Reading your article made me realize how the orgs still operate, and I felt it for you in that role so much, Judy! I’m glad you found it useful and took the lesson learned, but the bigger question is, did you get promoted after all this work?
I don’t think it was a reflection on my skills personally, but a general belief that engineers could do our jobs or perhaps fewer people who are centralized could do enough.
One of the hardest parts about being laid off right now is keeping up and continually closing the skill gap without the work & community at work.
Ah, I’m so sorry! The way you’ve translated that experience into such clear thinking really comes through in this piece, and it’s what the enterprise AI need. I’m glad Substack is giving you a new kind of community around this work.
Judy when you told me this made you think of Kafka I thought you were being dramatic and then I read the part about spending 20 hours on a 1-page email status only to achieve "nirvana" and then never getting feedback again 😂 that's a Kafka short story, literally.
I forget how much of enterprise AI adoption is just all about surviving the organization around the AI....
I hope to never be overly dramatic, but I will be guilty as charged at times.
It is quite something to navigate new things in old systems.
The hardest parts for me are when you disappear between authority boundaries and feel organizationally disconnected to the work, even though you’re actually doing the work.
I think about Kafka working at that insurance company in 1920 probably feeling the same way sometimes.
This is such a useful distinction between experimentation and actual organizational work.
The PoC vs. Pilot framing is especially clear. A lot of AI work gets praised while it is still safely in “various stages of maybe,” but the harder question is whether it can survive contact with ownership, workflow, adoption, and governance.
And the hard thing is the shape of experiments constantly evolves so knowing how to compare them and think about them seems to get more complex, but I think the grounding factors are still the good product managers thinking about value & cost - maintainability.
Or the trickiest one yet, do we even have the level of data quality to support AI in this space?
The experiments keep changing shape, so comparison itself becomes a moving target. But I agree that the grounding questions still sound very product-minded: what value is actually being created, at what cost, and can this be maintained without quietly becoming operational debt?
And the data quality question may be the least glamorous and most important one. Before asking whether AI can improve the system, we may need to ask whether the system has produced data trustworthy enough to be improved by AI at all.
I’ve only read snippets of The Metamorphosis, but meant to also read The Trial, The Castle, and perhaps his diary collections (although these days I’d be afraid of going down too deeply into psychological rabbit holes, should have read these when I was too happy 13 years ago)
The part about AI work happening outside the normal roadmap process stood out to me. I've seen plenty of excitement around the experiment itself.
Figuring out ownership and what success actually looks like is usually the difficult part i think.
Reading your article made me realize how the orgs still operate, and I felt it for you in that role so much, Judy! I’m glad you found it useful and took the lesson learned, but the bigger question is, did you get promoted after all this work?
No, I got laid off along with my whole team.
I don’t think it was a reflection on my skills personally, but a general belief that engineers could do our jobs or perhaps fewer people who are centralized could do enough.
One of the hardest parts about being laid off right now is keeping up and continually closing the skill gap without the work & community at work.
Another reason why I appreciate Substack so much.
Ah, I’m so sorry! The way you’ve translated that experience into such clear thinking really comes through in this piece, and it’s what the enterprise AI need. I’m glad Substack is giving you a new kind of community around this work.
Truly appreciate that. I really like writing as much as possible without AI. And sharing my experience like this gives me that opportunity.
Judy when you told me this made you think of Kafka I thought you were being dramatic and then I read the part about spending 20 hours on a 1-page email status only to achieve "nirvana" and then never getting feedback again 😂 that's a Kafka short story, literally.
I forget how much of enterprise AI adoption is just all about surviving the organization around the AI....
I hope to never be overly dramatic, but I will be guilty as charged at times.
It is quite something to navigate new things in old systems.
The hardest parts for me are when you disappear between authority boundaries and feel organizationally disconnected to the work, even though you’re actually doing the work.
I think about Kafka working at that insurance company in 1920 probably feeling the same way sometimes.
Oh I bet he would!
This is such a useful distinction between experimentation and actual organizational work.
The PoC vs. Pilot framing is especially clear. A lot of AI work gets praised while it is still safely in “various stages of maybe,” but the harder question is whether it can survive contact with ownership, workflow, adoption, and governance.
That is where the real story begins.
And the hard thing is the shape of experiments constantly evolves so knowing how to compare them and think about them seems to get more complex, but I think the grounding factors are still the good product managers thinking about value & cost - maintainability.
Or the trickiest one yet, do we even have the level of data quality to support AI in this space?
Yes — that may be the hardest part now.
The experiments keep changing shape, so comparison itself becomes a moving target. But I agree that the grounding questions still sound very product-minded: what value is actually being created, at what cost, and can this be maintained without quietly becoming operational debt?
And the data quality question may be the least glamorous and most important one. Before asking whether AI can improve the system, we may need to ask whether the system has produced data trustworthy enough to be improved by AI at all.
Nice to see you converged on a workflow from POC to production, after a Kafkaesque journey.
You made me realize I have not read enough Kafka!
It’s interesting to think he joined the Workers’ Accident Insurance Institute in 1908 and witnessed the fog of bureaucracy seeping into systems.
Curious, if you have any recommendations on what I should read from him.
I’ve only read snippets of The Metamorphosis, but meant to also read The Trial, The Castle, and perhaps his diary collections (although these days I’d be afraid of going down too deeply into psychological rabbit holes, should have read these when I was too happy 13 years ago)
Exactly. I found metamorphosis terrifying but I am going to try to do the castle and the trial.