Uppersky Podcast

The episode contains a handful of concrete operational details (Productboard's 2019 ML model, 3 - 5k monthly feedback pieces, 40% of work tied to AI) but is padded with well-worn PM platitudes - vitamins vs. painkillers, founder-as-first-PM, and the 80% unused features stat. The filler-to-insight ratio drags the score down significantly.

“we have roughly, uh, roughly 1112, uh, product managers, product team, we would get three to 5,000 pieces of feedback monthly”

“roughly 40% of what we are doing is associated to AI today”

The distribution→engineering→discovery bottleneck framing is the most interesting structural argument in the episode, but the vast majority of content recycles familiar PM orthodoxy: validate before building, pain vs. vitamin, context matters. There is little that would challenge or reframe a practitioner's existing mental models.

“before Internet the distribution of the product in a software...was the biggest problem...And then...engineering in like building stuff used to be bottleneck...Then it boils down to the initial component and it's the uh product discovery”

“a lot of the early failures would definitely come from...you built something. Maybe it isn't the right thing because you haven't done the...discovery”

Martin Felcman is a genuine practitioner who joined Productboard at ~25 people and grew with it through scale-up, giving him credible insider perspective on a real B2B SaaS product org. However, he is not in a C-suite or definitively senior strategic role, and his domain is narrow; the transcript never fully exploits the depth his tenure should provide.

“I joined Product board back in 2018. Uh originally my story is a little bit complicated, might get there later. But uh it was when the company had some run 25, 27 people”

“we are drinking our own champagne. So we are using a product board to build product boards”

There are a handful of real specifics - the 2019 ML model build, 3 - 5k monthly feedback items, the paused capacity planning feature, ~40% AI investment share - but most examples remain illustrative rather than measurable, and financial outcomes, customer counts, and timeline data are conspicuously absent.

“one particular uh thing that we have done a discovery on and we have decided to pause it...is the functional door capabilities around capacity planning”

“we actually started relatively early on...back in 2019, uh, probably like when we actually built our own machine learning model”

The host makes occasional genuine attempts to press for specifics ('I will push you again,' 'Do you have any example from product board') and covers a reasonable range of angles, but the follow-ups frequently dissolve into 'Got it, got it' and broad topic pivots rather than drilling into vague or incomplete answers. No meaningful pushback occurs.

“I will push you again like how and how this compare for example with I know about J Pop”

“Do you have any example from product board like maybe recently that you decided to discard not to build something”