Incorta NoLimits Recap

Michael Ross, June 6, 2024

Last week, Incorta hosted their NoLimits event in San Francisco. The full day event contained sessions on the current trends in the generative AI space, as well as Incorta customer stories and product demos. Incorta’s new Operational GenAI offering was announced by Incorta CEO Osama Elkady early in the day, setting the tone for emphasizing Incorta’s commitment to helping customers explore the benefits of GenAI in their data analytics journey.

A common theme throughout the day was the need for AI solutions to have access to granular data in a timely manner. This lines up nicely with Incorta’s typical deployment strategy of loading data into the platform in the same structure as it exists in source systems, effectively creating a digital twin of the source system, and refreshing that data on an hourly basis. Summarized and stale data will only lead to frustration with GenAI solutions as inaccurate results will quickly erode trust in the content delivered to users. Coupling the GenAI capabilities with Incorta’s fast data ingestion and dashboarding capabilities means customers will be able to see value from their investment in the Incorta data platform in days and weeks rather than months.

Most Incorta users will end up consuming the results of GenAI in the form of interactions with the Incorta Copilot. The Copilot chat bot interface will allow users to ask questions about their data, and Copilot will generate insights related to those questions. There was a demo of this functionality that walked through a financial analysis example where the CFO wanted to explore operating income margin, not just for their own company, but compared to competitors as well. In a series of questions to Copilot, the CFO was able to determine how their company performed over the past couple years based on data from their own company’s ERP data. Unstructured data from other public companies’ 10-K and 10-Q reports was then used as a point of reference to examine if the CFO’s company’s drop in operating income margin was specific to their company or was related to a broader market downturn. All of this played out in a conversational manner with Copilot providing generated textual context in response to the questions the CFO was interested in exploring. It’s not too hard to envision a point in the future when large Excel reports with dozens of columns are replaced with a conversational AI assistant that will highlight areas that users should be concerned about.

Incorta Copilot can also be used to generate SQL code that can be used to enrich the data sets stored in Incorta. As someone that is used to writing SQL, I’ve been impressed with the speed and quality of the code generated by Copilot. There have been examples where I would’ve had to google the proper SQL syntax to write my code properly, but Copilot was able to take my plain English request for what I wanted and turn it into functioning code in seconds. Even if I knew the exact code I wanted to write word for word, it would’ve still been faster to have Copilot generate it for me. This plays into the theme of AI serving as a helper to workers; not a replacement. Beyond code generation, there’s even support for generating documentation for existing code, and the ability to get performance improvement recommendations for code using Copilot.

PMsquare was offering demos of the Incorta Copilot functionality at the event. Even though the event is over, we’re still offering to take your company’s data and provide a demo of Incorta Copilot using data you are familiar with. Contact us to learn more about this offer.

Incorta Copilot Demo Flyer

Learn about SaaS Anywhere for Incorta

Learn about IncortaOne

 The event also featured sessions hosted by the two companies Incorta is partnering with to make GenAI a reality in the platform, Vectara and aiXplain (pronounced explain). Vectara’s RAG (Retrieval Augmented Generation) will be used to evaluate the generated responses from Copilot to limit hallucinations and inaccurate results. The agent framework and model library that aiXplain provides will ensure that the latest and greatest models are available for Incorta customers to use and test against. In a world where LLMs (Large Language Models) are evolving every day, it will be critical to evaluate how the models perform on customer data. In general AI trends, we’ve seen some models handle certain tasks better than others, and there’s even variance where release to release the models can fluctuate in accuracy. Incorta demoed an example where a user could switch the underlying model from a drop down to see how each model reacted to a certain request. The model that generates the best results for that task could be selected after testing.

There were a few sessions related to AI trends in the enterprise space, including the keynote from Sol Rashidi. She shared some of her experience as a C-Suite executive as it relates to AI. The key insight she shared around not using business value as the sole factor in deciding which AI initiatives to undertake was interesting to me. There may be AI use cases that could deliver tremendous value to a company, but the underlying data quality and effort to prepare the data to be usable could take so long that the initiative will not yield tangible results for a long time, if ever. If there’s an opportunity to generate $1M in additional revenue through an AI project, but it costs $2M to deliver, it’s obviously not going to be a worthwhile endeavor. Maybe there’s a more niche project that could generate $100K in revenue, but only costs $30K to deliver. That project is still profitable and can be used as a starting point for evolving AI’s footprint at the company.

Arun Chandrasekaran, VP Analyst at Gartner, had the anecdote that resonated with me the most throughout the day. He obviously gets exposed to a lot of new products in the AI space through his research at Gartner. His sentiment around AI changes day-to-day as some days he sees the high-profile failures of AI (like Google search recommending adding glue to pizza to prevent the cheese from sliding off the crust) which causes him to doubt the extreme hype surrounding AI. But he’s also seen the cutting-edge use cases that companies are developing, which make him think we are just scratching the surface of the potential for AI. I feel a similar way about AI at this point. We are still in a nascent stage of the hype cycle around AI. Things are evolving so quickly that we can only guess where we will end up. No matter what, I think we are in for an interesting journey with high highs and low lows.

Explore everything Incorta has to offer. Contact us today to set up your own Incorta environment in minutes!