Our Biggest Takeaways on AI & Alternative Data in 2025

At BattleFin Miami, BattleFin CEO Tim Harrington, Nimble Way CEO Uriel Knorovich and Tiny Fish CTO Mark Peng sat down to discuss what advances in AI technology means for financial alt data and what to expect in 2025.

Here are our top takeaways 

1. Two Paths to AI-Driven Advantages in Alt Data 

a. Truly Alternative Data with AI

What’s truly alternative data in 2025? Increasingly syndicated data brings greater access to data than ever but also raises the question: is your alternative data truly alternative, or available to most players in the field and thus mainstream? Meanwhile, AI is enabling data consumption and parsing at unprecedented speed and scale, redefining “the art of the possible” in alternative data. 

The key question – what datasets, previously thought inaccessible, can now be brought into the fold? And is your team the one capable of sourcing these insights first? 

b. Extracting Greater Insights Faster from Alternative Data

• Cleaning and Structuring Data:

Turning complex, messy sources into ready-to-use data is often costly and slow. AI enables this process at a new speed and scale. For example, Tiny Fish uses AI to both extract and transform data from complex websites and files into data tables in one step. Nimble uses AI Agents to support Enterprise needs, from governance & compliance in the data gathering phase, to structuring, cleaning and validation in the processing phase.

• Advanced Analytics and Specialized Tools:

Tools like the Exabel platform exemplify the new paradigm of alt data analytics and consumption. Such specialized platforms consolidate datasets, map and tag these datasets to KPIs and leverage AI to produce insights and signals. Whereas the alpha generation process at each firm can be unique, having a platform like Exabel to get you 80% there seems to be the proven choice.  

2. The Specialization of AI and Enterprise Use Cases

The era of generalized AI training is nearing its peak. In 2025, we expect a significant shift toward post-training specialization, where AI systems are adapted to address the unique requirements of specific domains, such as finance, supply chain, or healthcare.

For instance, in finance, entity resolution is key—whether it’s resolving tickers or mapping relationships in corporate hierarchies. Contrast this with a supply chain scenario, where the focus might be on parsing multilingual procurement data. AI’s ability to specialize enables it to deliver contextually relevant insights across industries.

Small, hard-to-maintain automation scripts are increasingly being relegated in favor of solutions that offer Enterprises the customization, scale, compliance, and reliability they require. Nimble is building solutions towards rising Enterprise needs. For instance, Coca Cola faced a data interpretation challenge on whether to process multiple online storefronts of a business as unified or fragmented entities. Nimble’s AI-driven interpretation is tailored to use case needs.

3. The Biggest Challenges Ahead

a. Accuracy and Reliability 

Across sectors, delivering accurate and reliable datasets remains a foundational challenge. Consistency across every run is not just a preference—it’s a necessity for any viable enterprise solution.

Open AI Operator and Google Project Mariner provide a glimpse into the future of agentic automation. In 2025, computer-using-agents face a similar challenge to build the framework and infrastructure to be enterprise ready.

b. Trust and Adoption: Bridging the Gap

The AI industry still faces significant cultural and technical hurdles in building trust. Panelists drew parallels with self-driving cars: no matter how reliable the numbers suggest these systems are, public and institutional trust takes time to establish.

AI systems must incorporate intelligent fact-checking and validation to enhance reliability. Yet, the human-in-the-loop model remains essential. As AI rapidly evolves, its adoption will continue to require robust guardrails, transparent annotations of data processing, and the assurance of human verification. 

c. Enterprise Requirements

Enterprises have unique working requirements, from stringent regulation and data accountability to high working volumes and unique use cases. Scaling up AI while maintaining performance, cost-effectiveness, and compliance is a substantial challenge facing the industry

‍