ChatGPT at Gartner Analytics Summit

The Year of ChatGPT

Before we get to ChatGPT, the opening keynote of the event made it clear that "Skills and staff shortages" is the main roadblock in data initiatives. Emphasis was added to "technology is not the solution to our most significant challenges" but "technology is foundational to everything we do, technology alone is not enough", I couldn't agree more.

Therefore, with skill shortages, there is a real need to deliver more value at the personal, team and organization layer. Keyword value, which we should be able to understand and communicate.

Gartner proposes communicating value through the Enterprise Value Equation consisting of mastering business language, building business cases, and measuring value.

Interestingly, an example of what NOT to do is to invest in belief cases based on cool technologies. Think Hadoop back many years ago and ChatGPT in 2023 -- "You will get funding for ChatGPT", cause everyone wants shiny new things but perhaps think twice.

So it is definitely a bad idea to fund ChatGPT projects just because they are shiny, don't be like Grogu.

One way to create value is by spreading a data culture across the organization. Remember, the main roadblock is a shortage of skills. The model Gartner proposes is opening a "Data Franchise", close and perhaps equivalent to the "Data Mesh" concept others reference.

Alright, so let's now come back full-circle. In order to empower business users with data & analytics we need to promote a data culture while also empowering them with the right self-service analytics tool.

We need self-service analytics tools for business users. Hard to ignore ChatGPT in this context, right? At Hal9 , we have found ChatGPT-like technologies applied to data analytics problems create significant business value. For instance, we saw data engagement grow 10x but that's for another post.

ChatGPT for Enterprise

This section highlights key points from "The Enterprise Implications of ChatGPT and Generative AI" by Arun Chandrasekaran . Starting with the takeaway:

Foundation models represent a huge step change in the field of AI, due to their massive pretraining, which makes them effective at few-show and zero-shot learning, enabling them to be versatile

Paraphrasing Arun, foundation models (e.g. ChatGPT) are a disruptive technology for the enterprise that we should all start exploring.

However, you do NOT want to train your own Large Language Model (LLM) -- leave that to others specialized in training them (e.g. OpenAI).

Alright, great! So all we need to do is call OpenAI's GPT API, right? Not so fast, there is more to this than to YOLO your data infrastructure with API calls:

We know this first hand at Hal9 , users will submit very random and sometimes even inappropriate prompts, we need systems to monitor those for sure. Even if they are properly used, how do we know the results are actually accurate? That should also be monitored and fixed. Finally, we also want to improve systems over time and collaborate beyond everyone generating the same results over and over again. That's quite a good amount of work and new tools will be needed.

This not to sound pessimistic, but just to point out this is just the beginning of a very exciting journey.

ChatGPT in Analytics

Finally, there were a ton of demos from different analytics provider that integrate ChatGPT-like functionality, I'll call out a few of them from their presentations or vendor booth chats:

• Microsoft announced Copilot for Excel and with the Build conference around the corner, I wouldn't be surprised to see a similar integration announced for PowerBI.
• Tableau announced a ChatGPT App for Slack.
• ThoughtSpot announced the ability to perform analytics using GPT with ThoughtSpot Sage.

I'm missing many more in the analytics space and even business apps like HubSpot with ChatSpot, but going beyond analytics is beyond the scope of this post.

My take here is that almost every software vendor will explore integrations with ChatGPT this year, full stop.

So where does that leave brand new companies like Hal9 who specialize in providing analytics with ChatGPT-like interfaces? At a great place! We believe best-of-breed solutions, as usual, will find innovators and early adopters to push the boundaries of what's possible together.

Thanks for reading this post, I hope it was useful. If you have 5 minutes, please take the hal9.com/survey survey to help us learn more about you.

All the best, Javier