AI Analytics
Since the likes of ChatGPT or Midjourney made their debut, there has been a lot of buzz around the next frontier of generative AI: Generative Analytics. Business leaders and analysts alike have envisioned a world where they can simply ask questions of their enterprise data warehouses to have charts, analysis, and answers created for them by AI. The benefit to organizations is to open up analytics to everyone, of course, as almost every employee in the modern enterprise should be data driven. However, not every employee needs to be a data scientist. Rather than burdening resource constrained analytics teams, leaders are envisioning a world in which ad hoc analysis can be facilitated by modern conversational AI.

Not only is this dream now a reality, it’s about to become table stakes to play at at the enterprise level. And fast.
Thank you for your interest, and enjoy the whitepaper!
Feel free to reach out to with any questions.
Thank you for your interest!
Continue Reading
Something went wrong while submitting your email.
Throughout the last year, we’ve spoken with hundreds of data leaders from governments and the Fortune 500. All of them are thinking about AI. Most are piloting new technologies. Many homegrown projects have already unraveled or never launched. As we explain here, most of these are going to fail.

With that said, a new wave of technology is coming and those organizations that harness it correctly will have a distinct market advantage. Those that remain stagnant will absolutely be playing catch up.

The devil, of course, is in the details. Here we discuss what such a world looks like, why you should care, and where where the challenges lie.
A Whole New World
So what does Generative Analytics look like tangibly? In the beginning, chatbots and co-pilots for everything data related. Want to query your database? Just ask your SQL copilot to generate a query for you. Want to create a chart? Ask your BI tool’s chatbot to generate a new visualization for you.

The problem with these surface-level approaches is that they conform to the old paradigms of analytics and are only a marginal improvement on existing processes for existing users of such tools.

The next phase of analytics is to teach the enterprise data assets across current platforms to AI, converting them into a conversational format so that all employees (with correct credentials, of course) can access fragmented insights and knowledge more naturally. This is where organizations will begin to draw value from the 97% of enterprise data that sits unused (cue platitude about only seeing the tip of the iceberg).
This is just the beginning, however. Once AI is natively able to understand enterprise data, nuances to particular tables, executive’s concerns, and the drivers of the business, the natural next step is to have AI proactively surface insights you never thought to ask. That is, AI-assisted insights generated in real time and at the speed of relevancy. This has long been the holy grail for most executive teams, but the technology is nearly there.

Prudent organizations should be laying the groundwork to get their enterprise data AI-ready without getting locked into the old way.
Why Care?
Executives have been dreaming of the day where “have the report on my desk by Monday” becomes instantaneous. Relatedly, analytics teams are tired of fielding the impromptu asks for for every stakeholder’s “absolutely pressing” question instead of focusing on deeper projects. From our interviews, analysts spend anywhere from 25-40% of their time fielding ad hoc questions. For a team of 10, that costs up to half a million dollars a year at the current market rate.

With the right security and governance in place (more on this below), the universe of self-service possibility has just grown tenfold. In fact, 67% of today’s managers still aren’t comfortable using existing tooling for analytics. The data revolution came for some but not for most. With the assistance of more intuitive, natural-language data interfaces, there is no excuse not to be data driven. And those that rise to the occasion will win.

What this means is that organizations can begin to address the unknown unknowns. These are the questions that executives and stakeholders choose not to ask because analytics resources are limited or because they may not line up with a given project or business priority. Today’s analytics serve to answer questions versus help you ask them. Most analytics projects are by definition reactionary, where the questions are to the effect of: “help me understand why X product category is lagging this quarter,” “what are the 'best' profiles of the customers I target,” “where can we find cost savings in our supply chain?”

When the barrier to inquiry is lowered, teams can begin to explore data rather than use it in service of what they are already partially aware of. This opens up an entire universe of insights currently left undiscovered. Current ad hoc investigation and self-service tooling attempts to achieve this, but fall woefully short. Executives instead choose to self-filter and most of the value remains below the proverbial surface.
The Challenge
Using AI to draw insights from one data source like an Excel file is hard but doable. Teaching AI the entirety of the enterprise data stack, where the right information is, and the nuance of how said information is to be used, is much, much more challenging. Logistical data questions of scale, governance, security, and integrity also have to be considered and respected. And that’s before you even to the AI part.

Generative AI introduces a new set of challenges while also exacerbating the pitfalls present in more traditional forms of AI. To start, generative AI has no concept of facts when generating content, which means they have a tendency to “hallucinate” or generate plausible sounding but entirely false information. When using AI in analytics, its absolutely crucial to build robust and battle tested frameworks to avoid misleading or false insights. Our team comes from the defense and intelligence world where such falsehoods could mean the difference between life and death, and our products carry forth the learnings we’ve gathered along the way. Generating insights is trivial. Assuring they are factual and useful is the real challenge.

Additionally, as with more rudimentary forms of AI, even correct insights can be misleading if they are not explainable. Which tables and columns were selected or filtered to generate a particular insight? Are there data quality or integrity issues I should be aware of when interpreting a chart? Where does the data even come from and am I using the right thing? Am I using the same assumptions everyone else looking at the same data are?

As with the classic example of the survivorship bias, today’s AI is entirely unaware where it may be sampling on the dependent variable or introducing other forms of bias.
At Clarative, we build our AI analytics systems to consider all of the questions, surface potential sources of bias, and loop in human intuition at the right time. Reach out if you’re interested in hearing more.

Our team consists of data experts from Palantir, Google, and Benchling with years of experience building data and AI systems for the Intelligence Community, Department of Defense, healthcare, leading financial institutions, and the world’s largest enterprises. If you’re considering bringing AI analytics into your organization, we provide free consultations with no commitment. Schedule time to talk with us here or drop us a line below.
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.