Episode 84 - Agentic AI & Retention in 2026

Welcome to episode 84 of the Retention Blueprint.

Over the last few years, I have gone pretty big on the impact I expect AI to make on retention. 

In some forums, my ideas raised a few eyebrows; the panel I was on at the Optimove client dinner in December 2024 comes to mind, as does another in autumn 2024 in Berlin with Airship. 

Of course, the technology was always going to be ahead of execution in enterprise, due to the complex nature of large organisations, inter-departmental politics, technical debt, sometimes shaky data foundations and business readiness. 

Turning Point in History

Now in 2026, driven by huge advances in reasoning models (both in efficiency and reliability) and improvements in function calling*, we are at a turning point in history that will transform how brands deliver customer experiences powered by AI agents.

*Function calling is where an LLM/Agent can read a situation, decide which system action and output a structured, machine-readable instruction and then safely execute. 

The evidence of this is all around us: 

1. The Magnificent Seven are investing heavily in AI and core infrastructure. Amazon, Alphabet, Meta, and Microsoft have all signalled sustained, multi-year investments running into the hundreds of billions of dollars across data centres, custom silicon, models, and AI platforms, while Apple has announced plans to invest up to $500 billion over the next four years across R&D, silicon, and AI system development.

2. Customer service and CX are moving from reactive workflows to agent-led resolution. AI agent-driven chatbots can now interpret intent, access multiple systems, and resolve issues end-to-end.

3. Platforms like Gmail are increasingly using AI to summarise emails, surface highlights, and prioritise messages, reducing the control brands have over subject lines, preview text, and inbox placement before users engage.

Fundamental changes to how we manage customer relationships, customer experience, service experiences, messaging, product experiences, operations, and organisational models are underway, powered by AI agents. 

Foundational Elements

And while there is a lot of noise. Some camps (where I sit) are massive advocates of the technology and see huge competitive advantages if executed well, while others are more cautious (although those voices are quietening by the day).

But what most commenters agree on is that there are two foundational elements required for success with agents: 

1. Data. To leverage deep learning neural network models that agentic AI relies on, organisational data readiness is critical. 

2. Customer strategy. Some people seem to think we will simply turn on agents and no longer need to think. That is simply not the case; human beings need to set customer strategy, establish guardrails, define HITL (human-in-the-loop) protocols, and define the truth layer and decision parameters within which agents operate. 

To truly understand why these elements are important, it's important to understand the layers that exist to power agentic AI: 

Major Agentic AI Trends

Having been through some fairly large-scale AI-first retention transformation programs in 2025, I see two major trends emerging in retention: 

1. Leveraging ‘embeddings’ to predict movement away from value, thereby predicting customer defection/value loss before the customer even consciously thinks about leaving/downgrading. Agents that detect drift before drop-off can automate earlier, lighter interventions across product, service, CX, and messaging. 

2. Business teams set the strategic direction and guardrails, while agents build dynamic, outcome-based journeys for individual customers, with bespoke content, messaging frequency, and experiences driving movement toward value based on goals and not predefined rules.

My belief is that this will mean: 

1. Foundational data technologies are structurally safe and are critical in the agentic world. If your business operates with misaligned definitions, multiple sources of truth, a lack of data governance, an unreliable data cadence, a lack of a clear data strategy, or an operationalised data management plan, this is an urgent fix. 

2. Customer retention execution platforms will remain, and while there will be some disruption, fundamentally, they are not at risk.

   1. Agents still need to instruct the messaging tools and CMS to execute, in accordance with the guardrails. 

   2. Agents still need to decide when to bring humans into the loop and feed relevant customer-level contextual information to customer service, sales and operational front ends (based on guardrails set by humans). 

3. Tools that simply predict churn, orchestrate journeys, provide risk scores, or run rules engines are structurally at risk and are likely to be absorbed into agent layers. Capabilities like journey builders within execution platforms, and the skills required to operate them, will also become much less important, as this work is absorbed into agent layers, too.

If you want to join the conversation, check out this post on this subject (particularly if you disagree!).   

How to Leverage AI Agents

To leverage AI agents effectively 

1. Data foundations, if shaky, must be a core focus area. 

2. Understanding the underlying drivers of customer retention in your organisation is critical to setting an accurate agent truth layer and guardrails. It is critical to have a multifunctional, cross-departmental customer retention strategy powered by customer truth (not internal agendas).  

3. Retention agents should be rolled out in phases. Identify low-risk, high-impact use cases, and build, scale, test, and iterate in controlled environments. Brands cannot jump to organisation-wide agentic AI in one project. Start with one agent and build from there. But the key is you must start.  

Your Opinions

I’d love your take. If you would like to discuss this further with me directly, reply. I read and respond to every email.  

Or if you would rather continue the conversation on LinkedIn, follow me or leave a comment on this post.  

Foundational Frameworks

Over the last month, I have built 3 foundational frameworks that can be used to build your customer retention strategy and to set guardrails around your AI agent builds. You can access them here:  

Framework 1: Retention is an ecosystem 

Framework 2: Base Management & Lifecycle 

Framework 3: Moments of truth 

Important Note

If you want to keep seeing this newsletter in your inbox, just click a link or hit reply with a hi. This tells your email tool that this content is useful. If it’s not for you, feel free to unsubscribe. I won’t be offended. I only want to show up for people it helps.

One Last Thing: 

I’m running a short agentic AI session on Google Meet in early March for B2C C-suite and Director-level folks responsible for retention in recurring-revenue businesses.  

I’ll walk through how agentic AI can help teams:

• Spot risk before defection signals even appear (with embeddings)

• Use agents to design earlier, smarter interventions

• Shift from reacting to actually shaping retention outcomes

It’s practical, not theoretical. 45 mins + discussion, no pitch. On Google Meet. If you fit the profile and would like to join, hit reply and apply with your name, role and organisation. 30 other leaders are confirmed from the worlds of streaming, telco, gambling, publishing, subscription mobile apps, subscription e-commerce and more. 

Until next time, 

Tom