AI Rewrites the Script

From Instincts to Insights

United Talent Agency (UTA) operates at the center of a rapidly transforming digital media landscape. In an ecosystem where new talent can emerge overnight through platforms like TikTok, Twitch, or YouTube, and traditional indicators like box office numbers or Nielsen ratings no longer apply, agencies like UTA need a better defense to confront mounting complexity in tracking market trends and guiding careers.

Michael Keithley, UTA's Chief Information Officer, observed that the conventional methods were no longer sufficient in an era dominated by media platforms and data analytics. In response, he spearheaded an AI-driven transformation centered around predictive modeling, real-time data ingestion, and deep signal processing across a fragmented digital landscape. 

This transition was not merely about adopting new tools but redefining the agency's approach to talent representation. By leveraging data, UTA aimed to provide more informed guidance to its clients, ensuring they remained competitive. The result was a proprietary intelligence system aggregating cross-platform metrics and redefined how UTA identified, evaluated, and represented talent.

Challenges of Traditional Talent Representation

Historically, talent agencies depended on methods such as A&R (Artists and Repertoire) to scout and sign new talent. This approach involved attending live performances, relying heavily on personal networks, and making decisions based on subjective judgments. “In the old days of music…you would go to clubs and watch all the bands, and maybe you would listen to the radio. Well, now with so many digital tools, metrics, and platforms, they do a lot of that work for you,” said Keithley.

This model delivered diminishing returns in a post-broadcast world. Relying on physical attendance at events restricted the pool of talent that agents could access, personal biases often influenced decisions, and predicting an artist's potential reach or success in the market was challenging without concrete metrics.

Even platforms where creators gained traction, like Instagram, Spotify, and YouTube, offered no unified view of influence. Additionally, streaming giants like Netflix and Amazon actively withheld performance data. This lack of transparency made it difficult for agencies to make informed decisions on behalf of their clients.

Without the ability to capture second-by-second attention trends, agents were increasingly flying blind in a metrics-first world: “If anybody has kids in their house, they understand that their view of entertainment is profoundly different than previous generations,” Keithley said.

Building the AI Core

Recognizing these challenges, UTA embarked on a mission to overhaul its talent representation strategy by integrating data analytics into its core operations. Under Keithley's leadership, the agency sought to harness data across various sources to gain a comprehensive understanding of market trends and audience preferences. This involved aggregating data from social media platforms, streaming services (where available), and other digital channels to build predictive models and insights.

Keithley and his team knew the solution would require not just aggregating data, but also operationalizing it. UTA developed a distributed data platform engineered to capture, clean, and correlate creator signals in real time. The system resembles edge-processing networks, extracting actionable intelligence from disparate, distributed nodes (social media platforms, content channels, and fan interactions), while avoiding latency and overload in centralized systems.

Architecture highlights included:

• Multichannel data ingestion: APIs from YouTube, TikTok, Twitch, Instagram, and public engagement metrics.
• Contextual modeling: Data normalization and enrichment using third-party intelligence tools.
• Predictive analytics: ML-based forecasting to determine creator trajectory, likely engagement arcs, and revenue modeling.

The system also enables adaptive modeling, a core requirement for a space where algorithms and consumption behaviors evolve weekly. “Digital technology is going to have profound effects on every aspect of the entertainment industry, and it's my job to help our company be prepared for that,” said Keithley.

The key to success was embedding the platform into agent workflows. Rather than deploying a monolithic dashboard, the team built modular components that mapped to specific client journeys like talent evaluation, deal negotiation, and audience development. Implementation took 12 to 18 months, with phased rollouts designed around agile development cycles.

Impact Across Talent Lifecycle

Once deployed, UTA’s intelligence infrastructure generated compound returns across talent discovery, representation, and dealmaking. Their accelerated scouting system flagged rising creators weeks before they reached viral inflection points, allowing UTA to engage with them early and with context.

Once signed, agents were armed with empirical insights into fan behavior, audience reach, monetization trends, and content performance when entering into negotiations and content strategy work for their clients. Campaigns were then tailored to specific fan cohorts, content types, and platform formats, increasing engagement rates and ROI for talent across verticals. With more profound insights into audience preferences and market trends, agents could tailor career strategies to meet current demands.

Internally, the platform also reshaped collaboration. Cross-functional teams began working from a shared intelligence baseline, improving decision velocity and minimizing friction between talent, business affairs, and digital strategy teams. This approach improved operational efficiency and positioned UTA as a forward-thinking entertainment agency.

Lessons from the Entertainment Edge

UTA's transition underscores the importance of adaptability and innovation in the face of industry evolution. For Keithley, the takeaway wasn’t just about tooling—it was about strategy. He emphasized that in volatile environments, distributed intelligence systems give organizations resilience. 

Looking ahead, UTA is exploring deeper machine learning layers to run scenario modeling for clients. The goal is to simulate outcomes before making strategic decisions, not after: “we pivot, we change, we react, we test stuff, we try it over and over again.”

Keithley recommends this to other enterprise leaders: start with questions, not tools. “I think you have to be a voracious reader and learner, and you need to be curious and want to be a sponge on all kinds of new stuff.”