convoke-agents 2.0.0

package/_bmad/bme/_vortex/workflows/learning-card/learning-card.template.md

@@ -0,0 +1,179 @@
+---
+title: "Learning Card: {experiment-name}"
+date: {date}
+created-by: {user-name} with Max (learning-decision-expert)
+experiment-name: {experiment-name}
+hypothesis: {hypothesis-summary}
+status: DRAFT
+confidence: {HIGH|MEDIUM|LOW|EXPLORATORY}
+outcome: {validated|invalidated|partially-validated|inconclusive}
+tags: [{tags}]
+version: 1.0
+---
+
+# Learning Card: {experiment-name}
+
+> **Confidence:** {confidence-rating} | **Outcome:** {outcome} | **Date:** {date}
+
+## One-Sentence Summary
+
+{one-sentence-summary}
+
+---
+
+## Experiment Context
+
+### Hypothesis
+{hypothesis-statement}
+
+### Method
+- **Type:** {experiment-type}
+- **Sample:** {sample-size-and-description}
+- **Duration:** {experiment-duration}
+- **Recruitment:** {how-participants-selected}
+
+### Pre-Defined Success Criteria
+
+| Metric | Target | Rationale |
+|--------|--------|-----------|
+| {metric-1} | {target-1} | {why-this-threshold} |
+| {metric-2} | {target-2} | {why-this-threshold} |
+| {metric-3} | {target-3} | {why-this-threshold} |
+
+### Strategic Context
+{why-this-experiment-mattered}
+
+---
+
+## Raw Results
+
+### Quantitative
+
+| Metric | Target | Actual | Met? |
+|--------|--------|--------|------|
+| {metric-1} | {target-1} | {actual-1} | {yes/no/partially} |
+| {metric-2} | {target-2} | {actual-2} | {yes/no/partially} |
+| {metric-3} | {target-3} | {actual-3} | {yes/no/partially} |
+
+### Qualitative Highlights
+
+**Key Quotes:**
+{direct-quotes-with-context}
+
+**Observed Behaviors:**
+{notable-behaviors}
+
+### Unexpected Findings
+{unexpected-findings}
+
+### Data Quality Assessment
+- **Sample representativeness:** {assessment}
+- **Duration adequacy:** {assessment}
+- **Confounding factors:** {factors}
+- **Missing data:** {gaps}
+
+---
+
+## Analysis
+
+### Success Criteria Assessment
+
+| Criterion | Result | Rating | Reasoning |
+|-----------|--------|--------|-----------|
+| {criterion-1} | {result-1} | {CLEAR PASS/MARGINAL PASS/MARGINAL FAIL/CLEAR FAIL/INCONCLUSIVE} | {why} |
+| {criterion-2} | {result-2} | {rating} | {why} |
+
+### Key Patterns
+{pattern-analysis}
+
+### Qualitative-Quantitative Alignment
+{alignment-analysis}
+
+### Alternative Explanations Considered
+{alternative-explanations}
+
+---
+
+## Validated Learnings
+
+### Assumption Status
+
+| Assumption | Status | Confidence | Key Evidence |
+|-----------|--------|------------|--------------|
+| {assumption-1} | {VALIDATED/INVALIDATED/PARTIALLY/UNRESOLVED} | {HIGH/MEDIUM/LOW} | {evidence-summary} |
+| {assumption-2} | {status} | {confidence} | {evidence} |
+
+### Core Learning Statements
+
+1. **Learning:** {learning-statement-1}
+   **Evidence:** {supporting-evidence}
+   **Conditions:** {when-this-holds}
+
+2. **Learning:** {learning-statement-2}
+   **Evidence:** {supporting-evidence}
+   **Conditions:** {when-this-holds}
+
+3. **Learning:** {learning-statement-3}
+   **Evidence:** {supporting-evidence}
+   **Conditions:** {when-this-holds}
+
+### New Questions Raised
+
+| Question | Why It Matters | How to Test | Priority |
+|----------|---------------|-------------|----------|
+| {question-1} | {importance} | {experiment-idea} | {HIGH/MEDIUM/LOW} |
+| {question-2} | {importance} | {experiment-idea} | {priority} |
+
+### Evidence Quality Rating
+**Overall:** {HIGH/MODERATE/WEAK/ANECDOTAL}
+**Justification:** {why-this-rating}
+
+---
+
+## Implications
+
+### Product Implications
+{product-implications-with-build-dont-build-modify-investigate}
+
+### Strategy Implications
+{strategy-implications}
+
+### Assumption Cascade
+**Now stronger:** {reinforced-assumptions}
+**Now weaker:** {undermined-assumptions}
+**New to test:** {new-assumptions}
+
+### Decision Trigger
+{does-this-trigger-a-major-decision}
+
+---
+
+## Recommended Next Actions
+
+| Timeframe | Action | Owner | Rationale |
+|-----------|--------|-------|-----------|
+| Immediately | {action} | {who} | {why} |
+| This sprint | {action} | {who} | {why} |
+| This quarter | {action} | {who} | {why} |
+
+---
+
+## Connections
+
+- **Builds on:** {related-learning-cards}
+- **Contradicts:** {conflicting-findings}
+- **Enables:** {future-experiments}
+
+---
+
+## Revision History
+
+| Date | What Changed | Why | Updated By |
+|------|--------------|-----|------------|
+| {date} | Initial creation | First synthesis | {user-name} |
+
+---
+
+**Created with:** Convoke v2.0.0 - Vortex Pattern (Systematize Stream)
+**Agent:** Max (Learning & Decision Expert)
+**Workflow:** learning-card

package/_bmad/bme/_vortex/workflows/learning-card/steps/step-01-experiment-context.md

@@ -0,0 +1,100 @@
+---
+step: 1
+workflow: learning-card
+title: Experiment Context
+---
+
+# Step 1: Experiment Context
+
+Before we analyze results, we need to clearly establish what was tested and why. This context ensures the learning card is useful months from now when you've forgotten the details.
+
+## Why This Matters
+
+A learning without context is just a random fact. Context establishes:
+- The original hypothesis and its importance to your strategy
+- The method chosen and why it was appropriate
+- The timeline and conditions under which the experiment ran
+- The success criteria that were defined BEFORE seeing results (this prevents moving goalposts)
+
+## Your Task
+
+Answer these questions to establish the experiment context:
+
+### 1. What experiment did you run?
+
+Give it a clear name and a one-sentence description.
+- **Name:** (e.g., "Landing Page Smoke Test", "Concierge MVP Week 3", "Price Sensitivity Survey")
+- **Description:** What did you do, in plain language?
+
+### 2. What hypothesis were you testing?
+
+Use the format: "We believe that [target users] will [expected behavior] because [rationale]."
+
+Be specific:
+- **Weak:** "We believe users will like our product"
+- **Strong:** "We believe that remote engineering managers with 5+ direct reports will complete weekly status reviews 40% faster using async video updates because they currently spend 3+ hours/week in synchronous standup meetings"
+
+### 3. What method did you use?
+
+Describe the experiment type and setup:
+- **Type:** (Interview, Survey, A/B test, Landing page, Concierge MVP, Wizard of Oz, Prototype test, etc.)
+- **Sample size:** How many participants/users/responses?
+- **Duration:** How long did the experiment run?
+- **Recruitment:** How were participants selected? (This matters for bias assessment)
+
+### 4. What were the pre-defined success criteria?
+
+What thresholds did you set BEFORE running the experiment?
+
+Examples:
+- "At least 30% of visitors click the CTA"
+- "7 out of 10 interview participants mention this pain point unprompted"
+- "Completion time under 5 minutes for 80% of participants"
+- "Net Promoter Score above 40"
+
+### 5. What was the strategic context?
+
+Why did this experiment matter? What decision was it supposed to inform?
+- What stream of the Vortex was this part of?
+- What assumption in your business model does this test?
+- What would you do differently depending on the outcome?
+
+## Example
+
+**Experiment Name:** Async Status Update Smoke Test
+
+**Description:** We created a simple Loom-based workflow where team members recorded 2-minute daily status videos instead of attending standup meetings. Ran with 3 pilot teams for 2 weeks.
+
+**Hypothesis:** We believe that remote engineering teams (5-8 people) will maintain alignment without daily standup meetings if they record 2-minute async video updates, because team members report standups are disruptive and poorly timed across time zones.
+
+**Method:**
+- **Type:** Concierge MVP (manual process using Loom + Slack channel)
+- **Sample size:** 3 teams, 19 total participants
+- **Duration:** 2 weeks (10 business days)
+- **Recruitment:** Volunteered from teams that had complained about standup meetings in last retrospective
+
+**Success Criteria (pre-defined):**
+- 80%+ daily participation rate (at least 8 of 10 days per person)
+- Team alignment score maintained or improved (weekly survey, 1-5 scale)
+- Average video length under 3 minutes
+- Managers report spending less time on status tracking (time log comparison)
+
+**Strategic Context:**
+- Part of Externalize stream: testing our value proposition before building
+- Tests core assumption: "Teams will actually do async updates consistently"
+- If validated: proceed to build MVP with recording + dashboard features
+- If invalidated: explore hybrid model (fewer meetings, not zero meetings)
+
+---
+
+## Your Turn
+
+Please provide the experiment context using the structure above.
+
+**Important:** If you didn't define success criteria before the experiment, acknowledge that. Post-hoc criteria are a red flag for confirmation bias -- we'll note that in the learning card so future readers know the evidence quality.
+
+## Next Step
+
+When you've provided complete experiment context, I'll load:
+
+{project-root}/_bmad/bme/_vortex/workflows/learning-card/steps/step-02-raw-results.md

package/_bmad/bme/_vortex/workflows/learning-card/steps/step-02-raw-results.md

@@ -0,0 +1,125 @@
+---
+step: 2
+workflow: learning-card
+title: Raw Results
+---
+
+# Step 2: Raw Results
+
+Now we capture the raw data and observations -- exactly what happened, without interpretation. This step is deliberately separated from analysis to prevent confirmation bias.
+
+## Why This Matters
+
+The most common mistake in experiment analysis is jumping straight to conclusions. Separating raw data from interpretation:
+- **Prevents cherry-picking** - You record ALL results, not just ones that support your hypothesis
+- **Enables re-analysis** - Others can draw different conclusions from the same data
+- **Catches surprises** - The most valuable learnings are often the unexpected ones
+- **Builds credibility** - Stakeholders trust conclusions grounded in visible data
+
+## Your Task
+
+Capture the raw results without analysis or interpretation. Just the facts.
+
+### 1. Quantitative Results
+
+Report the numbers against each success criterion:
+
+| Metric | Target | Actual | Met? |
+|--------|--------|--------|------|
+| {metric-1} | {target-1} | {actual-1} | Yes/No |
+| {metric-2} | {target-2} | {actual-2} | Yes/No |
+
+Include ALL metrics you tracked, even ones that seem irrelevant now.
+
+### 2. Qualitative Observations
+
+Capture direct quotes, observed behaviors, and notable moments:
+
+**Direct Quotes (verbatim where possible):**
+- "..." - Participant 1, context
+- "..." - Participant 2, context
+
+**Observed Behaviors:**
+- What did participants actually do (vs. what they said they'd do)?
+- Where did they get stuck, confused, or frustrated?
+- What surprised you?
+
+**Notable Moments:**
+- Any "aha" moments?
+- Any emotional reactions?
+- Any participants who used the product/solution in unexpected ways?
+
+### 3. Unexpected Findings
+
+What happened that you DIDN'T expect? These are often the most valuable data points.
+
+- Anything participants mentioned that wasn't in your hypothesis?
+- Any patterns you didn't anticipate?
+- Any metrics that moved in surprising directions?
+- Any segments that behaved differently from others?
+
+### 4. Data Quality Assessment
+
+Be honest about the limitations:
+
+- **Sample bias:** Were participants representative of your target user?
+- **Duration:** Was the experiment long enough to see real behavior?
+- **Confounding factors:** What else was happening that might have affected results?
+- **Missing data:** What couldn't you measure that you wish you had?
+- **Hawthorne effect:** Did participants behave differently because they knew they were being observed?
+
+## Example
+
+**Quantitative Results:**
+
+| Metric | Target | Actual | Met? |
+|--------|--------|--------|------|
+| Daily participation rate | 80%+ | 73% overall (Team A: 91%, Team B: 68%, Team C: 62%) | No |
+| Alignment score (1-5) | Maintain at 3.8+ | Week 1: 3.2, Week 2: 3.9 | Partially (Week 2 only) |
+| Average video length | Under 3 min | 2 min 14 sec | Yes |
+| Manager time on status | Reduce by 30%+ | Manager A: -45%, Manager B: -20%, Manager C: +10% | Mixed |
+
+**Qualitative Observations:**
+
+Direct quotes:
+- "I actually liked knowing I could watch updates on MY schedule, not theirs" - Manager A
+- "By day 4 I forgot to record. It's just not in my routine yet." - Dev, Team C
+- "The videos were better than standups because people showed their screens instead of just talking" - Tech Lead, Team A
+- "I don't want to be on camera. Can I just do audio?" - Dev, Team B
+
+Observed behaviors:
+- Team A adopted a pattern of recording first thing in morning (habit-stacking with coffee)
+- Team B participation dropped steadily over 2 weeks
+- Team C had one champion who recorded every day; others were inconsistent
+- 4 participants started watching videos at 2x speed by day 3
+
+Notable moments:
+- Manager A discovered a blocker through a video that had existed for 3 days unreported
+- Team B tried to game it by recording 15-second "nothing to report" videos
+
+**Unexpected Findings:**
+- Screen-sharing in videos was MORE informative than verbal standup updates
+- Async worked better for cross-timezone teams (Team A) than co-located teams (Team C)
+- Junior developers participated more consistently than senior developers
+- Two participants started using videos for non-status purposes (design reviews, bug reports)
+
+**Data Quality:**
+- Sample bias: All volunteer teams who already disliked standups (positive selection bias)
+- Duration: 2 weeks may not show long-term habit formation
+- Confounding: Team A was starting a new sprint (high motivation), Team C was in maintenance mode (low energy)
+- Missing: No customer impact data, no quality metrics
+- Hawthorne: Likely -- teams knew managers were tracking participation
+
+---
+
+## Your Turn
+
+Please share the raw results of your experiment. Resist the urge to interpret -- just give me the data. We'll make sense of it in the next step.
+
+**Reminder:** Include results that contradict your hypothesis too. Those are the most valuable learning opportunities.
+
+## Next Step
+
+When you've captured the raw results, I'll load:
+
+{project-root}/_bmad/bme/_vortex/workflows/learning-card/steps/step-03-analysis.md

package/_bmad/bme/_vortex/workflows/learning-card/steps/step-03-analysis.md

@@ -0,0 +1,125 @@
+---
+step: 3
+workflow: learning-card
+title: Analysis
+---
+
+# Step 3: Analysis
+
+Now we analyze the raw results. What does the data actually say? This is where we look for patterns, assess significance, and separate signal from noise.
+
+## Why This Matters
+
+Raw data doesn't speak for itself. Analysis transforms data into insight by:
+- Distinguishing between statistically meaningful results and random variation
+- Identifying patterns across different segments and metrics
+- Spotting contradictions between quantitative and qualitative data
+- Assessing whether results are strong enough to act on
+
+## Your Task
+
+Work through these analysis lenses systematically:
+
+### 1. Success Criteria Assessment
+
+For each pre-defined success criterion, assess the result:
+
+**Use this framework:**
+- **CLEAR PASS:** Met or exceeded the threshold with margin
+- **MARGINAL PASS:** Met the threshold but barely -- could go either way with more data
+- **MARGINAL FAIL:** Close to threshold but didn't reach it -- worth investigating why
+- **CLEAR FAIL:** Clearly did not meet the threshold
+- **INCONCLUSIVE:** Not enough data or too many confounding factors to judge
+
+For each: explain WHY you're assigning that rating.
+
+### 2. Pattern Analysis
+
+Look for patterns in the data:
+
+**Segment Differences:**
+- Did different user segments behave differently?
+- Was there a "best case" group vs. a "worst case" group?
+- What distinguishes users who succeeded from those who didn't?
+
+**Time-based Patterns:**
+- Did behavior change over the experiment duration?
+- Was there a learning curve? A novelty effect? A drop-off?
+- At what point did behavior stabilize (or didn't it)?
+
+**Correlation Patterns:**
+- Do any metrics correlate with each other?
+- Does high engagement in one area predict success in another?
+- Are there leading indicators of success or failure?
+
+### 3. Qualitative-Quantitative Alignment
+
+Compare what people SAID with what they DID:
+
+- Do the numbers match the quotes?
+- Are there contradictions? (e.g., "I loved it" but usage dropped)
+- Which should you trust more in this context -- the quantitative or qualitative data?
+
+### 4. Statistical Rigor Check
+
+Be honest about confidence levels:
+
+- **Sample size:** Was it large enough to draw conclusions? (Rule of thumb: for surveys, 30+ responses; for A/B tests, use a significance calculator; for interviews, 5-8 per segment for qualitative patterns)
+- **Confidence level:** If quantitative, what's the statistical significance? (p-value, confidence interval)
+- **Effect size:** Even if statistically significant, is the effect large enough to matter practically?
+- **Replication:** Would you expect the same results if you ran this again?
+
+### 5. Alternative Explanations
+
+For each finding, consider: is there another explanation?
+
+**Common alternative explanations:**
+- **Selection bias:** Did you only test with people predisposed to succeed?
+- **Novelty effect:** Did they engage because it was new, not because it was better?
+- **Hawthorne effect:** Did observation itself change behavior?
+- **Regression to the mean:** Were your baseline measurements unusually high/low?
+- **Confounding variables:** Was something else changing at the same time?
+
+## Example
+
+**Success Criteria Assessment:**
+
+| Criterion | Result | Rating | Reasoning |
+|-----------|--------|--------|-----------|
+| 80%+ daily participation | 73% overall | MARGINAL FAIL | 7 points short, but Team A hit 91%. Suggests it's achievable with right conditions, not universally. |
+| Alignment score 3.8+ | 3.2 -> 3.9 | MARGINAL PASS | Only met in Week 2. Learning curve effect likely. Need longer experiment to know if this holds. |
+| Video under 3 min | 2:14 avg | CLEAR PASS | Well under threshold with no outliers above 4 min. |
+| Manager time -30% | -45%, -20%, +10% | INCONCLUSIVE | Too small a sample (n=3) and too varied. Manager C's increase may be due to monitoring the new process. |
+
+**Pattern Analysis:**
+- Clear segment difference: Cross-timezone teams benefited most (Team A: +91% participation, highest alignment improvement). Co-located teams saw least benefit (Team C: 62% participation).
+- Time pattern: Week 1 participation was 80%, Week 2 dropped to 67%. Novelty effect likely.
+- Correlation: Teams with a "recording champion" (someone who recorded every day) had 25% higher overall participation.
+
+**Qualitative-Quantitative Alignment:**
+- CONTRADICTION: Team B members said they "liked" async updates in the mid-point survey, but their participation dropped from 78% (Week 1) to 58% (Week 2). Words didn't match behavior.
+- ALIGNMENT: Manager A's quote about discovering blockers matched the quantitative finding that screen-sharing revealed issues faster.
+
+**Statistical Rigor:**
+- Sample: n=19 across 3 teams is too small for quantitative conclusions. Treat all findings as directional signals, not proof.
+- Confidence: Cannot calculate meaningful p-values with this sample size.
+- Effect size: Alignment score improvement (0.7 points on 5-point scale) is practically meaningful IF it holds.
+
+**Alternative Explanations:**
+- Volunteer bias is the biggest threat: teams who already hated standups were predisposed to prefer any alternative
+- Novelty effect likely contributed to Week 1's higher participation
+- Manager monitoring likely boosted participation (Hawthorne effect)
+
+---
+
+## Your Turn
+
+Please analyze your raw results using the lenses above. Be rigorous and honest -- a learning card that says "inconclusive" is more valuable than one that pretends certainty.
+
+**Key principle:** It's better to learn that you can't conclude anything than to draw false conclusions. Uncertainty is data too.
+
+## Next Step
+
+When you've completed the analysis, I'll load:
+
+{project-root}/_bmad/bme/_vortex/workflows/learning-card/steps/step-04-validated-learning.md

package/_bmad/bme/_vortex/workflows/learning-card/steps/step-04-validated-learning.md

@@ -0,0 +1,139 @@
+---
+step: 4
+workflow: learning-card
+title: Validated Learning
+---
+
+# Step 4: Validated Learning
+
+Now we extract the validated learnings -- the actual knowledge gained from this experiment. What assumptions were confirmed? What was invalidated? What new questions emerged?
+
+## Why This Matters
+
+Validated learning is the currency of Lean Startup. It's the difference between "we tried stuff" and "we know something we didn't know before." Validated learnings:
+- Convert experiment effort into permanent organizational knowledge
+- Reduce uncertainty for future decisions
+- Prevent repeating the same experiments
+- Build confidence (or appropriate caution) for next steps
+
+## Your Task
+
+### 1. Classify Each Original Assumption
+
+For every assumption that was tested, categorize it:
+
+**VALIDATED** - Evidence strongly supports this assumption
+- What specific evidence validates it?
+- How confident are you? (High / Medium / Low)
+- What conditions are required for this to hold?
+
+**INVALIDATED** - Evidence contradicts this assumption
+- What specific evidence invalidates it?
+- Was it completely wrong or just partially wrong?
+- What's the corrected understanding?
+
+**PARTIALLY VALIDATED** - Evidence supports it in some contexts but not others
+- Under what conditions does it hold?
+- Under what conditions does it fail?
+- What's the refined hypothesis?
+
+**UNRESOLVED** - The experiment didn't generate enough evidence to decide
+- Why couldn't you resolve it? (Sample too small? Wrong method? Confounding factors?)
+- What would you need to resolve it?
+
+### 2. Extract the Core Learnings
+
+Distill your findings into clear learning statements. A good learning statement:
+- Is specific and actionable
+- States what you now KNOW (not what you hope)
+- Includes the conditions under which it's true
+- Is falsifiable by future experiments
+
+**Format:** "We learned that [specific finding] when [conditions], based on [evidence]."
+
+**Examples of strong learning statements:**
+- "We learned that cross-timezone teams adopt async video updates at 90%+ participation rates when a team champion models the behavior daily, based on Team A's 91% participation vs. Team C's 62% without a champion."
+- "We learned that novelty drives initial adoption but doesn't sustain it, based on participation dropping from 80% (Week 1) to 67% (Week 2) across all teams."
+- "We learned that we CANNOT conclude whether async updates improve team alignment, because our sample size (n=3 teams) and duration (2 weeks) were insufficient."
+
+**Examples of weak learning statements (avoid these):**
+- "Users like async updates" (too vague, no conditions, no evidence cited)
+- "Our product is better than standups" (not tested, opinion masquerading as learning)
+- "The experiment was successful" (binary thinking, misses nuance)
+
+### 3. Identify New Questions
+
+Every good experiment generates more questions than it answers. What new questions emerged?
+
+**Format for each:**
+- **Question:** What do you now need to know?
+- **Why it matters:** What decision does this inform?
+- **How to test:** What experiment would answer this?
+- **Priority:** How urgently do you need this answer?
+
+### 4. Rate the Overall Learning Quality
+
+Honestly assess the quality of evidence behind your learnings:
+
+**STRONG EVIDENCE:**
+- Adequate sample size
+- Pre-defined success criteria
+- Controlled for major biases
+- Results are replicable
+
+**MODERATE EVIDENCE:**
+- Directional but not definitive
+- Some bias concerns
+- Small sample but consistent patterns
+- Would benefit from replication
+
+**WEAK EVIDENCE:**
+- Very small sample
+- Post-hoc criteria
+- Significant bias concerns
+- Contradictory signals
+
+**ANECDOTAL:**
+- Based on individual stories, not patterns
+- No quantitative support
+- High risk of confirmation bias
+
+## Example
+
+**Assumption Classification:**
+
+| Assumption | Status | Confidence | Evidence |
+|-----------|--------|------------|----------|
+| Teams will do async updates consistently | PARTIALLY VALIDATED | Medium | Works with champions (91%) but fails without (62-68%) |
+| Async maintains team alignment | UNRESOLVED | Low | Alignment improved Week 2 but sample too small |
+| Managers will save time | PARTIALLY VALIDATED | Low | 1 of 3 saved significant time; 1 actually spent more |
+| Screen-sharing adds value | VALIDATED | Medium | Unprompted adoption + blocker discovery incident |
+
+**Core Learnings:**
+
+1. "We learned that async video updates CAN achieve 90%+ adoption when a team champion records daily and the team is distributed across time zones, based on Team A's results. However, without a champion and with co-located teams, adoption drops below 70%."
+
+2. "We learned that video screen-sharing is more information-dense than verbal standup reports, based on the unprompted adoption by 12/19 participants and the discovery of a 3-day-old blocker that verbal standups had missed."
+
+3. "We learned that initial enthusiasm does NOT predict sustained usage, based on the 13-point drop in participation between Week 1 and Week 2, despite positive survey responses throughout."
+
+**New Questions:**
+- Can we design a habit-forming mechanism to sustain participation beyond the novelty period? (HIGH PRIORITY -- blocks MVP decision)
+- Does async work for teams larger than 8 people? (MEDIUM -- scoping question)
+- Would audio-only options increase participation for camera-shy team members? (MEDIUM -- 2 participants raised this)
+
+**Overall Evidence Quality:** MODERATE - Directional findings with clear patterns, but sample too small for definitive conclusions. Should replicate with 8-10 teams over 4+ weeks before committing to building.
+
+---
+
+## Your Turn
+
+Please extract your validated learnings using the structure above. Remember: acknowledging what you DON'T know is as valuable as stating what you do know.
+
+**Key principle:** A validated learning is not "we were right." It's "we now understand this specific thing about how our users behave under these conditions, and here's the evidence."
+
+## Next Step
+
+When you've extracted your validated learnings, I'll load:
+
+{project-root}/_bmad/bme/_vortex/workflows/learning-card/steps/step-05-implications.md