And, thank you for having me in the women tech chief in tech summit.

Appreciate having you with us. We can see your slides, and the stage is all yours.

Thank you. Today's session is going to tackle a challenge top of mind for every executive. How to drive real business value from AI by starting with the foundation, AI ready data. Hello, everyone. A quick introduction on myself. As Anna made, I'm the CTO of Precisely, Tendio Chu. And for those of you who are not familiar with Precisely, we empower organizations around the world, including the 93 of the Fortune 100 to build trust in their data. We believe data is trustworthy if it is accurate, consistent, and contextualized. We call this data integrity. My perspective today comes from working with global enterprises and, as they navigate through their data journeys and real world AI deployments.

Trusted data is more important than ever to power your business as well as create customer experiences, driving operations, and enable timely confident decisions. But this happens only when it's fit for a purpose. In a survey conducted in partnership with the center of business analytics at Drexel University, 76% of respondents named data driven decision making as top priority. But here is the challenge. While data driven decisions are goal, 67% don't complete the trust, the data their organizations uses to make those decisions. This is the trust gap when confidence in strategy out places confidence in the data behind it. And it's one of the biggest barriers to realizing AI's full value. Despite that trust gap, AI adoption is moving fast. According to Stanford index, global private investment in AI reached 252,000,000,000 last year.

That's bigger than the entire GDP of some countries. 78% of companies report using AI in more than one business function, from support automation to AI assistance, knowledge systems, and increasingly AI agents. While agent based systems are gaining traction, the foundation remains the same. None of these initiatives succeed without trusted AI ready data. And here is the reality. Only 12% of organizations say their data is AI ready despite that growing AI adoption. That means the majority are building AI on top of their siloed, inconsistent, or incomplete data foundations. Let's pause for a Let me ask, is your organization's data AI ready? Without AI ready data, your business is exposed to a wealth of potential harmful impacts, especially with generative AI.

In fact, Gartner predicts that through 2026, organizations will abandon 60% of AI projects due to lack of AI ready data. For example, without access to all relevant or critical data your AI needs, you may deliver outcomes that are ageist, sexist, racist, basically biased. You may be familiar with the tutoring company that automatically rejected female applicants aged 55 and older, resulting in a major settlement and a six figure fine. Or the well known example in loan approval models that favored men over women simply because historical data reflected biased outcomes. As for accuracy, several lawyers submitted court filings with case citations generated by CHEP GPT only to discover the cases didn't exist. And relevance matters even more in the context of generative AI. For example, the number of recognized countries varies by source.

The United Nation recognizes 193 member states, while international standards organization lists 249 codes, including the territories. Both are correct within their context, but using them interchangeably can cause confusion. Trade data may require international standards organization codes, while diplomatic analysis would use UN membership. Without context, your model may get the right answer that's completely misaligned with business needs. Bias, inaccuracy, or irrelevance data problems. And they scale fast in the world of Gen AI. So how do we close this trust gap and set up the AI for success? It starts with data. Let's walk through three critical considerations for AI ready data, integration, quality and governance, and enrichment. Because enterprise data is generated and stored in very various legacy systems, not all relevant data may be easily accessible or usable by AI systems.

If you are missing a demographic for your customer profiles or missing sales information from a specific geography, your outcomes will not be reflective of the true state of your business. To minimize bias through integration, organizations should, one, integrate data across silos and multiple

Hi, Tindu. Can you hear me? It looks like you you're muted. We cannot hear you. We just kinda stopped hearing you for a No. I don't think so. I'll try to just unmute you. Hear me now? Yes. Yes. All good. Please continue.

Start from here?

Let's start maybe from just one, two sentences that you just shared.

K. If you are missing a demographic for your customer profiles or missing sales information from a specific region, your outcomes will not be reflective of the true state of your business. To minimize bias through integration, organizations should, one, integrate data across silos and several platforms, and now even integrate unstructured data using generative AI to ensure AI learns from a more complete representative view of your business. drive intelligence from metadata to detect potential bias. And ensure data is fresh and updated so insights reflect your current operating reality, not last quarter's view. This ensures all relevant data is available where AI needs it in a comprehensive, complete, and timely fashion. Even when data is accessible, it's often incomplete, inconsistent, and poorly governed, making it hard to trust and even harder to scale responsibly.

This is where data quality and governance come in as a strategic enabler. To increase trust, focus on democratizing access to data across the organization, making sure the training data is accurate, complete, standardized, and free of duplicates. Automate quality checks and monitor your inference data continuously to track the metrics. And you can use AI to automate creation of data pipelines. And most importantly, protect and govern govern both data and AI models, ensuring transparency, lineage, and compliance. Proactive governance doesn't slow AI down. It accelerates value while reducing risk. And even when data is accessible and accurate, AI can still miss the mark if it lacks context. That's where data enrichment plays a critical role. It's no longer enough to just use your organization's data.

You need to bring alternative party datasets such as consumer behavior, demographics data, point of interest, and address data, environmental data for more accurate outcomes. To increase relevance, organizations should, one, diversify data sources and types, also balancing between local and global data vendors. acquire highly created party data to ensure it's trustworthy. And again, make sure the party data is also fresh and updated. This is how your AI outcomes become current, contextual, and actionable. So how does this look like in practice? Let's look at two examples of organizations where precisely have built on this trusted data foundation and turn into real business value with AI. Let's start with a mortgage financing company who buys mortgages on the secondary market, pulls them, and sells them as mortgage backed securities to ensure the housing market is liquid, stable, and affordable.

They also work to ensure low income housing is available. This company needed to feed AI systems with complete transfer to data, and that wasn't an easy task across silos and legacy systems. They standardized, enriched their property data, and implemented the governance to track AI usage and compliance. They put in place end to end data governance to manage the data feeding into the AI models, ensuring transparency, auditability, and alignment with evolving guidelines like European Union AI Act. The result, they identified $7,000,000,000 in previously missed multifamily properties and expanded low income housing access by 32 34%. Here's another example with a major food delivery service. How many of you had the experience of wrong delivery? This food delivery company faced a costly challenge, failed deliveries due to location inaccuracy. Sometimes, actually, I benefit from that.

There were two times I had food delivered that was meant for others. They created point of interest data and real time location data to power hyper accurate AI driven last mile delivery. The result, $65,000,000 in saved revenue, higher customer satisfaction, and as a side benefit, better tips for drivers. If there's one message to leave you with, it's this. Building AI value starts with trust in your data. To drive a business impact at scale, organizations need to build AI foundation of comprehensive data integration, be proactive about data quality and governance, and contextualize data. This is how we move from data to trust and to impact. Thank you.

Thank you so much, Tendu, for this very insightful presentation. People asking if the recording will be available already for someone who want to revisit and to kinda deep dive. And, I'm gonna