In 2026, when you can take this sage like insight to prediction markets/stock market, I'll never understand the performative social media declarations w/o receipts.
Post receipts, or don't post at all.
@RandallKanna@jdnoc Just think about what that means for your employer or client.
I LOVE driving cars I grew up with, especially 90s sports cars.
I don't drive them daily because I have shit to do and FSD allows me to respond to emails on the road without threatening other drivers around me.
@RandallKanna Are you a consultant or an employee?
As a data & AI consultant, our ethics/responsibility, is to solve problems within constrained budgets and timelines.
It would be unethical to NOT used best in breed tools to deliver projects stronger, better, and faster.
@ovinophile@jahirsheikh8 No, it's really not. Experienced developers are still vital, they're just vastly reducing the time spent on mundane coding, but requirements gather, stakeholder alignment, oversight, etc., still incredibly important.
@acadictive I completely pivoted from being a small business owner into data & AI consulting... but I was in sales and built out Excel/Power BI analytics for small businesses for years as a side hustle so it wasn't like walking into an alien world.
LOL, it's EXACTLY the same thing.
Librarians used to utilize a far more manual system to keep inventory of books in a library.
The computers drastically reduced that burden.
Engineers used to write code manually end to end, and now much of that mundane coding is done by AI with the engineer servicing as an approval/validator.
@aneesmerchant 15 departments each with their own dashboards is exactly why it worked. Somebody finally had to agree on what the data means before the agents could use it, and that's the part most orgs have neglected for years.
@FisherNimrod This is the part most teams skip. Mapping what people actually do and what data they use is basically where a good semantic layer comes from, and if that part is sloppy the agent is just confidently wrong.
@jim_dowling@kirsten_lum_ I'd add a third one. The semantic layer is where the data gets translated into business language, so the agent answers with the same definitions the business actually uses. We run MCP between the LLM and the data and honestly that part matters more than anything else for us.
@markargyelan_ The build part sure, but somebody still has to know if the numbers are right. We've had real wins with AI on reports, but only when a senior data person is checking the model and the definitions behind it.
@PowerBITips Honestly never. The business moves too fast for data to ever be 100% ready. What's worked for us is getting the core stuff really clean and putting a solid semantic layer on top so the AI is speaking business language, not guessing at column names.
@dschewchenko Dad of three here too, a 1 year old and two girls, 3 and 6. Whatever gets done after bedtime is a bonus. Small consistent stuff beats waiting for big blocks of time that never come.
@Nehakathare It was the first field to experience head count reduction bc AI, but now the most difficult field to be a high performerโฆ bc AI has made connecting with customers and attribution far more difficult.
Well, it's raining shit across the American mid-market.
I'm happier that AI has forced mid-market leadership to the table to deal with their data culture, more than actually successful AI deployments.
Can't tell you the number of businesses that still balk at low six figure investments that bring them a single, trusted source for analytics and AI.
@MandelDuck I've got 3 under 7. A 1 year old boy and two girls, 3 and 6, plus a day job in data. Some weeks the win is just getting through work and bedtime without dropping a ball.