Rendering strategy, server actions, edge middleware, observability. The clients are media platforms, fintech products, and B2B SaaS, mostly running production with us inside an embedded team.
Building production web with Vercel.
Next.js applications shipped to real users, including the AI-feature work that pushes the platform.
What we ship on Vercel
The work covers Next.js application development, UI and UX, mobile-aware front end, AI features, and the architectural decisions underneath all of it.
The AI-feature work is the most active part of the practice right now.
Chat interfaces, retrieval-augmented generation, agent workflows running on the same stack as the rest of the product. The AI SDK and v0 are part of the toolkit. The work that determines whether an AI feature is production-ready is the work around them: data plumbing, evaluation, error handling, latency, and cost. We've shipped that work for clients, and we know where the gotchas live.
The practical effect of the partnership for clients is access.
Our engineers work directly with Vercel's solutions team on architectural questions and early product features. Thorny issues get answered faster.
Where Vercel fits, and where it doesn't
Vercel is the right answer for product teams shipping modern web applications who don't want to run their own platform team. It isn't the right answer for every workload. Long-running compute, large-scale data processing, and certain classes of regulated workload belong elsewhere.
We've also pushed back on the platform where we've found the limits. Our engineers have built custom GitHub Actions pipelines to work around Vercel's deployment rate limits on large mono-repos, and have fed back to Vercel on token granularity and OIDC support for CI providers. Real partnership runs both ways.
We're up front about which parts of an application belong on Vercel and which belong elsewhere. That's a judgement we make on each engagement, not a default we apply.
We've also pushed back on the platform where we've found the limits. Our engineers have built custom GitHub Actions pipelines to work around Vercel's deployment rate limits on large mono-repos, and have fed back to Vercel on token granularity and OIDC support for CI providers. Real partnership runs both ways.
We're up front about which parts of an application belong on Vercel and which belong elsewhere. That's a judgement we make on each engagement, not a default we apply.
Working with the Vercel team
Beyond client delivery, we run regular technical enablement sessions with Vercel's solutions team and participate in early-access programmes for new product features. Joint customer success stories feed both Vercel's marketing and our own.
Keep Reading
Partnerships
Cloudflare
The other half of how we ship at the edge.
Learn More
Governance
AI-Augmented Delivery
How AI changes the way we build software for clients.
Learn More
Engineering
Software Engineering
Experienced engineers building robust, scalable systems with a focus on what matters.
Learn More
Building or rebuilding on Next.js?
We'll talk through the architecture before talking about the engagement.