Lyminal Methodologies
Lyminal Methodologies
Lyminal Methodologies
From signal
to transformation.
to transformation.
to transformation.
to transformation.
to transformation.
to transformation.
We move from research to results with imagination, structure, and speed.
Our Hypothesis: When daily signals such as check-ins, conversations, reflections, and goals are guided through models shaped by cognitive science, hidden patterns emerge. We reveal states, surface triggers, and return guidance that awakens awareness and supports real change.
Methodology
Methodology
Methodology
Technology & psychology, made tangible
We give form to the invisible. Check-in rituals, protocol cards, calming soundscapes, dashboards, and even transparent objects become ways to see and touch the inner workings of the system.
We give form to the invisible. Check-in rituals, protocol cards, calming soundscapes, dashboards, and even transparent objects become ways to see and touch the inner workings of the system.
We give form to the invisible. Check-in rituals, protocol cards, calming soundscapes, dashboards, and even transparent objects become ways to see and touch the inner workings of the system.
In Practice: Inputs → engine → outputs
Inputs
01
We begin by noticing the small moments that shape the day.
Mood + brief context
Conversation content & tone
Declared goals & progress pings
Trigger & safety signals
Inputs
01
We begin by noticing the small moments that shape the day.
Mood + brief context
Conversation content & tone
Declared goals & progress pings
Trigger & safety signals
Inputs
01
We begin by noticing the small moments that shape the day.
Mood + brief context
Conversation content & tone
Declared goals & progress pings
Trigger & safety signals
Analysis Engine
02
Signals gather inside the engine and are translated into understanding.
Detects patterns & triggers over time
Estimates state from multimodal signals
Matches techniques to the moment
Prioritizes safety & escalation as needed
Analysis Engine
02
Signals gather inside the engine and are translated into understanding.
Detects patterns & triggers over time
Estimates state from multimodal signals
Matches techniques to the moment
Prioritizes safety & escalation as needed
Analysis Engine
02
Signals gather inside the engine and are translated into understanding.
Detects patterns & triggers over time
Estimates state from multimodal signals
Matches techniques to the moment
Prioritizes safety & escalation as needed
Outputs
03
What returns is tailored guidance designed for right now.
Timely reframes and micro-skills
Personalized resources and practices
Timely nudges and habit scaffolding
Progress reflections: Archive of Feelings
Outputs
03
What returns is tailored guidance designed for right now.
Timely reframes and micro-skills
Personalized resources and practices
Timely nudges and habit scaffolding
Progress reflections: Archive of Feelings
Outputs
03
What returns is tailored guidance designed for right now.
Timely reframes and micro-skills
Personalized resources and practices
Timely nudges and habit scaffolding
Progress reflections: Archive of Feelings
Methodology
Methodology
Methodology
Our Research Lens
(DV = f(IV))
Independent variables (IV):
Daily mood/context • Conversation tone • Q&A responses • Declared goals • Interaction volume • Trigger events • Overwhelm signals • Safety/risk signals • Computed state estimates • Technique-matching parameters
Independent variables (IV):
Daily mood/context • Conversation tone • Q&A responses • Declared goals • Interaction volume • Trigger events • Overwhelm signals • Safety/risk signals • Computed state estimates • Technique-matching parameters
Independent variables (IV):
Daily mood/context • Conversation tone • Q&A responses • Declared goals • Interaction volume • Trigger events • Overwhelm signals • Safety/risk signals • Computed state estimates • Technique-matching parameters
Dependent variables (examples):
Self-awareness index • Goal completion & habit progress • Engagement & retention • Frequency/duration of negative states • Short-term mood regulation • Skill acquisition & application • Appropriate routing to human help
Dependent variables (examples):
Self-awareness index • Goal completion & habit progress • Engagement & retention • Frequency/duration of negative states • Short-term mood regulation • Skill acquisition & application • Appropriate routing to human help
Dependent variables (examples):
Self-awareness index • Goal completion & habit progress • Engagement & retention • Frequency/duration of negative states • Short-term mood regulation • Skill acquisition & application • Appropriate routing to human help
People & Safety
People-in-the-loop design
We design with advisors and participants, testing for both accuracy and the feeling of being truly helpful. Consent is clear, feedback shapes the product, and contributions are recognized.
Safety & ethics
Safety is the foundation of our work. Crisis pathways are never tested, interactions remain brief and respectful, behaviors are transparent, and privacy is preserved through minimization, plain language, and strict controls.
Lyminal builds safe and transparent systems where exploration can flourish without harm.
Frequently Asked Questions
Product & Participation
Methods & Evidence
Data, Privacy & Safety
What is Lyminal, and how is it different from a typical mental-health app?
Lyminal is an R&D lab. We build transparent, low-burden tools that make mental processes observable and actionable, then we keep only what proves helpful. Unlike most apps, we favor glass-box explanations, brief interactions, and publishable results over opaque features and endless engagement loops.
How do I participate—call, app, or Discord—and can I switch later?
Join any H.E.Y. pilot via phone (HeyBro), iOS (HeyThere), or Discord (HeyChat). You can switch interfaces anytime; your preferences, consent, and data carry with you so the experience stays personalized.
How much time does it take each day, and can I pause or snooze nudges?
Typical use is 30–60 seconds/day (2–5 minutes for a call). You control quiet hours, snooze, and frequency caps. We design for minimal burden and never spam—nudges are capped and easy to turn off.
What are the H.E.Y. experiments, and how do I join a pilot?
H.E.Y. (Hybrid • Enhancement • Yield) is our series of real-world experiments across call/app/Discord. Sign up for the waitlist, complete consent, and you may be assigned to a study condition (e.g., different prompt timing). Safety flows never change, and you can withdraw at any time.
How does Lyminal define and measure “behavior change”?
We track outcomes such as self-awareness, goal completion, short-term mood shifts after nudges, frequency/duration of negative states, engagement/retention, and skill use. We analyze effect sizes with confidence intervals and report both intention-to-treat and per-protocol results.
What does “glass-box” or explainable modeling mean here?
Suggestions come with plain-language rationale (“Why this?”), key inputs (e.g., recent patterns), and model confidence where appropriate. The aim: you can see and question how guidance is produced, not just receive it.
How do experiments affect my experience?
You might receive different prompt wording or timing (e.g., fixed time vs just-in-time). Burden limits, privacy rules, and safety routing are identical for all groups. Participation is consent-based, and you can opt out without penalty.
Where can I read your results—methods notes or journal links?
See the Methodology page for study designs and metrics, and the About page for research notes, CEO letter, and journal/preprint links. We publish summaries in accessible language and share deeper technical appendices when available.
Do you specialize in any particular areas?
Absolutely. In fact, many of our most successful projects are built on close collaboration with internal R&D, data science, or innovation units. We integrate seamlessly, offering fresh perspectives while respecting existing knowledge and workflows. Our role is to complement, not replace.
What signals do you capture and why?
With consent, we collect brief check-ins, short chats, quick Q&A, declared goals, interaction patterns (streaks/lapses), and trigger/safety signals. These support state estimation and technique matching so guidance is timely and relevant.
How do you minimize data and keep it secure?
We practice data minimization, encrypt data in transit/at rest, and restrict access by role. We prefer derived features over raw content where possible and use de-identified aggregates for research. You can review settings and opt out of nonessential collection.
What happens if a safety/risk signal is detected?
We activate conservative routing: surface crisis resources, encourage contacting trusted supports, and—where available—offer warm handoffs. Safety pathways are never A/B tested, and we throttle or halt nonessential nudges during elevated risk.
How can I review, export, or delete my data—or leave a study?
Use the in-product privacy dashboard (or contact us) to review, export (e.g., CSV/JSON), or delete your data. You can withdraw from a study at any time; we confirm when de-identification or deletion is complete, subject to any legal retention requirements.
Product & Participation
Methods & Evidence
Data, Privacy & Safety
What is Lyminal, and how is it different from a typical mental-health app?
Lyminal is an R&D lab. We build transparent, low-burden tools that make mental processes observable and actionable, then we keep only what proves helpful. Unlike most apps, we favor glass-box explanations, brief interactions, and publishable results over opaque features and endless engagement loops.
How do I participate—call, app, or Discord—and can I switch later?
Join any H.E.Y. pilot via phone (HeyBro), iOS (HeyThere), or Discord (HeyChat). You can switch interfaces anytime; your preferences, consent, and data carry with you so the experience stays personalized.
How much time does it take each day, and can I pause or snooze nudges?
Typical use is 30–60 seconds/day (2–5 minutes for a call). You control quiet hours, snooze, and frequency caps. We design for minimal burden and never spam—nudges are capped and easy to turn off.
What are the H.E.Y. experiments, and how do I join a pilot?
H.E.Y. (Hybrid • Enhancement • Yield) is our series of real-world experiments across call/app/Discord. Sign up for the waitlist, complete consent, and you may be assigned to a study condition (e.g., different prompt timing). Safety flows never change, and you can withdraw at any time.
How does Lyminal define and measure “behavior change”?
We track outcomes such as self-awareness, goal completion, short-term mood shifts after nudges, frequency/duration of negative states, engagement/retention, and skill use. We analyze effect sizes with confidence intervals and report both intention-to-treat and per-protocol results.
What does “glass-box” or explainable modeling mean here?
Suggestions come with plain-language rationale (“Why this?”), key inputs (e.g., recent patterns), and model confidence where appropriate. The aim: you can see and question how guidance is produced, not just receive it.
How do experiments affect my experience?
You might receive different prompt wording or timing (e.g., fixed time vs just-in-time). Burden limits, privacy rules, and safety routing are identical for all groups. Participation is consent-based, and you can opt out without penalty.
Where can I read your results—methods notes or journal links?
See the Methodology page for study designs and metrics, and the About page for research notes, CEO letter, and journal/preprint links. We publish summaries in accessible language and share deeper technical appendices when available.
Do you specialize in any particular areas?
Absolutely. In fact, many of our most successful projects are built on close collaboration with internal R&D, data science, or innovation units. We integrate seamlessly, offering fresh perspectives while respecting existing knowledge and workflows. Our role is to complement, not replace.
What signals do you capture and why?
With consent, we collect brief check-ins, short chats, quick Q&A, declared goals, interaction patterns (streaks/lapses), and trigger/safety signals. These support state estimation and technique matching so guidance is timely and relevant.
How do you minimize data and keep it secure?
We practice data minimization, encrypt data in transit/at rest, and restrict access by role. We prefer derived features over raw content where possible and use de-identified aggregates for research. You can review settings and opt out of nonessential collection.
What happens if a safety/risk signal is detected?
We activate conservative routing: surface crisis resources, encourage contacting trusted supports, and—where available—offer warm handoffs. Safety pathways are never A/B tested, and we throttle or halt nonessential nudges during elevated risk.
How can I review, export, or delete my data—or leave a study?
Use the in-product privacy dashboard (or contact us) to review, export (e.g., CSV/JSON), or delete your data. You can withdraw from a study at any time; we confirm when de-identification or deletion is complete, subject to any legal retention requirements.
Product & Participation
Methods & Evidence
Data, Privacy & Safety
What is Lyminal, and how is it different from a typical mental-health app?
Lyminal is an R&D lab. We build transparent, low-burden tools that make mental processes observable and actionable, then we keep only what proves helpful. Unlike most apps, we favor glass-box explanations, brief interactions, and publishable results over opaque features and endless engagement loops.
How do I participate—call, app, or Discord—and can I switch later?
Join any H.E.Y. pilot via phone (HeyBro), iOS (HeyThere), or Discord (HeyChat). You can switch interfaces anytime; your preferences, consent, and data carry with you so the experience stays personalized.
How much time does it take each day, and can I pause or snooze nudges?
Typical use is 30–60 seconds/day (2–5 minutes for a call). You control quiet hours, snooze, and frequency caps. We design for minimal burden and never spam—nudges are capped and easy to turn off.
What are the H.E.Y. experiments, and how do I join a pilot?
H.E.Y. (Hybrid • Enhancement • Yield) is our series of real-world experiments across call/app/Discord. Sign up for the waitlist, complete consent, and you may be assigned to a study condition (e.g., different prompt timing). Safety flows never change, and you can withdraw at any time.
How does Lyminal define and measure “behavior change”?
We track outcomes such as self-awareness, goal completion, short-term mood shifts after nudges, frequency/duration of negative states, engagement/retention, and skill use. We analyze effect sizes with confidence intervals and report both intention-to-treat and per-protocol results.
What does “glass-box” or explainable modeling mean here?
Suggestions come with plain-language rationale (“Why this?”), key inputs (e.g., recent patterns), and model confidence where appropriate. The aim: you can see and question how guidance is produced, not just receive it.
How do experiments affect my experience?
You might receive different prompt wording or timing (e.g., fixed time vs just-in-time). Burden limits, privacy rules, and safety routing are identical for all groups. Participation is consent-based, and you can opt out without penalty.
Where can I read your results—methods notes or journal links?
See the Methodology page for study designs and metrics, and the About page for research notes, CEO letter, and journal/preprint links. We publish summaries in accessible language and share deeper technical appendices when available.
Do you specialize in any particular areas?
Absolutely. In fact, many of our most successful projects are built on close collaboration with internal R&D, data science, or innovation units. We integrate seamlessly, offering fresh perspectives while respecting existing knowledge and workflows. Our role is to complement, not replace.
What signals do you capture and why?
With consent, we collect brief check-ins, short chats, quick Q&A, declared goals, interaction patterns (streaks/lapses), and trigger/safety signals. These support state estimation and technique matching so guidance is timely and relevant.
How do you minimize data and keep it secure?
We practice data minimization, encrypt data in transit/at rest, and restrict access by role. We prefer derived features over raw content where possible and use de-identified aggregates for research. You can review settings and opt out of nonessential collection.
What happens if a safety/risk signal is detected?
We activate conservative routing: surface crisis resources, encourage contacting trusted supports, and—where available—offer warm handoffs. Safety pathways are never A/B tested, and we throttle or halt nonessential nudges during elevated risk.
How can I review, export, or delete my data—or leave a study?
Use the in-product privacy dashboard (or contact us) to review, export (e.g., CSV/JSON), or delete your data. You can withdraw from a study at any time; we confirm when de-identification or deletion is complete, subject to any legal retention requirements.
Still have questions? Get in touch with our support team and we'll discuss your unique requirements.
Welcome to Lyminal
Welcome to Lyminal
Welcome to Lyminal