Pooja Parthasarathy
@pparthasarathy0
strategic ops at @Stripe | prev @atlas @abstractopsco @narrativesci (acq @salesforce) @neubergerberman
i wonder how both the number of [D]AUs and # sessions per [day] spiked for @OpenAI @AnthropicAI and @GeminiApp after they added voice mode.
delightful @whimsical feature: pasting a link to diagram 1 (D1) into diagram 2 (D2) results in a pretty thumbnail of D1 getting dropped into D2. love.
In life, once you viscerally experience the magic of compound interest, you start to see it everywhere. In personal relationships, as a parent, in your health and your career. As you grow older, life = having fun dollar cost averaging, all day, every day.
as much context as is humanly possible (within token constraints [0]) + asking LLMs to co-write prompts and evals => at least 5X improvement in the quality of responses. [0] my guess is this will become / will need to become less of a constraint over time.
Listening to the @stripe Sessions keynote makes me wonder: in an MCP first world, what becomes the hub MCP server that drives all the spokes?
feature request for @googlesheets: would like to add a group by view as a persistent tab in the table. very hard to use + share otherwise.
one of my favorite things about @stripe: quarterly company-wide Q&As with our exec team. they zig-zag between questions on the macroeconomy, legislative trends, agentic commerce, card network economics, and [other complex topics] at dizzying speed. love sponging it all up.…
3 takeaways from the excellent @talraviv's workshop on building your PM AI copilot: 1. context is everything 2. there really are no 'best practices'; tinkering is the best way to learn 3. context is everything thanks for organizing @MarcBaselga!
interesting new feature in @googledocs--click to edit text using a custom prompt or shortcuts. idea of a 'prompt shortcut' makes sense. but doesn't seem like surrounding context is being passed in. output feels more thesaurus, less smart LLM.
