Pilot, the startup I work at, announced our Series B today.

This is really exciting for me. I’ve been here for a little over a year and a half now. I joined right before the Series A, and since then we’ve outgrown offices twice and hired to more than 4x what our size was when I joined. This funding round raises our valuation to $355M (!) and also includes a strategic investment from Stripe, aligning their interests with ours (!!).

I joined Pilot in order to learn more stuff about engineering and also to learn more stuff about organizational efficiency and effectiveness, and I’ve been really satisfied on both fronts. Right after I joined I sometimes described the decision to friends as “Well, this is the founders’ third startup together, and their first two were successful exits, so… [eyebrow wiggle suggesting probable future success].” Waseem, Jessica, and Jeff really do know what they’re doing, and I haven’t been disappointed.

Optimize processes, not just tech

At Pilot I’ve learned a bunch about the optimization of human-centered processes and why it’s important. As someone who came from a hard computer science background, this lesson is one that I’ve only really absorbed during the last year or two, and it came both from observing Pilot and from observing other highly functional institutions within my broader community.

Pilot works on making bookkeeping less painful for humans to do, so that we can provide it at scale. This means not only that the engineering team’s automation tools have to be really good, but also that our human-centered processes for onboarding new team members and for spreading knowledge around have to be really good. As Catherine Olsson summarized my own description to me the other day, Pilot is “actually in the process optimization business.”

Checklists, common knowledge generation, and open lines of interpersonal communication – these and other tools are all really important to designing well-functioning processes for organizations made out of people.

Aim at real problems, and propagate incentives

I’ve also learned that the powerful engine of process optimization absolutely must be aimed at real problems in the world. I think this is a crucial puzzle piece that distinguishes startups that succeed from startups that don’t. I’ve liked getting to observe how the real-world problems faced by Pilot’s clients get propagated all the way down to the level of engineers’ day to day efforts.

Our organization is actually set up such that the incentives are really well aligned throughout. Potential clients (the source of real-world problems) come in via the sales team. Fulfilling our service to our clients becomes the problem of our operations team, who is tasked with closing the books. We record metrics about what’s taking our operations team the most time, as well as free-form feedback, and then the product team distills that data into goals for engineering and design.

The entire product process is legible and transparent to people on the engineering team, so I can audit what I’m working on and why, and question decisions at the appropriate level if I feel something’s off. Team members are also empowered to propose product directions directly.

If anything, the one thing I’d worry about with our setup is a hypothetical world where the most stress to deliver our service to our clients is felt by the operations team, and where the engineering team doesn’t feel the fire as much as the operations team does. We have been addressing this by specifying specific goals for the engineering team to meet in order to properly support the operations team, so I can trust that we will succeed inasmuch as we craft those goals carefully. And since we’re using data and metrics to motivate our choices, plus the management team sometimes ends up doing a little bookkeeping if we find ourselves under-capacity, I can trust that the organization’s incentives are aligned to craft those goals well.

Culture can be engineered

I’ve learned by example what it looks like for a community to deliberately engineer itself to have a good culture. On day 1, I was impressed when Jessica replied to a member of the team pointing out something to fix about part of our infrastructure by responding with “Good callout,” encouraging honest feedback. Just last month, a member of the engineering team took on the task of writing up a doc about our code review standards, making implicit social norms explicit. From various other members of the team I’ve learned about blameless retrospectives, about the New York Times rule, and about two kinds of feedback.

Pilot also has some employees with roots in the Recurse Center, whose social rules and attention to inclusiveness are another example of an organization achieving a well-engineered culture.

While we don’t currently have an explicit set of company values, as an individual participant in company culture I can tell from everyday communications that we think of ourselves as valuing work-life balance, diversity, and open communication, that we take actions that promote those values, and that we talk about those values often, thereby elevating them to common shared culture.

Short feedback loops can be engineered

Although I already knew that short feedback loops are important for getting good engineering velocity (write unit tests! use debugging tools!), one thing I’ve learned at Pilot is that it’s possible to structurally engineer an organizational system such that it has short feedback loops built in. In particular, our engineering team writes software for our in-house users, which means I have the luxury of not needing to regularly set up A/B tests in order to infer what’s going on in users’ minds; instead, we can ask our users directly.

Processes can become stale – change them

I’ve also seen firsthand that as organizations scale, the old processes and ways of doing things don’t necessarily work anymore and you have to find new ones. I’ve been pleased that as we’ve been growing, the management team has been attentive to whether there are any growing pains showing up in our processes, and I’ve been pleased that we’ve been actively working on moving to newer ways of doing things that work better for our current situation, rather than sticking with old solutions because they’re the way things have always been done.

Since I have joined, we have actively improved our systems for triaging issues, for tracking projects, and for conducting engineering team meetings, each time preventing ourselves from getting stuck in old ways that weren’t serving us as well anymore.

This seems like a pretty hard category of problem in general, and I’m interested in learning more heuristics for how to notice it and address it. It seems that as organizations grow larger, oftentimes problems like this can slip through the cracks.


There are countless other things I’ve learned that I could mention as well. I’ve learned a lot of stuff about good habits for frontend development in Vue.js from Pilot’s first engineer. I’ve learned what good management feels like as an employee from our engineering manager’s care and diligence. I’ve learned new things about computer security directly from the founder of the Twisted project, including how SSL certificate issuance works and how to do formal threat modeling.

Overall, I’ve been really interested in institutioncraft lately, and Pilot feels like a success case in institutioncraft from which I’ve learned a whole bunch. Of course, the things I’ve learned so far, like all things I learn, are a work in progress that is continued every day. I’m looking forward to building more stuff with these amazing people and to learning more as we grow together.