Aurelioaurelio.ai
Learn
github.com/aurelio-labs
OSSsemantic-router 0.1 · graphai 0.0.7 · chunkers 0.0.10

The semantic mesh for real-world AI.

Aurelio is an applied AI lab. We build the open-source mesh production agents are routed, retrieved and orchestrated on — Semantic Router, Semantic Chunkers, and GraphAI.

semantic-router · playgroundlive demo · 0 LLM calls
USER TURN
""
encode · cohere · 1024didle
ROUTER DECISION
route:travel
confidence0.94latency0.07ms · 0 llm callsactionflights.searchguard✓ passed
// routing production AI at
MCGRAW HILLRED HATPINECONEINTERNET ARCHIVEUK GOVERNMENTOPEN BRAIN INSTITUTEOPEASUPERAGENTUNIVERSITY OF SÃO PAULOi.AI INCUBATORWAVEMAKERBOLNA AIRADIANT LOGICLMSYSTEMSBRAIN STATION 23BIG JOBSAE
The mesh

Three libraries.
One semantic mesh.

R
DECIDE
semantic-router
sub-ms routes
C
INGEST
semantic-chunkers
semantic chunks
G
ORCHESTRATE
graphai
stateful graphs
R
semantic-router
aurelio-labs / semantic-router
Deterministic decisions for LLMs and agents.

A superfast decision-making layer. Route tool-use and safety calls in semantic vector space — no slow LLM round-trips.

$ pip install semantic-router
3,745docs ↗
// flagship

Semantic Router.
The fast decision layer.

Rather than waiting for slow, unreliable LLM generations to make tool-use or safety decisions, route every user turn in semantic vector space — at sub-millisecond latency.

Safety
Deterministic guardrails on every turn.
Scale
Thousands of tools, one decision call.
Speed
Sub-millisecond, no LLM in the loop.
router.py
1from semantic_router import Router, Route
2
3router = Router(routes=[
4 Route("travel", utterances=["book a flight", "find a hotel"]),
5 Route("support", utterances=["talk to a human", "refund policy"]),
6])
7
8router("book me a hotel in lisbon")
9# → Route("travel") · conf=0.94 · 0.06ms
Aurelioaurelio.ai
© 2026 Aurelio Labs · applied ai lab · MIT licensed