Introducing Analyst Agent: Deploy custom AI agents for self-service analytics in minutes | Fabi.ai Fabi.ai Analyst Agent is the easiest way for data teams to create, configure and share AI data agents designed to answer self-service analytics requests. Analyst Agents can be built in minutes and hel...

Read the full story on how we're rethinking self-service analytics on our blog: www.fabi.ai/blog/introdu...

Data teams get complete control:

Configure agents by domain (marketing, product, CS)
Real-time data syncs
Transparent access to code
Enterprise-grade security

The best part? These aren't just text-to-SQL bots:

Run regression models
Calculate propensity scores
Handle complex analysis
All powered by Python & LLMs, no code required 💪

Our approach: Specialized AI agents that work within boundaries set by data teams.

⚡Deploy in minutes, not months
😌Work with curated datasets
☑️Multi-step validation
✨Nearly 100% accuracy

The old way:

Generic AI attempting all analytics
Perfect semantic layers required
Limited to basic SQL queries

The new way:

Focused agents with defined boundaries
Works with curated datasets
Multi-step problem solving
Python-powered analysis

Excited to announce Analyst Agent from Fabi.ai - a new approach to self-service analytics with AI that actually works.

We're live on Product Hunt today - would love your feedback!

🚀 www.producthunt.com/posts/fabi-a...

TIL: There's a NYC public squirrel census with available data.

www.thesquirrelcensus.com

You're welcome.

Congrats Taylor!

Playing around with MotherDuck’s prompt() for something I’m working on and man it’s cool. Haven’t played much with Snowflake Cortex but may have to check it out and compare.

Transformer Explainer: LLM Transformer Model Visually Explained An interactive visualization tool showing you how transformer models work in large language models (LLM) like GPT.

Came across this really cool tool that visually shows how transformers work built by a team at Georgia Tech.

If anything, this is an amazing implementation of visualization and a great use of Sankey diagrams.

poloclub.github.io/transformer-...

More individuals should be encouraged to remain on the the individual contributor route.

Conversely, more organization should create titles and comp bands that make being a world-class individual contributor as lucrative as being a manager.

Quick stocking stuffer 🎁

Conditional table formatting just got a whole new look. No more fiddling around with dropdowns, just ask the AI exactly what you want and let it do the work.

I kind of assumed the Irish would just buy beer to drink at home. But perhaps they appreciate the on-tap flavor more than Americans.

This is a case where the data feels wrong but… maybe it’s right?

Oh let’s go. Didn’t know this was going around. I want to get more involved. Following

#BSdata I want your opinion: if you have a file with 4K rows that has US and Canadian zip codes in one field and a numeric field, what would be your tool of choice to plot this on a map as quickly as possible?

Requirement: I need to be able to share it as a URL and the map needs to be interactive.

I never realized how much time was spent reading contracts and fine print as a founder.

#1 today 🎉

Every week or so I get a reminder that I live in a tech bubble. It feels like everyone I know is using the latest and greatest tech and data tool, when in reality most of the country (and world) still operates on spreadsheets and custom scripts floating around.

#2 on Microlaunch

Go check us out and leave a review!

I mean… most of the time. Throw in a good pivot and you’re golden.

That’s actually a really interesting read.

What type of front end are you building? Like chat bots?

100%

On a related note, I’m waiting for the day where it’s understood what data is used for training. Because if they’re crawling web data, there’s no way they’re supervising everything, so if I write a blog saying that “Acme sells the best widgets in the entire world” does the AI use that?

This makes perfect sense and I’m actually here for it. If they can nail this: 1. It actually completely upheaves google’s existing model which is good for everyone (arguably even good for G) 2. For businesses it can create a hyper-personalized targeting mechanism which is also better for the user.

The store is a Dead Sea. Prediction: it will be discontinued quietly over the next 12-18 months.

Too many developers tend to think issues like this are a user skill issue.

Love a good sankey diagram. I have a soft spot for them.

Ha well then…

You pay for OpenAI but not Anthropic? That’s my situation, I always debate switching.