When a six-person startup’s bottleneck isn’t the AI
Stably, a six-person startup, builds AI agents that run autonomous end-to-end tests. The agents deploy preview URLs, read code diffs, and validate whether changes actually work, integrating directly with coding agents like Claude Code. The pitch: AI can write code fast, but without automated validation, teams burn time checking everything by hand.
Yet Stably’s own constraint wasn’t model quality. It was the overhead of running infrastructure. Every feature meant deployment decisions. Every new agent brought a fresh round of DevOps anxiety. For co-founder and CEO Jinjing Liang—who previously built infrastructure on the Google Chrome team—and co-founder Neil, an Uber safety team alum, reliability was a baseline requirement. But with a team that small, spending engineering cycles on infrastructure wasn’t sustainable.
Their solution was to run the entire stack on Vercel. The frontend and backend sit on the platform, using AI Gateway for scalability and large TPM limits, plus Sandbox for agentic code execution. The team evaluated vendors thoroughly for each component, and Vercel kept winning on the same calculus: better developer experience, fewer bugs, fewer restrictions, and faster feature additions.
Liang’s philosophy: “You should just trust the community unless you have a reason not to. Your default should be: I should justify why I shouldn't build on Vercel.”
From “say something” to “just do it”
Once infrastructure stopped being a decision point, the team’s rhythm changed. The old motto was “if you see something, say something.” Now it’s “if you see something, just do it.” Engineers stopped asking whether deployment would block them. Product ideas turned into shipped features the same day.
Internal tools that once felt like “someday” projects now get built in hours. Stably has shipped Slack bots that search databases and answer questions, dashboards, and new agents on the same Vercel stack—even internal AI agents that deploy entire services autonomously. As Liang describes it, someone can decide they want to add a new product and simply go ahead: “You can do all the DevOps. It's no longer hard to do. And you can also just ask our Slack bot, which is running completely on Vercel, and that bot will go in, build everything for you, and put up a PR for you.”
Impact at a glance
Launch new product lines in hours instead of weeks
Ship internal tools (Slack bots, dashboards) without worrying about deployment
Scale from early product to enterprise contracts on the same infrastructure
Entire stack run by a six-person team, revenue-generating and lean
Lessons from the Accelerator
Stably entered Vercel’s AI Accelerator at an inflection point. Before applying, the team had spent years iterating on a product that treated testing as an afterthought—a separate flow tacked onto the end of the pipeline. Then came the realization that embedding testing directly into the developer workflow was the real breakthrough. They built an initial version of that newer product, got early feedback, and applied to the accelerator on that momentum.
The accelerator program shaped both product and business thinking. A fireside chat with Vercel, MongoDB, and HubSpot gave Liang clarity on pricing strategy, driven by the insight that “pricing is product.” The shift was personal: “It used to be I gave the pricing part to one of our interns,” she says. “But now, after that, I actually review the whole pricing part. I'm the only one on the team who knows more about the pricing code than anyone else. That was kind of eye-opening to see the amount of thought going into it.”
Winning the accelerator, along with investment from Vercel Ventures, opened doors beyond funding. Enterprise buyers saw the Vercel backing as a signal. “There have been many times where a customer said, ‘Oh, you guys were invested by Vercel, right? That's a great signal.’ It reinforced what we were already building, and gave enterprise buyers extra confidence.” The LinkedIn announcement alone generated multiple inbound leads, some now closing as contracts.
Since the program, Stably’s trajectory has shifted in two ways. First, increased open-sourcing of core product and internal tools, following what Liang calls the Vercel playbook—Next.js and Turborepo-style community building. Second, real revenue traction, with enterprise pilots converting to contracts.
“It always happens: Vercel just has the best developer experience, less bugs, less restriction, and we can see they're adding more and more features a lot quicker.”
Jinjing Liang CEO
Infrastructure that stays out of the way
For Stably, infrastructure was never supposed to be the story. The product was. Vercel’s role was keeping deployment concerns out of the way so a revenue-generating team of six could ship new product lines in hours instead of weeks, prepare to open-source its testing framework, and convert enterprise pilots to contracts.
The motto that started as an internal culture shift—if you see something, just do it—has become the operating principle for the entire company.
