The handbook · Working draft
The AI Personalization Handbook
A practitioner's reference for AI personalization, from the first decision to production.
Most companies do not have a data problem. They have a depth problem.
They know who is renewing, who is frustrated and who just hired a new ops lead, then send all of them the same email. This handbook covers the path from that gap to a system that remembers, decides and writes for one customer at a time.
This is a working draft. I'm revising every chapter in the open, so expect changes.
How to read it
- If you leadEvery chapter opens with the short answer: what to decide, what it costs, what to ask your team. You can stop there.
- If you buildThen it goes deep: the mechanism, the patterns, the failure story, a build checklist and the metrics to watch.
- If you work with an AIEach chapter ends with a structured block you can paste into your assistant, so it reasons with the same concepts.
Contents
Part I
The Case
- 01What Personalization Is, and What It Is WorthWhat is personalization, and why should a leader care now?23 min read
- 02The First EraHow did we personalize before language models, and what did it teach us?27 min read
- 03What Designers KnewWhat did UX research and HCI learn about adapting interfaces to people, and what does it teach AI personalization?25 min read
- 04Shallow to DeepWhy was deep personalization always limited to a few people, and what does "deep" actually mean?18 min read
- 05The New EraWhat exactly changed with AI, and where is it going?27 min read
Part II
Where to Start
Part III
Customer Data
- 07Getting the DataWhat data can we use, where does it come from, and are we allowed to use it?22 min read
- 08Preparing ItHow do we clean, enrich, and use unstructured data?21 min read
- 09One Customer, Many SystemsHow do we unify a customer across CRM, warehouse, and tools without merging two people?25 min read
Part IV
Unified Customer Memory
- 10Memory Built for AgentsWhat changes when the main reader of our customer data is an AI agent, not a person at a dashboard?30 min read
- 11Keeping Memory TrueHow does memory stay accurate, fresh, and traceable?26 min read
- 12The Cost of RememberingHow do we keep memory affordable at scale without losing accuracy?21 min read
- 13Agents That Run for WeeksHow do long-running agents use and maintain memory?22 min read
Part V
Deciding and Acting
- 14The Decision LayerWhat should happen next for this person, including nothing?26 min read
- 15Writing for OneHow do we generate for each person without inventing?24 min read
- 16Beyond the Message: Generative ExperiencesWhat happens when the page, the app, and the notification, not just the message, are composed for one person?27 min read
- 17Now or LaterWhen must personalization be real-time, and when is async better?19 min read
Part VI
Trust
- 18GovernanceHow do we govern what the AI says and does, at the level of the organization and at the level of each customer?27 min read
- 19Privacy and RegulationWhat should we never know, keep, or use, and what does the law require?25 min read
- 20Accuracy and QualityHow do we achieve and keep accuracy at scale?25 min read
Part VII
Proof and Deployment
Part VIII
Playbooks, by team
The foundations applied to one team's work: what to personalize first, what to leave alone, and how to know it worked.
- Marketing · M1Inbound and the WebsiteThey came to us. What should they see and receive next?
- Marketing · M2Accounts, Ads and EventsWe chose them. How do we reach them as individuals?
- Marketing · M3The Physical ChannelWhen each touch costs real money, who deserves one and what should it say?
- Sales · S1New CustomersHow do we open a relationship without sounding like every other AI message?
- Sales · S2Growth and ReturnWhat do we already know, and how do we use it to go deeper or win them back?
- Customer success · CSCustomer SuccessHow does customer success personalize onboarding, adoption and renewal for each account and each user?
- Support · SPSupportHow do we stop asking customers what we already know?
- Product · PRProductHow does the product adapt to each user, and what does the team learn?
- Software · SWSoftware CompaniesHow does a software company personalize when its richest signal is how people use the product?
- Research · RSResearch and StrategyHow does understanding individuals roll up into strategy?
- Across teams · P9One Customer, One ConversationHow do we stop marketing, sales, support, product, and our agents from contradicting each other, or exhausting the same person?
Companion paperGoverned Memory: A Production Architecture for Multi-Agent WorkflowsThe full architecture behind Part IV, with evaluation. arXiv, 2026.
Get chapters by email as they are revised