OBS 001
Company
Relay
Durable execution and observability for AI agents in production.
20 paying customers in six months
Source sky survey / public record
Source ofYOUPrimary object / SOY 0001
Priya Nair
Founder & CEO, Relay
Site San Francisco, CA
Log entry 0001
Instrument: self
Agents are easy to demo and hard to run. I build the infrastructure that makes AI agents reliable enough for production, so teams can ship them, not babysit them.
Chart 01
CV-RA
Career constellation
Epoch 2023 - present
Founder & CEO, Relay
Durable execution for AI agents in production.
Epoch 2020 - 2023
Staff Engineer, Cloudbase
Led the platform reliability team.
Epoch 2016
B.S. Computer Engineering, UC Berkeley
Chart 02
MAG-LG
Magnitude legend
Build developer infrastructure for AI agents.
Primary stars
- Infra
- Reliability
- Developer experience
Objects in focus
- AI agents
- Distributed systems
- DX
Chart 03
LOG-LF
Long-form observation
I spent seven years keeping a cloud platform up at scale, then watched every team around me struggle to run agents in production. The model was never the hard part; state, retries, and recovery were. Relay is the infrastructure I wished existed: durable execution, observability, and guardrails, so a run survives whatever happens halfway through.
Chart 04
OBS-WK
Observed works
OBS 002
Product
Durable execution engine
The core runtime that checkpoints agent state so a run survives crashes, restarts, and retries.
OBS 003
Engineering
Platform reliability at Cloudbase
Led the reliability team that owned the durable-execution primitives underneath a cloud platform.
Incident rate cut by half
Chart 05
PHO-ME
Photometric evidence
Measure 01
$6M seed
Measure 02
20 customers
Measure 03
Team of 12
Chart 06
TRG-SV
Observation program
01
Design-partner program
Works directly with a small number of teams putting agents into production to shape the platform.
02
Speaking
Talks on agent reliability and durable execution for engineering teams.
Chart 07
FLD-NT
Field notes
Reliability is the product
For agent infrastructure, the demo is not the hard part. Durability, observability, and recovery are the product. Everything else is a feature.
Agents are a systems problem
The hard part is not the model. It is state, retries, and recovery when something halfway through goes wrong.
Chart 08
QRY-AR
Query archive
- What does Priya do?
- She is the founder and CEO of Relay, which builds developer infrastructure that makes AI agents reliable enough to run in production.
- What is Relay?
- Durable execution and observability for AI agents, so a run survives crashes, restarts, and retries. Twenty paying customers in its first six months.
- What is her background?
- Seven years in distributed systems and platform reliability, most recently as a staff engineer leading the reliability team at Cloudbase.
Q01
Q02
Q03
Chart 09
SIG-LN
Signal links
Observatory·shown with a fictional demo record
See it on your record →