The Farmer's Dog Innovation & Technology Culture

How does innovation show up in your company culture?

Innovation at The Farmer’s Dog starts with a low barrier to putting ideas on the table. If you have something to propose, the path is simple: Write it up and share it. We have a public engineering channel where design docs and proposals flow constantly. The goal is fast calibration and getting feedback from people you might not naturally talk to, so decisions don’t happen in silos.

What I’m most proud of is that technical alignment doesn’t depend on hierarchy. With a clear, shared vision, our teams align pragmatically and refine ideas in the open. 

On the AI front, we’re intentional. We use AI for code generation, automated code review and design briefs. We’ve also enabled our product and design teams to generate prototypes with AI tools, creating a tighter partnership between engineering and the rest of the org. The expectation isn’t just “use AI.” The goal is for people to understand its capabilities and make informed decisions about where it adds value. That intentionality is what makes it stick.

What’s one recent innovation that improved user or employee experience?

We’ve invested in developer experience this year and it shows. Our CI/CD pipeline supports 300 to 400 deploys a week, with more than 100 on a busy day. That speed isn’t chaos — it’s the result of a reliable, well-managed system that lets engineers ship quickly and confidently.

We’ve also overhauled onboarding. Getting a new engineer’s local environment up and running used to take eight separate setup commands and a lot of documentation-hunting. We collapsed that down to a single command. We streamlined the supporting docs and built tooling that pulls directly from our internal knowledge base to get new hires productive faster, including onboarding to our AI workflows.

We believe that when engineers spend less time fighting tooling and more time building, the customer benefits downstream. Fast, reliable deploys mean we can iterate quickly and ship fixes the same day we find them.

How do you balance experimentation with stability?

We set ourselves up to move quickly, which means investing just as much in guardrails as we do in velocity. Our releases ship with monitoring that alerts our on-call engineers directly. Major changes go through release documents, ship behind feature flags and often run in shadow mode so we can validate consistency before customers see anything. If something goes wrong, we can roll back fast.

After incidents, we run reviews to understand what happened. The goal is to close the loop every time, turning what we learn from launches, bugs and code review into process improvements. We apply the same thinking to our AI workflows, rolling learnings back into our agents and tooling to catch similar issues going forward.

The piece that ties it together is making sure our people are growing alongside the tools. As we adopt AI across engineering, we want engineers actively engaging with their work, understanding the changes being made and using AI as a way to learn, not just to produce output. Experimentation works when the team is equipped to learn from what it ships.

What new technologies or frameworks are helping your teams move faster and build smarter?

Two things have really changed the day-to-day. The first is Claude, which our engineers use for development and for automating processes across the stack. The interesting part isn’t how much time it saves, it’s where the time went instead. We spend more of the day on design now and less of it typing out implementation, which means we’re thinking harder up front about how something will be extended and maintained a year from now. We’ve also started designing with Claude in mind, laying systems out so the next expansion is something Claude can pick up and carry, so new features arrive fast and don’t rot. 

The second is “temporal.” Subscriptions are long-running processes that have to survive retries, partial failures and deploys and for years that meant stitching together cron jobs and queues and hoping. Now we write them as durable workflows with failure handling built in. Our pipelines are steadier for it and we can ship customer experiences we wouldn’t have risked before.

How do you balance experimentation with reliability in your development process?

We shadow everything. Before an experiment reaches a customer, the new code runs in production right alongside the old code. The customer still gets the old result and we log a comparison of the two. It’s the difference between knowing how a change behaves under real traffic and guessing from a test suite. Once the data looks right we flip it: the new code serves customers, the old code keeps running in the shadows and we keep watching for differences.

While it’s shadowed, new code fails open, so a surprise error falls back instead of breaking someone’s order. We gradually ramp up rolling out the new code to everyone and rolling back is always a single toggle, never a runbook someone has to read at 2 a.m.

None of this is caution for its own sake. One incident can cost weeks of cleanup that a day of shadowing would have prevented. Shadowing is what lets us move fast without gambling with a dog’s next meal.

What role does collaboration play in turning innovative ideas into products? 

Collaboration isn’t a stage here, it’s the whole thing and the launch of Instant Fresh is the clearest example we have.

Our product development team built an entirely new food format, refining it on feedback from employees and, more importantly, their dogs. While that was happening, engineering sat with product managers and designers working out what the customer experience should feel like. That question turned out to require rebuilding how subscriptions work at The Farmer’s Dog. The old model assumed one format and we needed one that could flex, so we rebuilt it in a way that opens doors we haven’t walked through yet.

The whole time, we stayed in the room with product, design, marketing and customer care, discussing how we’d market it, how we’d onboard someone new, what an advisor would need on day one. Everyone weighed in at every stage instead of receiving a handoff at the end. That’s why what we launched was fully thought through and why the whole company was behind it.