Weave is an engineering intelligence platform aimed at CTOs and engineering leaders trying to figure out where their AI coding budget is actually going. The company plugs into Git and coding tools like Claude Code, Cursor, and Codex, then uses its own models to measure how engineers use AI on every pull request, benchmark that against other teams, and route each prompt to the most cost effective model for the job. The launch marks a $13.5M Series A led by Standard Capital , coming just months after a $4.2M seed earlier this year.
The pitch lands at a moment when frontier model spend has become a real line item and most companies have little visibility into what they are paying per engineering hour. Weave claims that most prompts do not need a frontier model, and its router is designed to exploit that by picking a cheaper option when it will not hurt output quality. The open source router sits at localhost:8080 in front of tools like Claude Code, Codex, and Cursor, and advertises 40 to 70 percent cost reductions from a single endpoint change. Customers referenced in the launch include Robinhood, PostHog, and Telnyx.
Weave was founded in 2024 by Andrew Churchill and Adam Cohen and is part of Y Combinator's W25 batch, with a team of 14 based in San Francisco. Cohen, who fronts the launch video, describes the company's mission on his profile as ending "tokenmaxxing psychosis," a nod to the broader argument here that spending on tokens should be measured with the same rigor as spending on people.
Comments (14)
Sign in to join the discussion.
P
Priya K.Jul 28, 2026
The fear-conquering bit in the video actually made me watch to the end, which is more than I can say for 99% of Series A announcement clips. Whoever edited that earned their paycheck.
T
Tomasz WroblewskiJul 28, 2026
Tokenmaxxing psychosis is the funniest possible category name and I hope you never let marketing sand it down.
K
Kenji AbeJul 28, 2026
Cool cool but where are the docs, rate limits, and a self-serve tier? Half of us can't get procurement to sign a Series A vendor contract before Q3.
Y
Yara HaddadJul 28, 2026
ok wait, is this evals, routing, caching, or all three wearing a trench coat? The site is vibes-forward but I need a picture of the boxes and arrows.
F
Fumi AdeyemiJul 28, 2026
First 3 seconds of the video: guy staring into void. Retention hook is elite, whoever storyboarded this understood the assignment.
G
Grigor PetrosyanJul 28, 2026
Respect for putting the founder on camera doing something uncomfortable instead of another whiteboard voiceover. The genre needed a shakeup.
L
Leo MarchettiJul 28, 2026
Every MLOps company says they save you millions on tokens. I'll believe it when I see a real bill diff, not a marketing dashboard screenshot.
H
Hans OlovievJul 28, 2026
Staff eng question: is the optimization layer inline on the request path or async? Because one of those numbers is a latency graph I actually care about.
D
Deepa RanganJul 28, 2026
We had an internal thing called TokenTrim in 2019 that did like 30% of this. Died in a reorg. Glad someone is actually shipping it to customers.
N
Ngozi BalogunJul 28, 2026
Pricing page is suspiciously absent. If you're saving me millions I assume you want a cut, so just tell me the cut.
A
Amir FarhadiJul 28, 2026
Building adjacent in the eval space and honestly the token-intelligence angle is a smart wedge. Slightly annoyed I didn't frame mine that way first.
M
Mei ZhangJul 28, 2026
Curious about the like-to-view ratio on that video, it's punching above its weight. The bungee gag is doing a lot of work.
S
Sven OstergaardJul 28, 2026
This might legitimately be the last dev tool launched without an agent doing the demo. Enjoy the vintage era while it lasts.
R
Rauf YildizJul 28, 2026
hot take: 'AI has taken over engineering and companies are wasting millions' is the single most fundable sentence of the year.
Work With Us
Book an intro call to explore our suite of services. We might have a waitlist.
Weave
Al Engineering and token intelligence to optimize the SDLC
Al Engineering and token intelligence to optimize the SDLC
Tags
TLVC Rating
Community Rating
No ratings yet
to rate this launch.
Weave is an engineering intelligence platform aimed at CTOs and engineering leaders trying to figure out where their AI coding budget is actually going. The company plugs into Git and coding tools like Claude Code, Cursor, and Codex, then uses its own models to measure how engineers use AI on every pull request, benchmark that against other teams, and route each prompt to the most cost effective model for the job. The launch marks a $13.5M Series A led by Standard Capital , coming just months after a $4.2M seed earlier this year. The pitch lands at a moment when frontier model spend has become a real line item and most companies have little visibility into what they are paying per engineering hour. Weave claims that most prompts do not need a frontier model, and its router is designed to exploit that by picking a cheaper option when it will not hurt output quality. The open source router sits at localhost:8080 in front of tools like Claude Code, Codex, and Cursor, and advertises 40 to 70 percent cost reductions from a single endpoint change. Customers referenced in the launch include Robinhood, PostHog, and Telnyx. Weave was founded in 2024 by Andrew Churchill and Adam Cohen and is part of Y Combinator's W25 batch, with a team of 14 based in San Francisco. Cohen, who fronts the launch video, describes the company's mission on his profile as ending "tokenmaxxing psychosis," a nod to the broader argument here that spending on tokens should be measured with the same rigor as spending on people.
Comments (14)
The fear-conquering bit in the video actually made me watch to the end, which is more than I can say for 99% of Series A announcement clips. Whoever edited that earned their paycheck.
Tokenmaxxing psychosis is the funniest possible category name and I hope you never let marketing sand it down.
Cool cool but where are the docs, rate limits, and a self-serve tier? Half of us can't get procurement to sign a Series A vendor contract before Q3.
ok wait, is this evals, routing, caching, or all three wearing a trench coat? The site is vibes-forward but I need a picture of the boxes and arrows.
First 3 seconds of the video: guy staring into void. Retention hook is elite, whoever storyboarded this understood the assignment.
Respect for putting the founder on camera doing something uncomfortable instead of another whiteboard voiceover. The genre needed a shakeup.
Every MLOps company says they save you millions on tokens. I'll believe it when I see a real bill diff, not a marketing dashboard screenshot.
Staff eng question: is the optimization layer inline on the request path or async? Because one of those numbers is a latency graph I actually care about.
We had an internal thing called TokenTrim in 2019 that did like 30% of this. Died in a reorg. Glad someone is actually shipping it to customers.
Pricing page is suspiciously absent. If you're saving me millions I assume you want a cut, so just tell me the cut.
Building adjacent in the eval space and honestly the token-intelligence angle is a smart wedge. Slightly annoyed I didn't frame mine that way first.
Curious about the like-to-view ratio on that video, it's punching above its weight. The bungee gag is doing a lot of work.
This might legitimately be the last dev tool launched without an agent doing the demo. Enjoy the vintage era while it lasts.
hot take: 'AI has taken over engineering and companies are wasting millions' is the single most fundable sentence of the year.