Be opinionated about what makes a tool great, but be pragmatic about how to achieve greatness.


Smaller, nimble teams. The more people you add, the easier it is to keep doing what’s not critically important. With a smaller team, it’s very hard not to stay focused. You constantly need to avoid doing what you want to do in the future for that which has to be done now.

Platform integration as self-help. Pocket’s platform mentality has been a huge part of its growth. Allowing companies to build on its platform easily has been the first step, but scale happens when they can do so independently.

Woo big partners tenaciously with your platform, then keep them with transparency. Simple platform integration and self-service doesn’t mean that startups should go on autopilot. Pocket stays on the radar of strategic companies that it believes will benefit from its product. Remember: getting an app or company on your platform marks the end of a deal, but the beginning of an official working relationship. For Pocket, channeling “generous transparency” is the way it retains its partners and continues to improve its product with such a small team.

The takeaway for lean startups is to construct a platform that can enable external parties to operate as extensions of your team: sales, engineering and business intelligence. In the case of Pocket, its platform is simple to integrate, built for self-service and brings with it an increasingly rich set of data, that when aggregated, benefits all its partnering companies.

Talk to people. Talk to your potential customers and people around you.

Marketing is important. It doesn’t help to build a cool product if people don’t know about it.

Don’t be afraid of sales. Your product is solving a problem, it’s only fair that your customers pay money for it.



  • Keybase, 20n, Civis Analytics, Planet Labs, Smyte, Spokeo, Flexport, Remix, Connectifier, Opendoor, Translation Exchange, and CrowdFlower have really interesting problems to work on. A pattern here is technology being applied in a relatively greenfield domain.
  • Samsung Accelerator is an interesting example of how giant enterprises can get into the entrepreneurship space.
  • Microservices Microservices Microservices Microservices and even more Microservices.
  • Went to an interesting Go/Microservices talk by the VP of engineering at Grabtaxi. Though not the focus of the talk, there was a brief mention of the familar shift from a monolithic Rails application used largely in the validation phase to a distributed, microservices Architecture with Node.js and now Go. Is there a right way to build software? We’ve seen this pattern in other companies too, such as Songkick, Soundcloud, and many others. It’s actually ok to use Rails to validate your MVP! Optimizing early by starting with microservices from the start is likely optimizing for the wrong thing.
  • Performance is the last reason people do microservices. It’s more about organizational-level work distribution and being able to develop services in parallel across multiple teams.
  • Monolithic approaches on the other hand, allow sweeping, organization-level changes more easily. See discussion about Google’s 2B LOC monolith here.

  • What if scheduling office cleanings, arranging a handyman to fix sinks, catering, hiring photographers, and other service transactions can be made via a single API call?
  • I hope to one day be as prolific in open source as maxogden and tj!
  • I played Undertale recetly and found it very charming. A strong cast of characters, great art and soundtrack, and solid writing makes a it fantastic experience. Recommended.
  • Hong Kong uses AI to schedule subway repairs and approve visa applications.
  • Magic, Operator, DudeGenie, FavorDelivery and many other services are popping up. They follow the intersection of the current biggest trends: messaging, mobile, and on-demand services.
  • There are some similarities between BotSquad and morph. It’s interesting how they used Heroku’s buildpack system to handle multiple languages and library dependencies!