1 min read

Coffee notes

These are a few stray notes from a Dreamforce 2019 session on architecture led by Peter Coffee. Hopefully you will find them insightful.
Coffee notes
Photo by Mike Kenneally / Unsplash

At Dreamforce 2019 I was able to sneak into a very intimate pre-conference session led by Peter Coffee that was geared towards architects operating in the nonprofit vertical. As it often happens with Peter, it was thoroughly inspiring.

I recently came across some of the ideas that I was able to jot down, and I wanted to share them more broadly, because I think that they are very good food for thought for those of us with architectural aspirations.

  • Architecture is about people and the systems that connect them. Systems must be reliable and trustworthy. Investing in systems must include investing in the system landscape's trust directive.
  • An architect does not need to be a "seller" to get projects and initiatives through. Listen. Demonstrate empathy. Ask great questions. Be smart.
  • Pain points are a great entry point for change, but let's talk about the problem, not about symptoms that need to be remedied.
  • To build consensus around transformational activities (e.g. projects or new features), focus on the needs rather than activities, speak in verbs, and name initiatives after the goal they fulfill.
  • Bring the human being into the equation, and lead with curiosity. Shift from "you are doing it wrong" to "why do you put up with this?".
  • The features that matter follow a Pareto distribution: 20% of the features represent at least 80% of the functionality that matters.
  • For better or worse, investing in productivity is not like buying nicer furniture for the office. For better, because it has the potential of being transformative; for worse, because it is way less self-evident than a cushy new sofa.
  • Requests for proposals (RFPs) introduce a pernicious dynamic in innovation/disruption initiatives, because the decision makers ask from existing vendors that know about the organization's biases and there is little learning on the client's end. This is conducive to situations in which we try to fix a broken system by attaching a new database to it.
  • For an AI to be trusted, it must be handled ethically.