Prukalpa Sankar said that over $250 billion was spent on data and analytics over the last decade and that about 75% of those projects failed.
“over $250 billion was spent on data and analytics. And 75% of those projects used to fail.”
Person
Prukalpa Sankar
1 episodes · 19 quoted claims
Sankar said Atlan gives the gap between AI pilots and AI production a name, although the transcript spelling of it is garbled.
“We call it the AI wild gusm.”
Sankar relayed a CIO's requirement that an HR chatbot be the only AI allowed to use payroll data.
“we are deploying an HR chatbot that can use my payroll data, nothing else is allowed to use my payroll data.”
Sankar named the EU AI Act as an example of AI regulation that enterprises must manage policies around.
“AI governance regulation, EU AI Act.”
Sankar said Palantir uses the term ontology, and others use knowledge graphs, for ideas related to a context layer.
“Palantir talks about this idea of an ontology. We've talked about ideas of knowledge graphs.”
Sankar said sales teams define top ten customers by revenue while customer success teams define them by ratings or NPS.
“if the sales team asks for top ten, it means top ten customers by revenue. If the success team asks for top ten, it means top ten customers by ratings or NPS.”
Sankar said one Atlan customer spent about fifteen months trying to get an AI system into production for its CEO.
“We had a customer that was trying to get AI into production for their C E O for about fifteen months”
Sankar said that 45 years ago an average S&P 500 company stayed on the index about 75 years.
“Forty five years ago, the average company on the S P 500 used to stay there for roughly seventy five years.”
Sankar said that in the last decade the average S&P 500 company stays on the index about 15 years.
“In the last decade, the average company in the S P 500 stays there for fifteen years.”
Atlan held a customer conference featuring speakers from Workday, Mastercard and Virgin Media O2.
“we just got off our customer conference. And we had some phenomenal customers, customers like Workday and Mastercard and Virgin Media O two”
Sankar said Workday's chief data officer stated that its well-governed business intelligence data could not be used by AI.
“their perfectly governed BI data wasn't consumable by AI.”
Sankar said Workday improved AI accuracy by over five times after moving to a context layer approach.
“they were able to improve accuracy by over five times, which is the difference between an AI and pilot and AI in production.”
Sankar said Virgin Media O2 onboarded over 6,000 employees to its data platform.
“VM O two talked about how they onboarded over 6,000 employees”
Sankar said Mastercard's chief data officer described an agentic governance framework and that over one third of its products are AI-first.
“they're building an agentic governance framework for, uh, you know, Very cool, as they have over, uh, one thirds of their products are AI first products today.”
Sankar quoted Mastercard's chief data officer as saying the company is actually a data company rather than only a finance company.
“you would think we're a we're a finance company, but we're actually a data company.”
Sankar said Atlan can bootstrap an evaluation set by pulling metadata from business intelligence tools and from Slack questions.
“we can pull metadata from your Slack where people are asking you questions.”
Sankar said customers saw accuracy on their test set rise ten times within a month of starting with Atlan.
“within a month, their accuracy has gone up ten times on that test set.”
Sankar said metadata architecture should be Iceberg-native, open, interoperable and ready for real-time workloads.
“the metadata architecture needs to be iceberg native, and it needs to be open and interoperable”
Sankar said that attempts to build a perfect 'golden' semantic layer, which were hyped in 2020, tend to fail because they lose business backing and funding.
“I think golden approaches fail.”