A system underneath.
The research that informs MyWhatIf, the mechanisms we are designing around, and what we are actively testing.
Trauma changes prediction. Danger feels more likely, possibility feels less real, and the future becomes harder to imagine.
This is the part of trauma that measurement and treatment reach least often. PTSD and depression affect how people see themselves, interpret the present, and imagine what is ahead. People living with PTSD often find it harder to construct positive future scenarios at all, and that difficulty tracks with symptom burden.
Can the capacities involved in imagining and moving toward a possible future be deliberately activated?
Everything below is organized around it: what the literature establishes, what we hypothesize, and what our research program is testing.
Why hope, agency and purpose? Because in the research they are not moods. They are capacities with structure, and two of the three have been studied as such for decades.
In Snyder's hope theory, hope is not optimism. It is the combination of agency thinking, the belief that one's actions can influence outcomes, and pathways thinking, the ability to identify routes toward a goal. That is why our first two targets are inseparable: agency is a component of hope as the literature defines it, not a second virtue added beside it.
A different future can feel possible.
Hope has been identified as a protective factor associated with greater resilience and lower PTSD symptoms, and in trauma-focused treatment both baseline hope and gains in hope have predicted symptom reduction.
I can influence what happens next.
Agency thinking is one of the two components of hope in Snyder's model, and choosing one's own question is itself an exercise of it. Its independent relationship to trauma outcomes is less directly established.
There is something worth moving toward.
Meaning and purpose sit at the center of post-traumatic growth research and of narrative approaches to recovery. We include purpose because a future has to be worth moving toward, and we are honest that our evidence for it is the least direct of the three.
We would rather show the asymmetry than manufacture symmetry.
The metaphor is a switch. The biology is a network.
There is no hope center in the brain and no switch. We keep the word because it makes one idea graspable to someone who is not a neuroscientist: a capacity that has gone quiet is a capacity, not a character flaw.
What sits under the metaphor is a set of interacting systems implicated in episodic future thinking, self-referential processing, valuation, goal-directed cognition, motivation, emotion regulation and meaning. Imagining a future and remembering a past recruit overlapping networks, which is part of why trauma can affect both.
No single center. The brain's prediction model is based on multiple interacting systems involved in future-oriented thinking, regulation, motivation, purpose and meaning.
That repeated, personalized narrative experience can help re-engage this capacity of hope and future thinking, and that the change shows up in validated measures. Proposed, not yet demonstrated in RCTs. It is what the trials exist to test.
A literature is not an intervention. Turning one into the other means deciding what to understand about a person, what to offer them, what to measure, and what to refuse to do.
Language is the human operating system. Every person runs on a unique internal code, built from memory, meaning, beliefs, values, relationships, culture and imagination.
If those patterns can be understood, an intervention can respond to the individual rather than the average. That is the design ambition MyWhatIf operates from.
We use the language of code because it makes one idea legible: a pattern that keeps producing the same output can be examined and changed. People are not software, and the metaphor stops where the person begins.
A structured map of a Hero's narrative and psychological patterns: memory, meaning, beliefs, values, relationships, culture and imagination. Every person runs on a unique version of it. Today it is built mainly from words and voice; facial expression and behavioral signals are still being developed and validated through our research program.
The Hero's Emotional Relationship Structure: a non-generative, predictive framework that reads a Hero's Human Code and decides what kind of narrative movement is appropriate for them now, so personalization holds across a whole arc rather than a single reply.
The Basic Narrative Ontology: the narrative-generation model that operates inside HERS, writing language within the bounds HERS sets rather than deciding on its own where a story goes.
Human Code describes why every person is different. HERS is how MyWhatIf responds to that difference.
Personalization that holds across a whole arc rather than a single reply. What HERS allows is the part that matters here; how it is engineered is proprietary, and publishing it would not make the case more credible.
A model can write the language. It does not get to decide the trajectory.
In an open-ended chatbot, the model that writes the prose effectively decides where things go next. MyWhatIf is built the other way round: generation is separated from control. Language models do specific jobs, and what kind of narrative movement is appropriate for this person now is determined outside them, by our own understanding of the Hero together with separate personalization, prediction, safety and feedback mechanisms.
It is also what distinguishes MyWhatIf from therapy-oriented bots, which use conversational technology to deliver, guide or support an established therapeutic framework. Here, generative AI sits inside a purpose-built narrative intervention architecture organized around understanding the individual, future possibility, hope, agency and purpose, personalized narrative simulation, participant response, measurement and controlled adaptation. AI is used to understand emotional and narrative patterns, to personalize the intervention and to support measurement.
Early internal validation reported symptom and hope improvements. Self-report, no control group. Signal, not proof.
Multi-site randomized controlled trials with academic and clinical partners, governed by a Clinical Development Board, with safety governance, risk management and quality systems built alongside them.
Whether the intervention safely produces meaningful, durable change in these measures, and for whom.
We don’t stop at whether it felt good. We ask whether anything changed.
Measurement was designed in from the start, using validated instruments rather than instruments of our own: the PCL-5 for PTSD symptoms, the Snyder Hope Scale for hope, and the ACE questionnaire for adverse childhood experience. A subset of participants may take part in fMRI substudies examining neural changes associated with narrative reframing. We optimize for outcomes, not engagement, not screen time, not profit.
A regulatory pathway is part of this work rather than a precondition for it. MyWhatIf can be offered as a non-clinical hope intervention where that is appropriate; FDA authorization matters for regulated clinical contexts, reimbursement, and some government and institutional deployments. It has not been granted, and it is not assumed.
Safety here is a design constraint, not a disclaimer. It shapes what the system is allowed to do, and it stretches beyond the model to clinical judgment, privacy, and the human beings around the platform.
Safety is not something we expect AI to provide alone.
Four distinct dimensions. They interact. They are not interchangeable.
A Clinical Development Board of trauma-recovery clinicians and researchers challenges assumptions, methodology and clinical direction. Studies run with academic and clinical partners under the appropriate IRB structures. Contraindications and escalation are defined, not improvised.
The experience is a defined intervention with a defined shape, which is what makes it reviewable. HERS constrains what the narrative can become; BNO writes language within those bounds. The system does not require anyone to repeatedly recount the traumatic event; avoiding unnecessary re-exposure is a principle the Foundation was built on.
MyWhatIf is HIPAA compliant. You can see what is held about you and ask us to delete your data when you are done. Sensitive information is separated, access is controlled, and privacy and cybersecurity are funded workstreams, not afterthoughts.
Human review when needed, escalation to appropriate professional and crisis resources, and pathways to peer and community support where appropriate. Peer support is not clinical care, and neither is emergency care. MyWhatIf is not emergency care; anyone in immediate danger needs local emergency services.
As part of the human-support layer, MyWhatIf is developing pathways for situations in which technology should not be the only response. STAR Network Foundation, a peer-support and suicide-prevention collaborator, has offered to participate in the development of a formal crisis-response pathway, drawing on its lived-experience communities and suicide-prevention experience. Programs such as TAR Anon and Caregiver Connections illustrate the peer/community infrastructure this collaboration draws on.
Clinical efficacy is being evaluated and has not yet been established.
The boundary between what is known and what we hope is the most important line on this page.
Cited here as the scientific context our model is built on. MyWhatIf does not deliver narrative exposure therapy, and none of these studies evaluates MyWhatIf.
Studies, sites, partners and status.
What we measure now, and what we are working to demonstrate.
For researchers, health systems, universities and AI teams.
Learn about opportunities to take part through our studies and partner programs.
