The Play Date and all nine follow-up modules, redesigned for delivery by Yismach's voice AI system. The science is unchanged. The exercises are unchanged. The human who used to run the room has been replaced by something that never gets tired, never misses a cue, and is available anywhere in the world at any hour.
Every session in this curriculum is delivered by Yismach's voice AI agent — a warm, calm, Hebrew- and English-fluent voice that participants interact with through their phones or a shared speaker in the room. The AI does everything a human facilitator does: welcomes participants, delivers exercise instructions, tracks time, manages transitions, provides the "why this matters" framing, assigns homework, and follows up between sessions.
Three things the AI does that a human facilitator cannot: it is available at any hour in any location, it maintains perfect consistency across every session it runs, and it builds a continuous model of each participant's relational patterns across every interaction — from the intake through the Play Date through every module — refining its understanding without forgetting a word.
One thing the AI handles differently than a human: the co-regulation function. A calm human presence in a room produces physiological effects — through voice, gaze, and body language — that stabilize anxious nervous systems. The AI approximates this through voice tone calibration, pacing, and ambient sound design, but the approximation is not identical to the biological reality. The system addresses this honestly: the AI-facilitated environment is slightly more cognitively demanding for highly anxious participants, and the distress detection and escalation protocols described below are the compensation.
The AI monitors voice tone, response latency, and keyword patterns in real time throughout every session. When distress signals exceed threshold — silence after an emotionally loaded prompt, vocal indicators of acute anxiety, or explicit statements of overwhelm — the AI pauses the exercise, acknowledges what it detected, and offers three options: continue at a slower pace, skip to the next exercise, or connect immediately with a human Yismach coordinator. The coordinator pathway connects within 90 seconds. No session ends without the AI confirming participant wellbeing. Any distress flag is logged and surfaced to the Yismach team within 24 hours.
The voice agent runs on Yismach's Love Map infrastructure. Every spoken response during sessions is processed against the participant's existing Love Map profile — updating relational pattern models in real time. Pairings, exercise sequencing, and post-session follow-up are all informed by this continuous data layer. Participants interact with a single consistent AI voice across all sessions — the same voice that conducted their intake, their Play Date, and every module. Continuity is by design.
First. Play. Then the date.
The Play Date is a structured, AI-facilitated evening for 12–16 participants — equal numbers of men and women. The AI voice agent runs the entire evening through a shared speaker system in the room plus individual earpieces for private exercise instructions. Pairings are determined before participants arrive, based on Love Map cross-matching. Participants experience the evening as a series of guided exercises. The matching is invisible. The encounter is real.
Before the first participant arrives, the AI has already done three things. It has completed a deep intake with every participant — building their Love Map, capturing not just stated preferences but relational scripts, attachment patterns, and the gap between what they say they want and how they describe what has gone wrong. It has cross-matched every participant against every other. And it has generated a pairing sequence — a ranked list of highest-potential encounters, structured to give each participant at least one high-probability pairing during the evening.
The room itself is prepared according to AI-generated instructions sent to the venue coordinator in advance: soft lighting, no long tables, no name tags, stations arranged for paired exercises, ambient sound at a calibrated volume. The AI cannot adjust the room in real time. The room must be right before the evening starts.
"Welcome. I'm glad you're here. Tonight isn't a mixer, and it isn't a test. It's a series of exercises — things we're going to practice together. You don't need to perform. You don't need to impress anyone. Your only job tonight is to show up without the mask. I'll guide you through everything. Let's begin."
Before any cross-gender interaction, the AI runs same-gender warm-up groups through individual earpieces — separate audio streams for men and women in adjacent spaces. The warm-ups activate the ventral vagal state: the neurophysiological condition in which genuine social engagement becomes possible. The AI monitors voice tone during warm-ups and adjusts the pacing if anxiety levels remain elevated. No one moves to the mixed exercises until the AI confirms the group is ready.
"Before we begin, a few things I want you to know. Everything that happens in this room stays in this room — the AI system holds no shareable record of what you say to your partner tonight. You can pass on any exercise at any time by saying 'skip' aloud — I'll hear it and move us forward. If at any moment you need a break, say 'pause' and I'll stop the clock and check in with you privately. You are in control of your experience tonight."
People form lasting judgments in the first five seconds. Most of these judgments are wrong — but they are powerful, and they are almost never examined. The AI delivers the Five-Second Entrance exercise via individual audio, instructing each participant through the sequence and reading a curated selection of index-card observations aloud in a neutral, non-identifying voice.
One at a time, each participant steps into the circle, makes eye contact with the group, and introduces themselves. The AI instructs the group to write two observations: "I noticed…" (behavioral) and "I felt / assumed…" (impact). The AI collects responses via voice input from each participant, synthesizes them, and reads three to five back to the person — framed as "data, not destiny."
"Write down your first impression of the person across from you. Don't share it. Just write it. Now set it aside — not because it's wrong, but because it's a hypothesis, not a verdict. The evening begins here."
A series of improv exercises that develop nonverbal attunement, active listening, spontaneity, and collaborative building. The AI delivers all instructions and timing cues. It monitors for synchrony via voice and instructs pairs to adjust pace when one partner is moving significantly faster than the other.
AI instructs pairs via individual earpieces. One leads; switch at 60 seconds; then "no leader" for 60 seconds. AI voice prompt: "Your job is to make your partner feel easy — not impressed. Slow is kind."
AI initiates the spine verbally and cues each turn. It monitors for one partner dominating and prompts the other if more than two consecutive turns pass without their input.
AI delivers the setup and timing. Content-free — trains reading affect and expressive range without words. AI observes response latency as a proxy for engagement.
The mask comes off. The AI pairs participants based on Love Map analysis and delivers the Closeness Ladder prompts at calibrated intervals. It monitors disclosure depth via response length and emotional vocabulary, and adjusts pacing — slowing when one partner is disclosing significantly more than the other, accelerating when both are moving through prompts quickly.
AI instructs the repetition sequence and monitors duration. If repetition becomes emotional, AI lowers its own voice and gives a single quiet prompt: "If it becomes emotional, stay simple. Don't explain. Just repeat and let it land."
AI delivers ten structured, escalating prompts — two minutes each — selected from its library based on the pair's Love Map compatibility profile. Sample prompts: "What's something you're proud of that your résumé wouldn't show?" · "What helps you feel calm when life is stressful?" · "What are you working on in yourself right now?" Any prompt can be swapped by saying "next" — the AI moves forward without comment.
Trust is built through demonstrated responsiveness. The AI coaches the AVA structure — Acknowledge, Validate, Ask — in real time, offering a brief prompt if it detects a response that skips directly to advice or rebuttal.
AI reads estimation questions, manages timing, and reveals answers. It monitors vocal tone during disagreement and prompts: "Watch for the instinct to blame. Replace it with collaborative warmth."
AI instructs one partner to share a mild stress. Coaches the responding partner through the three moves. After each exchange, the AI gives one calibration note — privately, via earpiece: "You moved to advice before you validated. Try again: what's the feeling underneath what they said?"
The body speaks before the mouth does. The AI delivers all instructions verbally and monitors voice patterns for engagement. It cannot see the room — it infers nonverbal engagement through vocal response patterns and response latency.
AI instructs one partner to make three small bids — verbal or nonverbal — and coaches the other to turn toward them. AI monitors whether the responding partner follows up with a question.
AI coaches the calibration of distance, posture, and tone toward "warm + safe — no touch." After 60 seconds, AI prompts both partners to name what helped — voice input, processed privately.
The shidduch crisis is partly a commitment language crisis. People who cannot say what they feel will not get engaged. The AI normalizes the difficulty of direct expression before the exercise begins — and coaches directness without pushing past a participant's readiness.
AI coaches: Preference — "I enjoyed ___ about being with you." Intention — "I would like to ___ next." Request — "Would you be open to ___?" After each sentence, the AI holds silence for five seconds before moving forward — training tolerance of the silence after a clear statement.
AI coaches the participant through five repetitions of a short commitment phrase with small variations. It notes privately: "By round three or four, the nervous laughter usually drops and the tone clears. That's the skill arriving." The AI monitors tone across the five rounds and reflects the change back to the participant after the exercise.
A sustainable marriage requires both safety and vitality. The AI runs two three-minute conversations and a debrief, coaching both partners to identify what "secure excitement" actually feels like in a single encounter.
Edge (novelty): AI prompt — "What's something you'd want to try that scares you a little — in a healthy way?" Home (security): AI prompt — "What does reliability look like day to day for you?" Then: "What did you feel in each mode? Can you hold both at the same time, with this person, right now?"
Participants check in via the Yismach app. AI confirms their Love Map is current and completes a brief 3-minute voice check-in with each person privately. Safety norms spoken by AI over the room speaker. Ambient sound activated.
Separate rooms. AI delivers all exercise instructions via room speaker (group instructions) and individual earpieces (private coaching). The mask starts to slip.
AI announces pairings privately via earpiece. No one learns their pairing publicly. Stations set by participants per AI instructions. No-touch norm restated.
AI-curated pairs. All exercise instructions via earpiece — private, paired, simultaneous. The AI runs up to eight pairs simultaneously with no degradation in instruction quality.
AI contacts each participant privately via earpiece: "Continue with your current partner, switch to a new pairing, or open the evening for informal conversation — say one, two, or three." Re-pairing is executed silently. No one is publicly labeled. No one is stranded.
For pairs with mutual interest, the final exercises deepen. For others, pairings adjust. AI monitors all pairs simultaneously.
AI goes silent. Low-pressure human conversation. The AI monitors ambient sound for distress signals but does not intervene unless triggered.
AI collects consent-based yes/no via private app input. Feedback gathered. AI schedules follow-up contact with each participant within 48 hours. Matches with mutual interest are notified privately. The AI follows up — through the first date, through the fifth date, through engagement. Because that is what a shadchan does.
"The AI paired us without our knowing. By the time the energy told us what the algorithm already knew, we were too busy actually connecting to care how it happened."
Learning to fight for the relationship, not against each other.
3 hours · AI-facilitated · dyadic
Destructive conflict patterns are among the strongest longitudinal predictors of divorce — not the presence of conflict, but the way it is handled. Demand–withdraw cycles, contempt, defensiveness, stonewalling, and the failure to repair after rupture are the behavioral signatures of relationships heading toward dissolution. The good news: conflict competence is trainable. This module trains it.
The AI facilitator delivers all six exercises via the couple's shared device speaker, with private coaching to each partner via individual phone audio when needed. The AI is trained on Gottman Method and PREP frameworks and calibrates its language to the couple's cultural and religious context based on their Love Map profiles.
Before anything can be repaired, it has to be named. The AI prompts each partner to name one "solvable" tension and one "perpetual" tension — independently, via private audio — then facilitates the share using Speaker–Listener format.
"Write down what you keep circling around. One tension that feels solvable — it has a solution you could both live with. One that feels perpetual — built into the structure of who the two of you are. Don't solve anything yet. Just name it."
After both partners share, the AI reflects back what it heard from each and names the category: solvable vs. perpetual. It introduces the Gottman finding — approximately 69% of marital conflicts are perpetual — and frames the distinction as a relief, not a failure. It monitors for escalation during the share and invokes the de-escalation protocol if heart-rate proxy indicators (voice pitch elevation, rapid speech) exceed threshold.
When flooding occurs, no productive conversation is possible. The AI teaches and rehearses the time-out protocol and self-soothing toolkit. It can detect flooding in real time via voice analysis and invoke the protocol automatically when needed.
AI coaches each partner through the three-component script three times each: naming the state, specifying the duration, committing to return. The AI plays the role of the partner receiving the time-out — practicing calm acknowledgment without pursuit.
AI guides 4–7–8 breathing in real time (timing the counts aloud), five-senses grounding (prompting each sense in sequence), and pause phrase design (suggesting three options, asking the couple to choose one and speak it aloud).
The AI adds the pause phrase to each partner's Love Map and will reference it in future sessions if flooding is detected. If a participant reports a heart rate above 100 during any future session, the AI will prompt the toolkit before continuing.
The AI uses a virtual "token" system — it announces "speaker" and "listener" roles audibly and enforces the reflection-before-rebuttal rule in real time. If the listener begins to rebut before reflecting, the AI interrupts gently: "Hold that thought — reflect first. What did you hear them say?"
"The rule is simple: the listener's only job is to reflect until the speaker says they feel heard. Not agree. Not rebut. Reflect. I'll keep time and let you know when the token passes. If I hear a rebuttal before a reflection, I'll ask you to try again. Let's begin."
Three rounds: low-stakes topic, solvable tension, perpetual tension. The AI calibrates round three carefully — instructing the couple to practice hearing the perpetual tension without defending, not to resolve it.
The AI provides a starter menu of repair phrases, coaches personalization, and runs the rehearsal — reading a phrase once flat, then prompting the participant to say it with eye contact, then simulating a low-stakes disagreement and cueing the phrase mid-conflict. The AI can play an escalating-but-not-abusive partner voice when simulating the disagreement.
"I'm sorry. That came out wrong." · "Can we start over?" · "I hear you. I got defensive." · "This is getting too hot. Let me try again." · "I love you. This is hard. Both are true." — Choose the ones that sound like you. Then say them until they feel like yours.
IBCT-adapted softening exercise. The AI guides each partner to identify their most persistent complaint, then excavates the vulnerable fear beneath it. It coaches the listener through the AVA structure from the Play Date.
"The complaint is the surface. What's underneath it? 'You never plan anything' — what does that feel like in your body? What are you afraid that means? Start there. The complaint is not what needs to be heard. The fear is."
The AI designs the weekly conflict meeting with the couple — prompting them to choose a day, time, and location, confirming the structure (one appreciation, one concern, Speaker–Listener format), and scheduling the first three weeks directly into their calendar via the Yismach app. It sets reminder notifications for each meeting.
Speaker–Listener practice: Two conversations per week, ten minutes each. The AI sends a WhatsApp prompt at the agreed time, asks for a one-word check-in ("done / skipped / hard") after each, and adjusts booster content based on patterns.
Repair attempt log: One intentional repair attempt per week. Participants log it via voice memo to the Yismach app. The AI acknowledges every entry within 24 hours.
30-Minute Booster: AI-facilitated, two weeks after the session. Reviews the logs, adjusts the conflict meeting design if needed, and runs one more Speaker–Listener round on the perpetual tension.
Stress is not the enemy. Facing it alone is.
2 hours · AI-facilitated · dyadic + individual
Emotion regulation predicts marital satisfaction longitudinally. But regulation is not only an individual skill — relationships are regulatory systems. When one partner is stressed, it spills into the dyad. Couples who develop shared coping strategies fare significantly better. This module builds both layers: the individual floor and the dyadic system that stands on it.
The AI facilitates all five exercises. For exercises requiring simultaneous individual and partner work, the AI splits into two audio streams — one coaching each partner privately via their phone while the shared speaker handles group instructions and timing.
Each partner writes three observable stress signals privately, then writes what they observe in the other when stressed. The AI facilitates the share and reflects on the gap between self-report and partner-report — which is almost always large. It adds the stress signals to each partner's Love Map as reference data for future sessions.
"Not 'I feel anxious' — but 'I get quiet and clean the kitchen.' Not the feeling. The behavior. Your partner needs to see the signal before they can respond to it. Give them the dictionary."
The AI also introduces the 1–10 stress number as a shared daily shorthand, coaches three practice rounds with real current stress levels, and adds the number system to the couple's daily check-in protocol.
The AI guides 4–7–8 breathing in real time (counting aloud), five-senses grounding (prompting each sense with a two-second pause), and pause phrase design. Partners take turns guiding each other through grounding — the AI coaches the guiding partner's voice tone: "Slower. A little warmer. You're the calm voice right now."
Three moves: stress communication, supportive coping, common coping. The AI coaches each move in sequence, monitors for the most common failure (advice when comfort was requested), and corrects in real time.
"You moved to advice. They asked for comfort. Wrong delivery is worse than no delivery. Try again: just be present. 'That sounds really hard. I'm here.' Nothing else."
For common coping (shared stressors), the AI runs a structured ten-minute problem-solving conversation: prompting the couple to define the problem in one sentence, generate three approaches without evaluating, then choose one together. It monitors for hierarchy — one partner taking over — and redistributes the conversation if needed.
The AI runs two full-cycle roleplays and coaches a deliberate mismatch repair — instructing one partner to give the wrong type of support, then coaching the receiving partner through the correction script. The AI normalizes the correction: "In real life, you will misread your partner's signal. The skill is not perfection. The skill is the recovery."
The AI designs the daily ritual with the couple, schedules it, and sets reminders. It rehearses one live six-minute check-in in the session, offering one calibration note at the end. The note is delivered privately to each partner — not as a grade, as a direction.
Six-minute check-in: AI sends daily reminder at agreed time. After 14 days, AI reviews the coping move log and sends a brief personalized summary: which move was used most, which was most effective, and one recommendation for the next module.
Self-soothing practice: AI sends a one-minute 4–7–8 audio guide each morning for the first week. Participants confirm completion via a single tap in the app.
A marriage is not a feeling. It is a life you build on purpose.
2 hours · AI-facilitated · dyadic
Attraction fades. Chemistry fluctuates. Even the deepest emotional connection will, at some point, run into a question it cannot answer with feeling alone: what kind of life are we trying to build? Gottman's Sound Relationship House places shared meaning at the very top. This module makes the invisible architecture visible.
Eight domains: Torah/Spiritual Life, Work & Career, Money, Family & In-laws, Community, Children, Lifestyle, Personal Growth. Each partner independently rates importance (1–10) and records a one-sentence vision — via voice input to the Yismach app, transcribed in real time. The AI generates a visual overlap map and presents it to the couple, naming alignment zones, difference zones, and tension zones.
"Alignment is the foundation. Difference is normal and manageable. Tension is where the work lives — not where the relationship ends. Every tension zone requires an explicit conversation. Let's have them now."
The AI stores the matrix in each partner's Love Map and cross-references it in all future modules where the relevant domain arises. If a financial conflict arises in Module 4, the AI will reference the money domain alignment/tension score from this session.
Each partner completes three prompts via private voice input: "I grew up imagining the husband's role as…" · "I grew up imagining the wife's role as…" · "The role I actually want for myself is…" The AI facilitates the share using Speaker–Listener format, holding the reflection requirement firmly. It flags — privately to the listener — when a response sounds like inherited script rather than stated preference.
The AI guides the couple to design three weekly rituals: a Shabbos ritual, a weekday check-in ritual, and an appreciation ritual. For each, the AI asks: name it, set the time and day, describe the format, commit to a four-week trial. It adds all three to the Yismach app's ritual tracker and sets weekly completion prompts.
"A ritual you don't schedule is a ritual that doesn't happen. I'm going to ask you to put these in your calendar right now, while we're talking. I'll wait."
The AI presents four decision-making frameworks — domain authority, consensus with time limit, influence acceptance, veto with explanation — and coaches the couple to choose one or design a hybrid. The chosen rule is dictated by the couple to the AI, transcribed, and added to their shared profile. The AI references it when decision-conflict patterns are detected in future sessions.
Values workbook: AI sends three prompts per week expanding each Life Domain into Year 1, Year 5, and Year 15 visions. Responses submitted via voice memo. AI synthesizes and shares a summary with both partners.
Ritual trial: App tracks completion of the three designed rituals for four weeks. After four weeks, AI sends a brief review prompt: which rituals stuck, which need redesign, which were skipped and why.
Money is never just about money.
2.5 hours · AI-facilitated · engaged couples and serious dyads
Financial disagreements are a uniquely strong predictor of divorce — stronger than disagreements about housework, in-laws, sex, or leisure time. Not because of the money itself. Money is a proxy for power, security, trust, and control. This module turns the values conversation from Module 3 into a concrete financial system.
The AI prompts each partner privately through four questions — voice input, transcribed — then facilitates the share using Speaker–Listener. It listens for emotional vocabulary during the share and reflects it back: "I heard fear in how you described scarcity. That's worth staying with for a moment before we move forward."
"In my home growing up, money felt like…" · "The unspoken rule about money in my family was…" · "My biggest money fear is…" · "What I need from a financial partnership is…"
The AI facilitates four disclosures: debts and obligations, income and assets, giving commitments, and risk tolerance. It establishes the emotional frame before any numbers appear: "This is not an interrogation. It is a mutual act of trust. The purpose is information, not evaluation. No judgment is appropriate here — only curiosity." For dating dyads not yet engaged, the AI adjusts the protocol to ranges rather than exact figures.
The AI walks the couple through three passes: categories, allocation, and cadence. It provides a standard category list and prompts additions. For each category, it asks both partners to name an estimate — mediating when the estimates diverge significantly before moving to the next category. The completed first-draft budget is stored in the Yismach app and becomes the baseline for the monthly money meeting.
"Wrong numbers are better than no numbers. You can correct the estimate. You cannot correct the silence. Let's start with your best guess and adjust from there."
The AI coaches the couple through purchase threshold agreements, giving rules, and a financial-specific disagreement protocol. It monitors for one partner dominating — if one partner's voice input exceeds 70% of the conversation time, the AI prompts the other: "I want to hear from both of you equally here. What's your instinct on this?"
The AI designs the 30-minute monthly meeting with the couple: review, adjust, appreciate, flag. It schedules the first three meetings in the app, sets reminders, and offers to send a structured agenda to both partners before each meeting. The agenda is generated automatically from the previous month's spending data if the couple has connected their budget tracker to the Yismach system.
Joint budget draft: AI sends weekly prompts to complete the budget skeleton with real numbers. After four weeks, AI generates a summary: categories where estimates were accurate, categories where they drifted, and one structural recommendation.
First money meeting: AI sends the structured agenda, joins the meeting as a silent timer and structure-keeper if invited, and sends a brief reflection afterward.
The invisible labor that holds a home together — made visible.
1.5 hours · AI-facilitated · engaged couples and early marriages
The mental load — the planning, anticipating, remembering, and monitoring that keeps a household functioning — is invisible by definition. When it is distributed inequitably, the carrying partner experiences chronic depletion and resentment. This module makes it visible and distributes it equitably. Helping is not partnership. Ownership is.
Each partner independently voice-inputs every household task they can name, sorted into visible and invisible labor. The AI transcribes and combines the lists, then presents the gaps: "These tasks appear on your list but not your partner's. That gap is the invisible labor one of you is carrying alone." The shock is the point.
Working from the combined inventory, the AI facilitates ownership assignment for each task — confirming that "ownership means the full cycle: noticing, planning, executing, maintaining." It flags when a partner says "I'll help with that" — redirecting: "Helping leaves the mental load with the other person. Is this yours, or theirs?" It identifies two stretch tasks for each partner and adds them to the task tracker.
The AI presents the standard list of life transitions that require full redistribution: job change, pregnancy and postpartum, illness, children starting school, and custom triggers named by the couple. For each, the couple pre-commits verbally — and the AI logs the commitment. When a trigger event is reported in a future session, the AI initiates the redistribution conversation automatically.
The AI designs the 15-minute weekly meeting and schedules it. Three questions: What did I handle this week? What's coming up that needs coordination? Is anything feeling unbalanced? The AI can join the stand-up as a structured timekeeper if invited — keeping the three questions on track and flagging if the "unbalanced" question is skipped.
The conversation most couples never have — and every marriage needs.
3 hours · two sessions · AI-facilitated · engaged and early-married couples · culturally gated
Sexual communication is positively associated with both relationship satisfaction and sexual satisfaction in meta-analytic research. The quality of how couples talk about intimacy matters more than the frequency of intimacy itself. This module does not teach technique. It teaches communication — the language to express needs, hear differences, navigate them with dignity, and build rituals of affection that sustain the physical dimension of the relationship across decades.
Session A (separate gender groups) is delivered by a gender-matched AI voice — the system uses a female voice for the women's group, male voice for the men's. Voice matching is the AI's approximation of the gender-matched facilitator. For participants who flag significant trauma history during intake, this module is paused and a referral to a human therapist is offered before proceeding. All exercise content in Session B is exclusively dyadic and private — the AI audio stream is available to the couple only, never to a group.
The AI is trained to recognize disclosure that exceeds the session's scope — trauma history, significant anxiety, distress beyond the exercise parameters — and to respond with warmth, a pause, and a referral pathway rather than continuing the exercise. No exercise in this module continues past a distress threshold.
Three private prompts via individual audio: what was taught, what is feared, what is hoped. The AI invites voluntary voice sharing and responds to each with warmth and normalization — never with advice or evaluation. It names the most common fears aloud before asking for input: "Nearly everyone in this group has a version of this fear. You are not the only one."
The AI presents the learning curve framework and names common anxieties explicitly — reading them so no participant has to be the first to say them. It introduces the concept of desire discrepancy as normal and asks the group to discuss (voluntarily): what would help you feel safe enough to tell your partner what you need? The AI does not record or log voluntary group discussions. Individual responses to structured prompts are logged to Love Map profiles.
The AI coaches the three-category response system — Yes, Not Now, Let's Talk — via private couple audio. It coaches three practice rounds per partner and reflects on tone: "'Not now' carries less weight when it sounds like 'no, never.' Try again with a signal of continued care at the end."
Private prompts via individual audio, then Speaker–Listener share facilitated by the AI. The AI holds the reflection requirement firmly and intervenes if a partner moves to reassurance before reflecting: "Don't reassure yet. Just reflect what you heard. What did they say they feel when this happens?"
The AI guides the design of contact-based and non-contact affection rituals, with explicit attention to the niddah cycle — designing non-contact rituals that maintain emotional closeness during the periods when physical closeness is paused. All designed rituals are added to the app's ritual tracker alongside the Module 3 rituals.
Your marriage is not a branch office of your parents' home.
1.5 hours · AI-facilitated · engaged and early-married couples
Discordance in perceptions of in-law relationships predicts divorce risk in longitudinal research. Not the quality of the in-law relationships — the discordance. This module maps the loyalty binds, forms a united alliance, rehearses boundary language, and prepares the couple for the inevitable moment when a boundary is crossed. A boundary drawn with love protects everyone on both sides of it.
Each partner writes privately then shares via Speaker–Listener: where they feel pulled, what their default pattern is, and what they need from their partner in a loyalty bind. The AI reflects back what it hears from each partner and names the pattern — avoidance, compliance, explosion — without judgment: "That's one of the most common patterns. Naming it is the first step toward choosing differently."
The AI coaches the three-step protocol — private discussion, joint decision, own-parent delivery — through three scenarios of increasing stakes. It plays the role of a persistent parent when simulating the delivery conversation and monitors for the most common failures: throwing the partner under the bus, caving under imagined parental pressure, and decisions that were not truly joint.
"We decide together, then inform. The word 'we' is load-bearing. Not 'she wants' or 'he said.' We. Practice until 'we' is the first word out of your mouth."
The AI presents the boundary phrase menu, coaches customization, and runs three rehearsals for each partner: reading from the script, from memory, and in a simulated conversation where the AI plays a persistent parent. It monitors for the JADE pattern (justify, argue, defend, explain) and interrupts: "You're explaining. The boundary doesn't need justification. Just: 'I hear you. We've made our decision.' Say it again."
The AI walks through the four-step repair protocol and pre-agrees the signal — a word or phrase the couple will use privately when a boundary is crossed. The signal is stored in both partners' Love Maps and the AI will reference it if boundary-crossing patterns are reported in future sessions.
Family Policy one-pager: AI generates a draft based on the session's decisions and sends it to both partners for review and editing. This becomes the couple's internal operating agreement for family-of-origin relationships.
One boundary conversation: After the conversation, participants log a voice memo to the app. The AI responds with one reflection within 24 hours.
The baby does not destroy your marriage. The unpreparedness does.
3 hours · AI-facilitated · expecting couples · postpartum check-in at 8–12 weeks
Marital satisfaction declines on average across the transition to parenthood — one of the most replicated findings in relationship science. Not because babies are bad for marriages. Because most couples face this transition with no preparation beyond the childbirth class. This module is the preparation. It does not teach parenting. It teaches couples how to stay a couple while becoming parents.
The AI presents the four domains of change — sleep, workload, spontaneity, identity — with research-grounded honesty. It then invites each partner to share one fear about the transition via Speaker–Listener. The AI names every common fear aloud before asking for individual input, normalizing each without dismissing it.
"Couples who expected the difficulty and planned for it report smaller drops in satisfaction than couples who expected it to be natural and easy. I'm going to tell you exactly what's coming — not to frighten you, but to inoculate you. Surprise is the enemy. Information is the preparation."
The AI presents supportive vs. undermining coparenting behaviors and coaches each partner to commit to three specific supportive behaviors and identify one undermining behavior they are most at risk for. It then facilitates the parenting disagreement protocol — in-the-moment support, debrief format, and ongoing perpetual-difference navigation — linking each step to the skills built in earlier modules.
"You will disagree about parenting. Every couple does. The question is whether your disagreements happen in private, between allies — or in public, in front of the child and the grandparents. The alliance is: we disagree in private and present a united front."
The AI reactivates the Module 5 task inventory and guides a full redistribution for the first 90 days. It is explicit about the biological constraints (nursing), coaches the night shift protocol, and establishes a weekly labor review cadence for the first three months. It stores the plan in the app and sets review reminders at weeks 2, 4, 6, and 8.
"The plan you make today will be wrong within two weeks. That is not a failure. That is parenthood. The plan is not the point. The habit of replanning together is the point."
The AI designs the ten-minutes-a-day ritual: check-in (3 min), appreciation (2 min), touch (5 min). It schedules it, sets daily reminders, and coaches the couple on identifying their support network for the weekly couple time. It is direct about the stakes: "The marriage that waits until it is easy to make time for each other will wait forever."
The AI schedules this before the couple leaves the session. At 8–12 weeks, it conducts a 30-minute check-in: labor review, coparenting pulse, and couple connection check. It also runs a brief postpartum mood screening using validated language, normalizing the screening as routine — "We check this with everyone, the same way you check blood pressure" — and provides a warm referral pathway to a human mental health professional if indicators are present. If the screening flags significant concern, the AI does not continue to module content. It pauses, acknowledges, and facilitates the referral.
Postpartum operating agreement: AI generates a draft document covering the first-90-days labor plan, coparenting alliance commitments, couple protection ritual, and in-law boundaries for the postpartum period. Sent to both partners for review within 24 hours of the session.
The strongest marriages are not the ones that never need help. They are the ones that ask for it before they are drowning.
1 hour · AI-facilitated · capstone · all couples who completed the curriculum
Relationship education shows average positive effects on communication — but the effects fade without maintenance. The curriculum you have completed is not a vaccination. It is a training regimen. The skills will atrophy if they are not practiced, the systems will drift if they are not maintained. This module converts everything built across nine modules into a sustainable, lifelong maintenance plan.
The AI presents the three-zone framework and coaches the couple through a mutual pre-commitment. It addresses stigma directly and in culturally specific language — naming the frum community's particular barriers to professional help-seeking and reframing help-seeking as consistent with Torah values: consulting a rav, visiting a doctor, hiring a tutor are all forms of seeking expertise when it is needed.
"Green: you're maintaining. Keep going. Yellow: something has drifted and you can feel it. Enrichment within two weeks. Red: contempt has arrived, or one of you has considered leaving, or there is deception. A therapist within one week. I'm going to ask both of you to commit to those timelines out loud, right now."
The AI monitors for zone-relevant signals across all future interactions — session check-ins, app logs, homework completion rates, distress flags. When patterns suggesting yellow-zone drift emerge, it surfaces them directly: "I've noticed a pattern I want to share with you." It does not wait to be asked.
The AI builds the referral map with the couple in real time — prompting them to name a rav or rebbetzin they trust for relational guidance, and presenting a curated list of Yismach-vetted couples' coaches and licensed therapists who are frum-competent and trained in evidence-based approaches. The couple chooses one name in each category and the AI stores the contact information in both their profiles. It also stores the agreed help-seeking thresholds — and will reference them if distress signals appear in the future.
The AI schedules the first annual checkup before the session ends — asking the couple to choose a fixed annual date and adding it to both calendars. The checkup structure: a three-minute CSI satisfaction pulse (administered by the AI), a systems review (checking which maintenance rituals from all nine modules are still active), a zone assessment, and goals for the year. The AI sends a structured checkup guide two weeks before the scheduled date every year — indefinitely.
"The AI doesn't get tired of checking in. It doesn't forget. It doesn't assume everything is fine because you haven't complained. Every year, on the same date, it asks: how are you? And it means it."
Interpersonal fluency. The foundation of being seen and seeing. One evening. AI-facilitated entirely.
Fight well. Repair fast. Protect the bond even when you disagree. 3 hours + AI-monitored booster.
Carry stress together. Self-soothe. Match support to the request. 2 hours + 14-day AI-monitored practice.
Dream together. Align values. Design rituals. Make decisions on purpose. 2 hours + AI-generated workbook.
Build a shared financial system with honesty, transparency, and fair process. 2.5 hours + AI-maintained budget baseline.
See the invisible labor. Own tasks, not just help. Renegotiate when life changes. 1.5 hours + AI task tracker.
Talk about closeness. Navigate difference. Build affection rituals. 3 hours (split) + AI-monitored check-in. Culturally gated.
Form an alliance. Set boundaries with love. Repair when lines are crossed. 1.5 hours + AI-generated family policy.
Reset expectations. Build a coparenting alliance. Protect the couple through the most demanding transition. 3 hours + AI-facilitated postpartum check-in.
Know when to ask for help. Build a referral map. Install an annual AI-facilitated checkup. Maintain for life. 1 hour.
Total investment: approximately 21 hours of structured learning, plus daily and weekly rituals that take minutes. Available anywhere in the world, at any hour, with no human facilitator required. The same AI that met you at intake will still be checking in on your marriage a decade from now — on the same date, every year, asking the same question: how are you?