Portraitor: the analysis prompt and content engine behind a private by design AI product
Portraitor reads a chat conversation a user chooses to share and returns a written portrait of communication and behaviour patterns, built on established psychological frameworks, without ever storing the conversation. Michael S.U. Hudson designed the analysis prompt at its core and built the content engine behind its growth.
Outcome
0
Conversations, portraits, accounts or email addresses held by Portraitor.
Available
Web, App Store and Google Play
Source
Portraitor Privacy Policy, version 1.3, in effect from 30 September 2026
The problem
General purpose AI models give the same personality answer again and again, and usually miss the person. People who want to understand how they communicate also need to know that what they share is not kept.
Role
Michael created the analysis prompt Portraitor runs on, then tested and optimised it until it performed reliably with the underlying models. He built the content engine that drives Portraitor’s growth and helped create the website. Title: Prompt Architect and Content Engine Lead.
What was built
A privacy filter that runs on the user’s own device and replaces names, email addresses, phone numbers, postal addresses, links and credentials with anonymous placeholders before anything is sent. The key that maps them back never leaves the device. If masking fails, nothing is sent and the payment is refunded automatically.
An analysis prompt that reads the conversation through four lenses: relationship context, the Big Five personality model, Transactional Analysis, and change over time. Every finding carries a confidence score and its stated limitations.
A content engine that publishes to the Portraitor blog every day.
Outcome
Portraitor is live. Analysis is fully automated; no person at Portraitor can read a conversation or a portrait. Portraits are stored only on the user’s device. UK and EU data rights are extended to every user worldwide.
What it shows
A fixed, tested prompt grounded in established frameworks gives more consistent and more honest results than open conversation with a general model, and privacy can be a design property rather than a policy.