One thing I'd add to the test: check whether the task involves judgment under ambiguity or just retrieval and formatting. AI features nobody asked for tend to cluster around the second category, because it's easy to demo and hard to justify skipping. The features users actually adopt usually replace a step where they were making a guess, not a step where they were just waiting for data to load.
So, so true - thank you Elena - I'm seeing so many leaders drawn in to launching EVERYTHING and doing MORE now that engineering is (seemingly) cheaper - but it creates a mess not just for maintenance but for users. Adding all 25 requested features is not a strategic path to deepening PMF or a coherent user experience. Applying energy to known stall points is a good guide!
Thanks, Rebecca! 🤗 It is so tempting to say yes to everything right now because spinning up a new feature takes hours instead of months. But if it doesn't solve a real friction point, you are just trading engineering debt for maintenance and token noise.
Yes! If you’re able to sunset features and really you’re building a little validation test, that’s one thing - but most people (myself included) have such a hard time removing features. The idea that we’ll just stop doing things is usually unreasonably optimistic 🤣
Your DraftKit and advent calendar comparison is the useful part because it shows that "built with AI" and "needs AI inside it" are separate decisions. In marketing, I add one question to the Stall Map: what happens if the output is wrong? A draft outline can wait for review. A customer message or budget change needs a smaller test and a named approver. Same potential value, very different blast radius.
LOVE it Zarif! adding "what happens if the output is wrong?" to the test is the ultimate sanity check, especially when customer messaging or billing is involved.
If anyone else is staring at a roadmap full of speculative AI features, run them through the five questions. If you can't name the exact stall point it fixes, cut it. 👌
One thing I'd add to the test: check whether the task involves judgment under ambiguity or just retrieval and formatting. AI features nobody asked for tend to cluster around the second category, because it's easy to demo and hard to justify skipping. The features users actually adopt usually replace a step where they were making a guess, not a step where they were just waiting for data to load.
Right, Mohamed! Retrieval and formatting features make for great demos, but rarely these survive first contact with real users!
So, so true - thank you Elena - I'm seeing so many leaders drawn in to launching EVERYTHING and doing MORE now that engineering is (seemingly) cheaper - but it creates a mess not just for maintenance but for users. Adding all 25 requested features is not a strategic path to deepening PMF or a coherent user experience. Applying energy to known stall points is a good guide!
Thanks, Rebecca! 🤗 It is so tempting to say yes to everything right now because spinning up a new feature takes hours instead of months. But if it doesn't solve a real friction point, you are just trading engineering debt for maintenance and token noise.
Yes! If you’re able to sunset features and really you’re building a little validation test, that’s one thing - but most people (myself included) have such a hard time removing features. The idea that we’ll just stop doing things is usually unreasonably optimistic 🤣
Your DraftKit and advent calendar comparison is the useful part because it shows that "built with AI" and "needs AI inside it" are separate decisions. In marketing, I add one question to the Stall Map: what happens if the output is wrong? A draft outline can wait for review. A customer message or budget change needs a smaller test and a named approver. Same potential value, very different blast radius.
LOVE it Zarif! adding "what happens if the output is wrong?" to the test is the ultimate sanity check, especially when customer messaging or billing is involved.
If anyone else is staring at a roadmap full of speculative AI features, run them through the five questions. If you can't name the exact stall point it fixes, cut it. 👌