Hyp.
Articles

Articles

Notes on Agentic Marketing.

5 Layer agentic marketing design

My take on 1-2-1 personalisation at scale (the 5 layer technique)

Real 1-2-1 personalisation at scale isn’t a single segment label. It’s five layers working together: Shopper State, behavioural clusters, intent trajectory, catalogue knowledge, and an agentic orchest

May 27, 2026

Martech grid illustration

The martech chart is crazy, and marketers are feeling the pressure more than ever

14,000 martech vendors, most of them dashboards with a chatbot bolted on. The bar should be higher: an agent that takes action, autonomously, and drives revenue without being asked.

May 25, 2026

Push notification from a dashboard illustration

Insider buying Bluecore validates the destination

Insider buying Bluecore is a $1B signal that autonomous, end-to-end customer engagement is a real category.

May 25, 2026

Illustration of poker chips and cards

AI engineering reminds me of playing online poker in 2006

In 2006 I used to stack four poker tables on my CRT and call it a flex. Two parallel AI agent teams is my coding equivalent. The skill isn't how many you run, it's knowing which ones to fold.

May 20, 2026

Table aggregation

Architecting the ledger table pattern

Tried to populate my dashboards with a year of test data and watched the aggregator silently ignore all of it. The ledger table I'd been resisting on day one was already the answer.

May 20, 2026

Shopify PDP page illustration

Looking cool vs delivering value

Cool AI demos look like magic but rarely deliver value. Useful AI hidden behind a dull dashboard never gets believed. The real question: when does invisible work have to become visible?

May 19, 2026

Email production line

We optimised the workflow and broke the deliverable

Modern CRM platforms can trigger a campaign in seconds. But the thing that lands in the customer's inbox looks more average than ever. We optimised the wrong layer.

May 18, 2026

Hierarchy of AI decisioning

Real-time personalization isn't an agent-per-user problem

An LLM agent for every customer every hour doesn't run at scale. The unit of decisioning has to be the micro segment, layered with ML and database-querying agents, recomputed regularly.

May 14, 2026

Scaling Shopify architecture

Trying not to architect for a scale problem I don’t yet have

Yesterday I down to refactor Hyp's Shopify integration today and caught myself trying to architect for 1,000 B2B customers. I haven't sold ten.

May 13, 2026

Abstract Shopify PIxel dot

Why I'm building for Shopify first

The Shopify pixel skips the data-layer problem that kills most ecommerce SaaS onboarding. The small-team operator is the harder test that proves the product before it goes anywhere near enterprise.

May 12, 2026

Abstract drawing of pipes delivering an email

What if CRM workflows got out of the way?

Most CRM teams spend a month wiring up a flow and a week writing one email for it. The next wave of CRM puts the effort back into the deliverable itself.

May 12, 2026

Abstract image of data flowing to a dashboard

What if the dashboard isn't the answer?

Most Shopify owners follow the same routine: open the dashboard, nod, and change nothing. We’re building Hyp to turn endless charts into a short list of plain-language decisions.

May 11, 2026