Notes on agentic personalization.
Field notes from the agentic experience layer: cohorts, holdouts, determinism, and shipping one decision to every channel.
The system of record won. The experience layer is still up for grabs.
CRM consolidation is over and the record layer got its winners. The layer that decides what a visitor sees in the next 200 milliseconds is a different discipline — which is why suites keep buying it, renaming it, and losing it.
Read →Mutiny retired personalization. Now what?
In April 2026 the company that defined B2B website personalization stopped selling it. If you're holding one of those contracts — or shopping like it's 2024 — here's the honest map of what replaces it.
Read →Intellimize is Webflow Optimize now. You have a decision to make.
The acquisition turned a platform-independent optimization tool into a feature of a website builder. If your site doesn't live on Webflow, "keep using it" quietly became "replatform your site."
Read →Leaving Adobe Target without buying another suite
Nobody quits Adobe Target over its testing engine. They quit the ecosystem it's welded to — the contract, the consultants, the velocity. The alternatives sort cleanly by what kind of team you are.
Read →Your agent is hired. Which martech tools can it actually operate?
Every vendor now claims AI. The checkable claim is an MCP server — and the delta between "has one" and "an agent can run the product" is where the 2026 martech evaluation actually happens.
Read →The martech suite era is ending. Good.
Twenty years of consolidation bought you seats, dashboards, and integration projects. The next stack is a grab bag of small, sharp, agent-operable tools — composed by your agent, not a vendor's roadmap.
Read →Your CRM should be free. The thinking is what costs.
The CRM's database was never the value — the discipline of keeping it current was. When an agent does the record-keeping, per-seat pricing on a table stops making sense.
Read →We gave the agent 130+ tools
The API era made software talk to software. MCP makes software operable by rented intelligence. Here's what a personalization platform looks like when an agent can run the whole thing.
Read →If your traffic is thin, we hide the verdict
Most personalization "lift" is a model grading its own homework. Ours is a randomized holdout, a z-test, and a scorecard that would rather say nothing than flatter you.
Read →The six-month implementation is a choice
Enterprise personalization takes two quarters to launch because someone's margin lives in those two quarters. A crawler, a vertical pack, and self-healing selectors get you live the same day.
Read →Personalization that collects nothing
The industry answered privacy pressure with consent banners stapled onto identity harvesting. The real answer is architectural: personalize the session, collect no one.
Read →Personalization didn't fail. The operating model did.
First-generation personalization engines mostly worked. The services army required to operate them was the real product — and when it left, the ability to answer "is this working?" left with it.
Read →How to not get gaslit by your platform rep
Three questions that separate real runtime decisioning from a rules engine wearing an AI badge — and why a holdout is the only honest proof of lift.
Read →You don't need to know who they are to know what they want
Identity-first personalization has nothing to say to the anonymous 90% of your traffic. You don't need to resolve who someone is to act on what they're doing right now.
Read →Your "AI personalization" stack is solving the wrong half of the problem
Most "AI personalization" tools help you make content faster. Deciding what a specific visitor sees in the moment is a different job — and the one that moves the number.
Read →No hallucinating agent in your decision chain
A language model in the live decision path demos beautifully and bills you three taxes in production — cost, confidence, and a standing eval burden. The fix is AI upstream, a deterministic hot path.
Read →When a vendor brags about 13 agents, run
The agent count is marketing, not capability. What the agents are wired into tells you everything.
Read →The three-stack tax on customer journeys
A CDP, an A/B tool, and an automation platform: three contracts, three teams, and a seven-figure tax on the seams between them.
Read →Your segmentation is backwards
You wrote a thousand rules and still only spoke to half your audience. The fix isn't more rules. It's letting the signal draw the lines.
Read →Notes in your inbox, when they ship.
No cadence games. Just new thinking on agentic personalization.