Beyond Chatbots: How Agentic AI Is Rewriting the Playbook for Media and Advertising

At the 2026 CIMM Summit, VideoAmp Chief Product & Technology Officer Josh Hudgins delivered a compelling message about the future of media: artificial intelligence is undergoing a massive shift from answering questions to executing complex workflows.

In recent years, enterprise AI adoption has focused on chat interfaces with humans prompting chatbots to retrieve information. Today, the industry is entering a new era powered by autonomous AI agents capable of orchestrating end-to-end media operations.

To put this transition into perspective, consider how AI workloads have flipped over the past year. In AI, data processing is measured in “tokens.” According to Andreessen Horowitz (a16z), AI agents officially overtook human data and token usage in February of 2026. Today, autonomous software agents running behind-the-scenes workflows account for 14 times more activity than human typing, processing a staggering 7.3 trillion tokens.

In plain terms: AI has officially evolved from a conversational tool into an engine that gets work done on its own.

The Shift from Chatbots to Autonomous Workflows

Chat interfaces have become the default way companies test AI. But when applied to cross-channel media and measurement data, simple chatbots quickly hit hurdles:

• Raw Data Isn’t Insight: Large language models (LLMs) cannot turn unformatted rows of viewership or exposure data into meaningful answers without identity resolution, deduplication, and projection.
  
• Conflicting Metric Definitions: The same metric often means completely different things across disparate datasets. Without contextual rules, AI models pick definitions by chance.

• False Confidence: LLMs are notoriously confident even when incorrect, producing plausible-sounding answers that are often only detectable by deep domain expertise.

The Deterministic Foundation AI Requires

Transitioning safely to autonomous agents requires the exact same foundational work needed to make chatbots accurate. At VideoAmp, we’ve operationalized three core principles to build deterministic trust into AI measurement:

1. Calculations via API: AI models should never calculate metrics directly from raw data. By forcing models to interact exclusively with API outputs, VideoAmp is able to apply statistical methods, weighting, and methodology consistently.

2. A Unified Semantic Layer: By encoding precise data definitions directly into a semantic layer, the AI reasons from single-source-of-truth rules rather than guessing what a metric means.

3. Guardrails That Say “I Don’t Know”: Built-in evaluation frameworks and validation checks catch partial or out-of-scope data, forcing the AI to flag uncertainty rather than hallucinate a figure.

By building on this deterministic foundation, VideoAmp enables users to move past endless spreadsheets and ask direct questions, receiving trusted answers accompanied by clear, actionable visualizations.

Scaling Outcome-Based Advertising

When autonomous agents are built on reliable data, previously prohibitive advertising workflows become possible at scale.

Consider advanced outcome-based deal structures. Historically, pricing, forecasting, pacing, and guaranteeing campaigns against lower-funnel business outcomes (like conversions or sales) required immense manual effort. Executing these deals was typically limited to small-scale pilots.

With multi-agent frameworks (using emerging protocols like the Model Context Protocol or MCP) agents can communicate across clean rooms, identity spines, and planning systems:

• Agents Orchestrate: AI agents automate audience definition, clean room joins, forecasting, CPM/CPA pricing, and real-time pacing adjustments.

• People Decide: Humans are freed from manually moving spreadsheets, allowing strategy, creative direction, and high-level decision-making to drive campaign performance.

Preparing Your Organization: Encoding Institutional Knowledge

To succeed in an agentic future, organizations shouldn’t focus solely on selecting new AI tools. The most vital, urgent step is documenting and encoding internal domain expertise: the business rules, edge cases, logic, and judgment calls that currently live inside employees’ heads.

Josh illustrated this point with a quick historical parallel: when Gutenberg introduced the printing press, text production exploded, but scaling books hit an immediate bottleneck because the industry lacked scalable paper production. Today, agentic AI is our printing press, capable of executing workflows at unprecedented speeds. But our modern bottleneck is organizational clarity.

Without pulling domain logic out of team members’ heads and structuring it into explicit rules, semantic layers, and context, AI agents will yield unpredictable, garbage results. The organizations that thrive in this transition won’t just be the ones buying AI platforms, they will be the ones that deeply audit their own processes and encode that institutional knowledge into clear instructions so agents can run workflows accurately, securely, and effectively.

Looking Ahead

The future of media measurement is moving rapidly from passive reporting to active execution. By pairing a deterministic data foundation with agentic orchestration, VideoAmp is leading the industry beyond raw metrics to deliver scalable, outcome-driven advertising. 

Get in touch to find out how you can get started and stay ahead of the curve: let’s chat.