At Channel99, we’re training our AI models the same way you’d onboard a top-performing team member—step by step, with a strong foundation in B2B marketing, technology, and data.
Over the past year, Channel99 has been working diligently to develop the most powerful Marketing Performance AI agent, aimed at making insights more accessible for data-driven marketers. On this journey, we’ve come to realize that the term “AI” is often overused and misunderstood—it’s used to describe everything from tools that help rewrite a blog post to mythical systems that seem capable of solving any problem without training. The reality is that AI must be carefully honed and guided in specific directions to provide true value to end users.
At Channel99, we’ve taken a phased approach to training our AI models, treating them much like an employee—teaching them the necessary technologies, terminologies, and marketing best practices. The major advantage, of course, is that the AI learns at scale, using far more data than any human could process. Let me take you through the journey of how we started and where we’re heading in the next three phases of development.
Phase 1 – Enablement and Support
We began by training our AI to help enable and guide customers in setting up and navigating our platform. This phase focused on improving the onboarding process for new users while simultaneously teaching the AI about marketing technology and industry-specific challenges.
During this stage, we fed the AI with support materials, definition libraries, and FAQs to ensure it developed a robust understanding of our subject matter. The result was an AI that could effectively answer questions, streamline onboarding, and provide a strong foundation to support users.
Phase 2 – Data Analyst
In the second phase, we shifted focus to transforming our AI into a capable Data Analyst. The goal of this phase was to enable the AI to interpret user queries—whether written or verbal—and convert them into actionable data queries. These, in turn, would retrieve performance data from the customer’s historical metrics.
To further enhance accuracy, this phase was underpinned by marketing playbooks and best-practice guides. These resources enabled the AI to use deterministic logic to recognize patterns and evaluate performance. For example, the AI needed to understand when smaller numbers are better (e.g., Cost per Visit) and when higher numbers signify success. Without these foundational instructions, the AI would lack the necessary context to provide meaningful insights.
This is where we are currently—in Phase 2—with the AI acting as a skilled data analyst that not only tells you how your marketing campaigns are performing but also provides actionable recommendations to drive improvement.
Phase 3 – Business Analyst
Our next phase involves evolving the AI into a Business Analyst, leveraging advanced machine learning and predictive modeling. The focus here is on enabling “What If” scenario simulations—empowering marketers to forecast future outcomes and make data-informed decisions based on prior performance and incremental analysis.
This phase is particularly complex because it requires vast amounts of data to build accurate predictive models, as well as the foundational contextual knowledge gained in the earlier phases. By teaching the AI to interpret and analyze data in real-world terms, we can ensure its predictions and recommendations are both credible and actionable for our customers.
This phase represents the culmination of the current training roadmap, and as the AI continues to learn and evolve, we’re confident it will become an even more sophisticated tool for marketers.
With the rapid pace of AI innovation, there’s no question that these tools will only grow more powerful and valuable in the years ahead. At Channel99, we’re committed to staying at the forefront of AI development, ensuring data-driven marketers are empowered with the best tools to unlock their full potential.
