Abhi Sivasailam

That ...wasn't my take 😅 Actually, the markets are smaller, GTM harder, and the most and NFC more entrenched, and this all says much more about the market weakness of the acquired than the strength/optimism/momentum of the acquirers

Not on my bingo card.

Dagster.

Lol it was actually a last-second re-add that I had removed. Anyway, I'm generally interested in AI enabled first-/middle-mile data work. Pretty uninterested in all the last-mile stuff like text to SQL. Y'all doing this stuff?

Does anyone have a cool demo or workflow of AI codegen for building and managing dbt projects?

No, I pay for Claude, I just use it a *lot*.

The cruelest, most painful words in the English language.

Would argue it's a lot easier to downshift from BigCo context than to extrapolate out from SmallCo. Basically every org has more to learn from AMZN than [SmallCo]

Happy Thanksgiving. Endlessly grateful for all the brilliant, thoughtful, passionate nerds that make #databs great.

At least, this gets you very far. And especially when you can tie those queries to context from the BI tool and get that "semantic" context from charts, dashboards, labels, etc..

Also, the notion that a "semantic layer" is necessary for text to SQL and AI affordances makes little sense to me (outside of, yes, making metrics easier to query)

bsky.app/profile/abhi...

I just think the "semantic" part is overblown. Call a "metrics layer" what it is.

Idk what got into Chris tonight but I'm here for it. Go off king.

Also will add that Idiosyncratic "levers" usually "hang on" the Common. Or in the parlance I use when talking about metric trees, the Idiosyncratic are usually "Influences" and the Common are usually "Components".

Some companies win through operational excellence; some win through innovation.

IME (read: anecdata), most winners do the former category, but most of the biggest winners do the latter.

I'd ditch the FE

(The right answers to (2) are obviously "Neither: Fritz's" or "frozen custard is overrated: Oberweiss" 😌)

Let's just get the rest of this out of the way:
1. Imos or CWP?
2. Ted Drews or Andy's?

lol I thought the same thing when I saw this a few weeks ago and keep forgetting to ask. I'm West County Rockwood :)

I keep forgetting to ask where you're from in MO (and if stl area, "where did you go to HS" 😀)

1. Making all those things you mention above work well (causal model, operating mechanisms, etc).
2. True, proactive exploration/"insight"-harvesting.
3. Shifting up the value chain and looking more like a profit center: customer-facing data work, data commercialization/monetization, etc...

More basic than any of those things is just having data models and metrics that are stable, reliable, complete, and expressive enough for a data team (and a would-be Staff DA) to not have to be in constant reactive mode. Rare, actually.

There _are_ staff DAs, there just aren't that many of them. In no small part, this is bc maximal Staff DA leverage requires an organizational data maturity that very few cos have realized.

Most certainly. AI/LLMs have lowered the friction for last mile analysis, and data mgmt tech has lowered the friction for managing every link in the data value chain => we'll probably see the roles consolidate into more data generalists

👋🏽

@spite.vc continuing to raise the 🐐 bar.

(He's going to be even more insufferable now isn't he?)

I want to heckle.

Only recorded or is it live somewhere?