Insurtech · AI · Digital Identity

mirror gpt: De-risking behavior through reflection

Origin: Pre-seed startup venture
Industry: Insurance
Partners: Qatar Insurance Group, Prefeena
My role: Founder · Strategic Designer · Venture Lead
Stage: Concept → Prototype → Partner validation → Investor pitching
Year: 2022–2023

Why this project started

Mirror GPT did not start as a client project. It started as a pre-seed startup hypothesis.

The core question was simple but provocative:

What if people could meaningfully talk to their own digital identities — not just to AI trained on others’ data?

At a time when AI systems were increasingly trained on aggregated, external datasets, we wanted to explore a different direction: AI as a reflective interface to one’s own data.

The hypothesis

We believed that:

  • People are surrounded by fragmented digital selves

  • Data exists, but meaning is missing

  • Reflection could become a new value layer

The hypothesis was that structured self-dialogue with personal data could unlock:

  • Reduced risk in high-stakes or habitual behaviors

  • More grounded, context-aware decision-making

  • Deeper self-understanding and self-empathy

  • New foundations of trust between humans and AI

Why Insurance Was Explored

Insurance was explored as a strategic context because its business model is directly shaped by human behavior over time.

Qatar Insurance Group sits at a unique intersection where:

  • Individual decisions accumulate into long-term risk

  • Life transitions influence exposure and vulnerability

  • Trust determines both customer loyalty and compliance

Our hypothesis was that supporting reflection could reduce risky behavior, which in turn could:

  • Improve risk profiles across customer segments

  • Enable more meaningful, trust-based risk communication

  • Increase long-term value through prevention rather than reaction

  • Strengthen internal decision-making in uncertain environments

What Mirror GPT was (and wasn’t)

Mirror GPT was not:

  • A chatbot giving advice

  • A prediction engine

  • A wellness or coaching app

Mirror GPT was designed as:

  • A reflective system that mirrors language, assumptions, and patterns

  • A way to explore how someone thinks, not what they should do

  • A bridge between personal data and meaning

AI’s role was not to decide — but to help users see themselves more clearly.

Data ownership & the Prefeena partnership

A critical pillar of the concept was data sovereignty.

We partnered with Prefeena to enable:

  • Secure, user-controlled data storage

  • Explicit consent over which data is used

  • Separation between AI interaction and raw data ownership

This allowed us to prototype a future where:

AI works with your data — without owning it.

What we built

  • Early conceptual architecture of reflective AI

  • Interaction flows for “talking to your digital identities”

  • Prototype scenarios combining stored personal data + AI reflection

  • Pitch decks and narratives for partners and investors

  • Strategic framing workshops with insurance stakeholders

The focus was learning — not scaling.

Validation & outcome

Through partner discussions, internal exploration, and investor pitching, we reached a clear conclusion:

While the concept showed strong philosophical and experiential value,
it did not demonstrate a clear strategic advantage for insurance to build at that time.

This was a valid negative outcome — and an important one.

The project was consciously not pushed further, avoiding solutionism without real pull.

What the project proved

Mirror GPT successfully demonstrated that:

  • People are willing to engage in reflective dialogue with AI

  • Talking to one’s own data feels fundamentally different than generic AI chat

  • The value lies in sense-making, not optimization

What it did not prove:

  • A sustainable insurance-specific business model

  • Clear short-term ROI for a regulated, risk-averse industry

My role as founder

  • Defined the venture hypothesis and narrative

  • Led partner discovery and validation

  • Designed the reflective AI interaction model

  • Translated abstract ideas into prototypes and pitches

  • Made the decision to stop when strategic value was not proven

Knowing when not to continue was part of the work.

Reflection

Mirror GPT deeply shaped my later thinking.

It reinforced a principle that continues to guide my work:

Not every meaningful idea should become a product —
but every tested idea should leave us with better understanding.

Many of the concepts explored here — reflection, orientation, identity, and AI as a mirror — continue to live on in my research, writing, and design practice

Previous
Previous

Three Years of Continuous Patient Journey Improvement

Next
Next

Tree rings