, evals.
The results
The engagement is ongoing. The numbers below are the results measured so far.
Agent response, p95
< 90 msDown from roughly 800 ms on the rules-based baseline it replaced
Agent boundaries
3 rolesAttendee, organiser, venue — each with its own tool surface
Covered by evals
Every actionScored in CI and on live traffic, not sampled after the fact
“Rules-based scheduling could not absorb a floor that changes every minute. Agents that negotiate inside a 90 ms budget can.”