Salesforce announced this morning that it is bringing generative AI to its data visualization platform, Tableau. The goal is to make data accessible to everyone. But we are also working on something that could be even more revolutionary.
“Only 30% of organizations use data to make decisions,” Tableau’s Chief Product Officer Francois Ajenstat said at a convention in Las Vegas this morning. “Data is still too hard. Not everyone has access to it.”
The new version of Tableau will leverage generative AI to enable people who don’t normally get data to ask questions and gain insights using natural language, the company said. Called Tableau GPT, the software automatically creates visualizations and suggests potentially relevant questions and queries. A new “experience” within Tableau is called Pulse, which provides instant insights including charts and graphs for users who would not normally use the software.
The announcement comes a week after Salesforce announced Slack GPT, which integrates conversational AI into its popular work communication and coordination platform. Competitors such as Microsoft have added his ChatGPT to their Microsoft Dynamics 365 customer relationship management tools, announcing support for it in March of this year. Zoho CRM also includes an AI assistant that analyzes data, generates charts and detects anomalies.
Pedro Arellano, SVP and general manager at Tableau, says all this innovation is happening at an incredible speed.
“The pace of innovation is not the normal pace,” Arellano told me this morning. “Usually when you see an invention, you have time to absorb it.”
Not anymore.
ChatGPT launched just four months ago on November 30, 2022. Since then, it has taken the world by storm, boosting the market share and stock price of investor Microsoft’s search engine search engine, and forcing software companies to innovate in generative AI and large-scale language models, creating a slew of competitors. .
A core innovation in the new Tableau technology is aimed at accelerating adoption of data insights across organizations by reducing barriers to entry, said Arellano.
But I don’t think humans are left out of the loop.
“Learning doesn’t come naturally,” he said. “Adding a layer of business context and insight requires human involvement.”
Tableau Pulse, a version of the company’s software designed for non-data-natives, won’t be available until a pilot release in the fall of 2023, but Salesforce already has something that could be more innovative. is being tested. A broad and “owned” AI engine for virtually anything.
GPT technology (Generative Pre-trained Transformers) was popularized by OpenAI’s GPT-4. But companies like Samsung have already banned engineers from using it because the data shared to ChatGPT resides on servers outside the company’s control. Amazon, like Goldman Sachs, has issued a warning.
Therefore, you need enterprise-specific AI tools. Perhaps it is in some way owned, developed, trained, and maintained by a company, even though it was ultimately built by a vendor like Salesforce. And these tools must learn, grow, and remember company-specific data, insights, trends, opportunities, and challenges, not just query by query, but month by month, quarter by quarter.
These AI tools are ultimately not just for point solutions in marketing such as text generation and image creation. And they’re not just for data analysis. Rather, they end up seeing and influencing just about everything within the company, knowing the company’s deepest secrets, and developing models that can be shared with different people at different levels within (and outside) the organization. I have. And you need ironclad rules and technology firewalls to keep all your data and insights safe within your organization.
Salesforce is working on that sort of thing, Arellano told me this morning.
“We already have a prototype,” he said.
Such an AI can tell you what products to launch, where to sell them, how common events in the world affect sales, and what marketing efforts are failing. and where companies should invest more or less. Ultimately, in an AI-first economy, companies with better AI will outperform those with weaker AI.
Building this out is a massive undertaking, and potentially far beyond what Salesforce has already started. And, as Alellano said, AI needs to be able to access the entire internet, and that comes with its own set of problems. But for Salesforce, this is an interesting possibility. That’s because the company already has key capabilities in social listening and marketing technology, enterprise sales, customers, and internal operations that can feed vast amounts of data to such AI.
Few companies can compete on this richness of functionality, with Microsoft being the main competitor, and perhaps IBM, SAP, Oracle, and a few others.
As Arellano emphasizes, a major challenge is determining which external data is accurate and reliable.
If the company could work it out on some level, the engineers behind OpenAI had to do just that to create something that did it right. many This opens up a world of possibilities not just for AI-first companies, but perhaps AI-driven companies as well.
For now, this is just an idea and a prototype, Tableau is cautious about GPT technology and corporate data, Tableau’s GTP part is on the side of what to do with the data, and Tableau’s part is on Tableau’s part. says. GPT on the data side here.
“GPT is not a deterministic analytics engine,” Arellano told me. “Use without humans increases the risk of hallucinations and erroneous results.”
follow me twitter Or LinkedIn. check out You can find my website and other works here.
