Rafael M. EhlersRafael M. Ehlers

Product Manager who builds, writes, and researches AI.

I came to product the long way: a decade of support, engineering, and staying close to customers. I turn messy real problems into things people actually want to use, whether that's an AI product, a side project, or a book. Lately I also write conceptual AI research papers on language agents — memory, individuation, and successor-model lifecycles.

I study in agents what I practice in product: quality — and identity — are formed by staying accountable to a specific counterpart over time.

Right now that work is my full-time chapter — and I'm open to my next product role.

Selected work

Ink-on-bone engraving of a branching quiz flow being rebuilt by an indigo chat thread.

Case study

Rebuilding the Quiz editor used by thousands of Shopify brands as AI-native

How I diagnosed the problem with my own first-use recording and an NPS survey, then led a ground-up redesign of every surface: the canvas, components, live preview, and an AI assistant that edits by chat.

Read

Ink-on-bone engraving of a magnifying glass and balance scale over a page of text — instruments of critical reading.

Featured project

Jiddu — read any text with a critical mind

Four tools in one for reading critically: a fallacy detector, a fact-check pipeline, a neutrality assessment, and a paper explainer. It runs in the browser, over a REST API, and through a hosted MCP server, grounded in two recent research papers.

See the project

Ink-on-bone engraving: six small framed plates, each with a distinct emblem, connected in sequence by a single indigo thread — six papers bound into one research program.

Conceptual AI research

Six papers on language agents, memory, and individuation

Independent conceptual work on what language agents inherit across model successions, whether scale alone yields a persistent self, how perspective forms, and why stable deference may need a self-model first.

Read the papers