Foundational ML, generative AI, agentic automation. Three distinct narratives with different buyers, different objections, and different levels of category familiarity. One through-line connecting them.
Most technical marketers have built narrative for products buyers already understand at some level. The category exists. The problem has a name. The buyer has a budget line somewhere close to what you are selling.
Narrating AI from 2019 through today has been something different. Not one technology to explain. Three distinct phases, each with its own buyer vocabulary, its own category readiness, and its own set of objections that did not exist the cycle before.
The hardest part of narrating AI is not explaining the technology. It is knowing which phase of buyer familiarity you are actually in, and building the story for that phase, not the one that feels most comfortable to the product team.
This issue is the story of how the arc actually got built, and what the transition between each phase demanded that no playbook warned me about.
02 / Three Phases
What each phase actually required
Each phase is not just a different technology announcement. It is a different conversation with a different buyer in a different state of readiness. The narrative work for each one looks almost nothing like the others.
Phase One / 2019 to 2022
Foundational ML and AI
The buyer has no frame. The category does not exist yet. Your job is to name the problem before you are allowed to name the solution. This phase is less about proving the technology and more about earning the right to introduce a new problem category the buyer will eventually recognize as something they already have. The risk is moving too fast to capability before the problem has been accepted as real.
Phase Two / 2022 to 2023
Generative AI and Now Assist
The category explodes. Every buyer has heard of it. Most of them have a strong opinion about it and have not used it seriously. The narrative challenge inverts: now you have to differentiate inside a crowded and noisy market rather than introduce a concept. The risk here is generic positioning. Everyone is saying the same things. The technical marketer's job is to find the specific claim that only your product can make and prove it visibly.
Phase Three / 2024 to Present
Agentic Automation
The buyer has seen promises broken. The expectation bar is high and the trust threshold is higher. Agentic AI narrative has to prove outcomes, not just capability. The buyer is not asking what it can do. They are asking what happens when it does something wrong, and who is accountable. Technical marketers who have not shifted their narrative to address accountability and governance are still in phase two.
03 / The Through-Line
What connects all three phases
The technology changes dramatically between phases. The buyer's underlying problem does not change at the same pace. Anchor every phase transition in what the buyer still recognizes as their pain, then introduce the new capability as the next level of answer to it.
In 2019 the pain was data quality and process inconsistency. In 2022 it was still data quality and process inconsistency, but now the buyer also had to contend with a market full of AI promises they did not know how to evaluate. In 2024 and beyond the pain is AI promises that did not deliver, and the need for something that works in production, not in demos.
Every capability in the stack evolves. The buyer's skepticism evolves with it. The technical marketer's job is to stay one step ahead of that skepticism, not one step behind it.
The through-line from phase one to phase three is not the technology. It is the credibility of the story. Phase one built the right to be heard on AI. Phase two built the right to differentiate inside the category. Phase three is where that credibility gets tested against real outcomes and real accountability.
04 / Phase Transitions
What the transitions actually required
Moving from phase one to phase two required letting go of the education-first narrative and trusting that the buyer now had enough context to evaluate differentiation. That was harder than it sounds. After years of building the category, the instinct is to keep explaining. The market had moved past needing the explanation. The narrative had to move with it.
Moving from phase two to phase three required something harder still. It required admitting, inside the narrative, that earlier AI promises had not always been kept. Building agentic AI narrative that acknowledges the failure modes of the previous generation is not comfortable work. But it is the only narrative that earns trust in a buyer environment shaped by broken promises.
No playbook covers that transition. The ability to hold the discomfort of an honest narrative, one that names what did not work in order to earn the right to describe what does, is not a skill most product teams will ask for by name. It is exactly what the moment requires.
Questions this issue answers
Phase one is foundational ML and AI, where the primary challenge is explaining a technology the buyer has no frame for yet. Phase two is generative AI, where the challenge shifts to differentiation inside a crowded and noisy market. Phase three is agentic automation, where the challenge is proving outcomes and building trust for AI that acts on behalf of users. Each phase requires a different buyer conversation, different objection handling, and a different level of product proof.
AI categories get invented in real time. The buyer often has no established frame for the capability, the market is still being defined, and the competitive landscape is being built simultaneously with the narrative. A technical marketer narrating AI has to create the category frame before making the product case. That requires more groundwork, more patience with abstraction, and a clearer read on which phase of buyer familiarity you are actually in.
The through-line is the problem the buyer is living with, not the technology you are introducing. When you move from ML to GenAI to agentic systems, anchor every phase transition in what the buyer still recognizes as their pain, then introduce the new capability as the next level of answer to it. The technology changes dramatically. The buyer's underlying problem changes more slowly.
Apply the arc to your demo narrative
The TDCI framework scores demo content for narrative coherence -- whether the story holds up across evaluation, not just in the room where it was built.
Demo Intelligence
Score a Demo
Run your demo through the TDCI framework and get a coherence score across nine formats.
Founder of ThinkRoot. Twelve years in Technical Product Marketing at ServiceNow across ITSM, SecOps, ITAM, and Agentic AI. Built and narrated enterprise AI narrative from foundational ML through the GenAI wave to agentic systems in production. U.S. Navy veteran.
