Momentic's Mo is an AI QA engineer that developers point at a web, iOS, or Android build to run a bug bash before shipping. Rather than writing or maintaining Playwright, Selenium, or Cypress suites, teams give Mo a URL (optionally attached to a PRD, Linear ticket, or Confluence doc), and the agent explores user flows, probes edge cases, and returns a report of what failed with reproduction steps and video evidence for each finding. The launch tweet claims that in internal evaluations against Codex with Playwright MCP, Mo surfaced 3.8x more bugs, 9x more bugs per hour, and cost roughly half as much per bug.
The timing is tied to the gap between how fast coding agents now ship features and how slowly verification has kept up, with Momentic positioning Mo as the counterpart to code-generation agents: agents write the code, Mo bug bashes it, agents fix what breaks. During a private beta, engineering teams at Notion, Glide, and Superpower ran Mo against their own apps, logging more than 1,100 bug bashes and 3,500 verified bugs , and the broader Momentic platform is already used by Notion, Whatnot, Retool, Webflow, Framer, and others.
Momentic is a Y Combinator company founded in 2023 by CEO Wei-Wei Wu, previously an engineer at Assembled and the first engineer at Nashi, and CTO Jeff An, who built cloud and test infrastructure at Splunk, Google, and Robinhood. The company previously spent two years building a lower-code test automation product before pivoting the pitch toward removing test suites altogether, which is what this launch of Mo represents.
Comments (12)
Sign in to join the discussion.
P
Priya Rajagopal10d ago
The 3.8x vs Codex framing is spicy but I want to see the eval harness. Who graded the bugs, Mo or a human in the loop?
T
Tomas L.10d ago
Naming it Mo is a power move. One syllable, flirts with the company name, works as a verb ("did you Mo it?"). Branding team ate.
Z
zane10d ago
Repro steps + logs + video per bug is the only part I care about. Half my triage time is just reproducing what QA swore happened.
A
Aminah K.10d ago
The split-screen in the demo video where you see all 100 agents poking at the app simultaneously is doing more marketing work than the tweet copy. Chef's kiss on the pacing.
D
Dmitriy O.10d ago
First 8 seconds of the launch video does the whole pitch without a voiceover. Rare to see devtool founders respect the scroll like that.
K
Kenji Watanabe10d ago
Adjacent founder here (we do e2e for mobile). Genuinely curious how Mo handles flaky selectors on apps with heavy client state. That's where every "AI QA" demo I've seen quietly falls apart.
H
Hank R.10d ago
Call me old fashioned but didn't we have humans for this. Anyway, filing this under "will revisit in 2027."
C
Chioma Eze10d ago
Dumb question from a junior PM: when Mo finds a bug that's actually intended behavior, how do I teach it the difference? Does it remember across runs or do I keep yelling at it forever?
R
ravi10d ago
How big is the team shipping this? If it's under 15 people I'm going to go cry into my own repo for a bit.
S
Sofia Marchetti10d ago
Already told my portfolio to try this before I even opened the website. The logo wall alone (Notion, Whatnot, Retool) is the whole sales pitch.
M
merkle.eth10d ago
Imagine each bug minted as an on-chain bounty NFT and agents stake to claim them. I'll see myself out but you know I'm right.
Y
Yuki Tanabe10d ago
Who's hiring eval engineers in SF? I know someone who built the test infra at a sim-robotics startup and would eat this problem for breakfast.
Work With Us
Book an intro call to explore our suite of services. We might have a waitlist.
Momentic AI
Mo, the AI QA engineer that bug-bashes your app before every release. Trusted by Notion, Whatnot, Retool, Webflow, Framer, and 1.6k+ others.
Mo, the AI QA engineer that bug-bashes your app before every release. Trusted by Notion, Whatnot, Retool, Webflow, Framer, and 1.6k+ others.
Tags
TLVC Rating
Community Rating
No ratings yet
to rate this launch.
Momentic's Mo is an AI QA engineer that developers point at a web, iOS, or Android build to run a bug bash before shipping. Rather than writing or maintaining Playwright, Selenium, or Cypress suites, teams give Mo a URL (optionally attached to a PRD, Linear ticket, or Confluence doc), and the agent explores user flows, probes edge cases, and returns a report of what failed with reproduction steps and video evidence for each finding. The launch tweet claims that in internal evaluations against Codex with Playwright MCP, Mo surfaced 3.8x more bugs, 9x more bugs per hour, and cost roughly half as much per bug. The timing is tied to the gap between how fast coding agents now ship features and how slowly verification has kept up, with Momentic positioning Mo as the counterpart to code-generation agents: agents write the code, Mo bug bashes it, agents fix what breaks. During a private beta, engineering teams at Notion, Glide, and Superpower ran Mo against their own apps, logging more than 1,100 bug bashes and 3,500 verified bugs , and the broader Momentic platform is already used by Notion, Whatnot, Retool, Webflow, Framer, and others. Momentic is a Y Combinator company founded in 2023 by CEO Wei-Wei Wu, previously an engineer at Assembled and the first engineer at Nashi, and CTO Jeff An, who built cloud and test infrastructure at Splunk, Google, and Robinhood. The company previously spent two years building a lower-code test automation product before pivoting the pitch toward removing test suites altogether, which is what this launch of Mo represents.
Comments (12)
The 3.8x vs Codex framing is spicy but I want to see the eval harness. Who graded the bugs, Mo or a human in the loop?
Naming it Mo is a power move. One syllable, flirts with the company name, works as a verb ("did you Mo it?"). Branding team ate.
Repro steps + logs + video per bug is the only part I care about. Half my triage time is just reproducing what QA swore happened.
The split-screen in the demo video where you see all 100 agents poking at the app simultaneously is doing more marketing work than the tweet copy. Chef's kiss on the pacing.
First 8 seconds of the launch video does the whole pitch without a voiceover. Rare to see devtool founders respect the scroll like that.
Adjacent founder here (we do e2e for mobile). Genuinely curious how Mo handles flaky selectors on apps with heavy client state. That's where every "AI QA" demo I've seen quietly falls apart.
Call me old fashioned but didn't we have humans for this. Anyway, filing this under "will revisit in 2027."
Dumb question from a junior PM: when Mo finds a bug that's actually intended behavior, how do I teach it the difference? Does it remember across runs or do I keep yelling at it forever?
How big is the team shipping this? If it's under 15 people I'm going to go cry into my own repo for a bit.
Already told my portfolio to try this before I even opened the website. The logo wall alone (Notion, Whatnot, Retool) is the whole sales pitch.
Imagine each bug minted as an on-chain bounty NFT and agents stake to claim them. I'll see myself out but you know I'm right.
Who's hiring eval engineers in SF? I know someone who built the test infra at a sim-robotics startup and would eat this problem for breakfast.