Human-centered AI
Human-Centered Reflection
Explore how structured interfaces can help people examine complex experiences while maintaining control over their own interpretations and decisions.
Early-stage reflection project
Reflexion
Structured reflection for clearer thinking.
Reflexion is an early-stage, non-clinical AI-assisted reflection project that explores how complex personal narratives can be organized into events, patterns, boundaries, and questions for further reflection.
The public demo illustrates the intended workflow using sample data. Live AI analysis is not yet available.
Structure
Events, causes, patterns
Boundary
What is mine, what is not
Direction
Judgment and next steps
Demo flow: narrative input to structure to summary
About Reflexion
Explore the reflection workflow
The sample-data demo previews three steps: describe an experience, review its structure, and consider follow-up questions. Outputs are illustrative and do not analyze what you enter.
01
Name what happened
Start from a user-authored narrative, not a diagnosis or mood score.
02
Separate the pieces
Organize events, causes, constraints, repeated patterns, and agency boundaries.
03
Consider follow-up questions
Review sample prompts about context, boundaries, and possible choices.
Research direction
Why It Matters
These research directions guide future development. Their reliability and usefulness have not yet been formally evaluated in Reflexion.
Trustworthy AI
Trustworthy AI
Investigate how reliability, transparency, uncertainty awareness, and clear safety boundaries can be incorporated into AI-assisted reflection systems.
Responsible Evaluation
Responsible Evaluation
Develop and evaluate structured reflection workflows through usability testing, human feedback, and safety-oriented assessment without making unsupported clinical or therapeutic claims.
The project does not diagnose, treat, or claim clinical authority. It studies structured, non-clinical interpretation of difficult affective and narrative material.
Research Publication
Reflexion as an early HCI / affective computing prototype
arXiv:2504.20342v2Early prototypeNo efficacy claims
Narrative-Centered Emotional Reflection
An Early Prototype for AI-Supported Emotional Self-Reflection
Reflexion is presented in the arXiv paper as an early conversational AI prototype for structured emotional self-reflection, integrating emotion detection, layered reflective prompting, and metaphorical storytelling generation.
The current version frames the system as a theory-informed HCI / affective computing prototype and does not make empirical efficacy claims.
Framed as conversational AI and structured self-reflection, not therapy, diagnosis, or clinical intervention.
Demo Workflow
A preview of the intended experience
Each view below uses sample data to illustrate the proposed workflow. Live AI analysis, saved reports, and analysis across sessions are future capabilities.
Input
Guided narrative input
Prompts help users describe what happened in their own words without reducing the experience to a mood score.
Order
Event and signal organization
Sample outputs illustrate how events, concerns, and language patterns could be organized. Live extraction is planned.
Pattern
Cause and consequence mapping
Sample mappings illustrate possible causes, constraints, and repeated themes. They are not generated from your input.
Boundary
Agency-boundary prompts
Follow-up prompts help users distinguish what they feel, what happened, what is within agency, and what is not theirs to carry.
Output
Structured reflection reports
Sample summaries preview the intended report structure. Generating and saving personal reports are planned.
Review
Review over time
Monthly and yearly views display sample patterns. Tracking and analyzing reflections across sessions are planned.
Project Team
Reflexion AI Team
Human-centered AI research and product development team
Reflexion is developed as a project-centered structured reflection prototype focused on narrative sensemaking, pattern recognition, agency boundaries, and inner order.
Structured reflectionNarrative sensemakingAgency boundariesNon-clinical interface design
Project Progress
A project-centered view of public-facing progress
01
Research direction
Explore how non-clinical AI can help structure difficult narratives into events, patterns, boundaries, and possible next steps.
02
Visible progress
Public website, interactive prototype, mobile workflow, signal summaries, pattern outputs, and roadmap concepts are already represented.
03
Development path
Backend reliability, privacy-conscious usage infrastructure, user feedback, and clearer product boundaries are the next implementation priorities.
Technical Status
A visible prototype with a clear path toward connected reflection
Reflexion currently demonstrates a fixed-data structured reflection workflow. The next phase focuses on live analysis, saved reports, privacy controls, and web/mobile continuity.
Sample-data demo
Mobile-first reflection workflow
Guided input screens, category selection, structured summaries, and export concepts are represented in the public demo experience.
Sample-data demo
Structured reflection framework
Sample outputs illustrate events, context, boundaries, and repeated themes. Automated extraction and analysis are not available in the public demo.
Planned
Backend and API integration
Live model calls, reflection storage, usage controls, and privacy-conscious data handling are planned. Update-list sign-in is separate from reflection accounts.
Future research
Evaluation and feedback pathway
Usability, human feedback, output reliability, and safety-boundary assessment are future evaluation directions. No study recruitment is open.
Prototype boundary
The public demo shows the intended workflow. It is not yet a live backend, clinical tool, or full account-based product.
Public landing page and prototype demo are live under the main site experience.
Current demo uses a fixed-data workflow to show how narrative input can become structured reflection output.
The roadmap separates live behavior from planned backend, web/mobile sync, and report features.
FAQ
Technical purpose, prototype boundaries, and public-facing development direction.
Overview
What is Reflexion?
Reflexion is an early-stage, non-clinical AI-assisted reflection project. The public demo illustrates an intended workflow using sample data; live AI analysis is not yet available.
Overview
What problem does it study?
Reflexion studies how AI can help organize complex lived experience into events, context, causes, agency boundaries, recurring patterns, and possible next actions.
Safety
Is Reflexion a clinical or diagnostic tool?
No. Reflexion does not diagnose, treat, assess clinical risk, provide medical advice, or claim clinical authority. It is designed as a non-clinical interface and analysis prototype.
Technology
What is the technical foundation?
The sample-data demo previews narrative structuring, prompt routing, and summaries. Live AI analysis is planned; reliability and safety evaluation remain future research.
Technology
How is it different from mood trackers?
Reflexion explores a narrative-based workflow for reviewing events, context, patterns, boundaries, and questions. The public demo illustrates that approach with sample data rather than analyzing user input.
Impact
How does Reflexion relate to public interest?
Reflexion addresses a public-interest gap in human-facing AI: clearer, non-clinical interfaces for organizing sensitive affective and narrative material without clinical overreach.
Impact
How will the workflow be evaluated?
Future evaluation will examine usability, human feedback, reliability, uncertainty awareness, and safety boundaries. The study design and timing are not finalized, and recruitment is not open.
Impact
How is Reflexion connected to trustworthy AI?
Because Reflexion works with affective content, reliability, interpretability, privacy, uncertainty-aware wording, and non-clinical boundaries are central to its design.
Status
Is this just a prototype?
Reflexion is a visible prototype demonstrating the mobile-first narrative workflow, signal summaries, export concepts, and structured reflection framework. Live AI analysis and reflection storage remain planned.
Team
Who develops Reflexion?
Reflexion is developed by a human-centered AI research and product development team focused on structured reflection, narrative sensemaking, and non-clinical interface design.
Connect with Reflexion
Demo
Try the interactive prototype
Explore the guided narrative workflow, signal summaries, and pattern views.
Open Demo
UpdatesFollow project progress
Read external press, public announcements, and Reflexion development milestones.
View News
ContactContact the Reflexion AI Support Team
For collaboration, product feedback, or questions about the project.
Email Support Team
Reflexion is a non-clinical structured reflection prototype. It does not provide diagnosis, treatment, medical advice, or crisis assessment.