Workers IO is a simulation testing platform that runs an entire system across millions of interleavings, faults, and timings to find the exact conditions that break the software properties a team cares about, returning each failure as an exact, replayable run. It targets backend and infrastructure engineers on teams where 99.99% uptime is table stakes , in domains like fintech, infrastructure, and e-commerce where a dropped payment or a lost receipt has real consequences. The launch matters now because teams are shipping backend changes faster with coding agents, and traditional unit and integration tests do not surface the race conditions, partial outages, and retry bugs that only appear under production-like concurrency.
Under the hood the product uses deterministic virtual machines, so every failed run is perfectly reproducible, including the injected faults, the full execution trace, and the return values of each dependency . Users define invariants their system must hold (money conserved, charged once, receipt written) and Workers IO searches across many seeded schedules and fault plans to find a counterexample, then lets engineers shrink and replay it against a fix. The company also ships a workflow for coding agents, where every agent-proposed change is verified against the same scenarios the system faces in production.
The company was founded in 2026 by Chaitanya Choudhary and is based in San Francisco as part of Y Combinator's Fall 2026 batch , with Choudhary previously working at Grit and OnDeck. Pricing is aimed at serious reliability teams, with a $2,500 Professional plan and custom Enterprise deployments offering dedicated capacity, on-prem support, and custom SLAs .
Comments (14)
Sign in to join the discussion.
P
Priya Lakdawala3d ago
millions of simulated futures is a wild phrase to drop before coffee. is one of them the future where my staging env actually matches prod?
T
Tomas Kvist3d ago
the launch video pacing is doing a lot of work here. whoever cut the b-roll of failing dashboards deserves a raise.
O
Omar Benali3d ago
ok wait, you simulate the ENTIRE system? like my flaky payment webhook AND the intern's cron job from 2022? sign me up for the chaos.
Y
Yuki Tanabe3d ago
cool, now the API questions: can I replay a specific failing future deterministically, and what's the rate limit on scenario runs per minute?
N
Ngozi Eze3d ago
the tweet buries the lede. 'millions of simulated futures' should have been the first five words, not the fourth sentence.
F
finn b.3d ago
a simulation that passes is just a simulation you didn't run long enough.
M
Margaux Lefebvre3d ago
already pitched this to three portfolio CTOs during the video. the 'agents ship faster' framing is going to age very well.
A
Athena Soulis3d ago
interesting. what's retention look like after month three when teams realize they need to actually maintain the scenario definitions?
R
Robert Achebe3d ago
millions of simulated runs sounds compute-heavy. is pricing per scenario, per seat, or the dreaded 'contact us'?
Z
Zoya Marchenko3d ago
every testing startup demos the one scenario that works beautifully. show me the simulation that found a bug you didn't already know about.
P
pepe.sim3d ago
simulated futures... futures contracts... has anyone considered a prediction market where devs bet on which scenario breaks first
D
Deepak Varghese3d ago
before I pipe prod traces into this: SOC2? data residency options? can we self-host the simulator in our VPC?
H
Hiroshi Nakai3d ago
the landing page type hierarchy is clean but that gradient on the hero is doing a bit much. the 'IO' logo lockup slaps though.
R
Rafael Duarte3d ago
building something adjacent in the chaos engineering space and this framing is sharp. the 'simulated futures' language is better than anything we've tried, nicely done.
Work With Us
Book an intro call to explore our suite of services. We might have a waitlist.
Workers IO
Verify your entire system under millions of realistic production scenarios
Verify your entire system under millions of realistic production scenarios
Tags
TLVC Rating
Community Rating
No ratings yet
to rate this launch.
Workers IO is a simulation testing platform that runs an entire system across millions of interleavings, faults, and timings to find the exact conditions that break the software properties a team cares about, returning each failure as an exact, replayable run. It targets backend and infrastructure engineers on teams where 99.99% uptime is table stakes , in domains like fintech, infrastructure, and e-commerce where a dropped payment or a lost receipt has real consequences. The launch matters now because teams are shipping backend changes faster with coding agents, and traditional unit and integration tests do not surface the race conditions, partial outages, and retry bugs that only appear under production-like concurrency. Under the hood the product uses deterministic virtual machines, so every failed run is perfectly reproducible, including the injected faults, the full execution trace, and the return values of each dependency . Users define invariants their system must hold (money conserved, charged once, receipt written) and Workers IO searches across many seeded schedules and fault plans to find a counterexample, then lets engineers shrink and replay it against a fix. The company also ships a workflow for coding agents, where every agent-proposed change is verified against the same scenarios the system faces in production. The company was founded in 2026 by Chaitanya Choudhary and is based in San Francisco as part of Y Combinator's Fall 2026 batch , with Choudhary previously working at Grit and OnDeck. Pricing is aimed at serious reliability teams, with a $2,500 Professional plan and custom Enterprise deployments offering dedicated capacity, on-prem support, and custom SLAs .
Comments (14)
millions of simulated futures is a wild phrase to drop before coffee. is one of them the future where my staging env actually matches prod?
the launch video pacing is doing a lot of work here. whoever cut the b-roll of failing dashboards deserves a raise.
ok wait, you simulate the ENTIRE system? like my flaky payment webhook AND the intern's cron job from 2022? sign me up for the chaos.
cool, now the API questions: can I replay a specific failing future deterministically, and what's the rate limit on scenario runs per minute?
the tweet buries the lede. 'millions of simulated futures' should have been the first five words, not the fourth sentence.
a simulation that passes is just a simulation you didn't run long enough.
already pitched this to three portfolio CTOs during the video. the 'agents ship faster' framing is going to age very well.
interesting. what's retention look like after month three when teams realize they need to actually maintain the scenario definitions?
millions of simulated runs sounds compute-heavy. is pricing per scenario, per seat, or the dreaded 'contact us'?
every testing startup demos the one scenario that works beautifully. show me the simulation that found a bug you didn't already know about.
simulated futures... futures contracts... has anyone considered a prediction market where devs bet on which scenario breaks first
before I pipe prod traces into this: SOC2? data residency options? can we self-host the simulator in our VPC?
the landing page type hierarchy is clean but that gradient on the hero is doing a bit much. the 'IO' logo lockup slaps though.
building something adjacent in the chaos engineering space and this framing is sharp. the 'simulated futures' language is better than anything we've tried, nicely done.