Conversations · apr 2026 → jul 2026

Meetings.

a record of what we said & what we'll do ↘

Meetings on record

13 conversations

apr 2026 → jul 2026

Action items

60 tasks

23 of 60 completed

Cadence

Mondays · noon

recurring sync · monday noon ct

Core attendees

Adam · Ari

+ jen, joe, ablenet slps

23 Jul 2026

Status Check-in — Four Weeks to Submission, and the Scenes Pivot

With AriJen Thalhuber

M2M3BetaProductTimeline

**Honest timeline: about a month, not the beginning of August.** Best case is roughly **two weeks to finish the remaining scenes, then about two weeks in Apple review**. The earlier hope of landing at the start of August does not survive contact with the work left. Better to say so now than to discover it at the deadline.
**Pivoting away from the "scenes" approach at the older tiers.** It was getting too complex and starting to obscure the app's core purpose — communication. Adam flagged it quickly, and the principle behind the call is the one worth keeping: **innovation still has to serve what the app is for**. The scene approach stays where it earns its place, at the youngest tier; the older tiers move to a more direct layout.
**The clinical feedback is genuinely positive.** Alyse and Nicole's early response is strong, and reads as a real innovation signal rather than politeness. That is the first outside confirmation the direction holds up with practising clinicians.
**Onboarding simplification is the top innovation priority coming out of that feedback.** The current flow is too daunting — for parents, for caregivers, and for SLPs. This is the single clearest thing the reviewers want changed, and it is the gate on whether families ever reach the app at all.
**Voice is a hurdle, not a blocker.** Apple's voices are a problem, and the AI voice vendor refused on principle — their lawyers blocked child voice synthesis. It is being treated as something to route around rather than something to wait on; alternatives are still being explored and no other work is being held up by it.

Action items 3 / 3 complete

✓

Finish the remaining scenes — about two weeks of work — then submit to Apple and allow roughly two more weeks for review

OWNERAri

M3-001

✓

Simplify onboarding — both beta SLPs flagged the current flow as too daunting for parents, caregivers and clinicians alike

OWNERAri

M2-009

✓

Keep pursuing a voice alternative — Apple's voices are not good enough and the AI voice vendor declined on principle

OWNERAri

M2-005

17 Jul 2026

SLP Beta Onboarding — Alyse & Nicole Get the App, and the Brief

With AriAlyse RoybaNicole Betts

M2BetaDesignClinical

**Onboarding call for the first two clinical reviewers.** Alyse and Nicole are the only testers for now — Adam's call. Wider team access follows once the narrow group has been through it.
**TestFlight install done live on the call.** The invitation had landed in junk mail; it was found and redeemed on the spot. Worth noting for every future tester, because a silent junk-mail catch looks exactly like a tester who lost interest. Adding testers is a manual step on Ari's side, so the team should email when they have names ready.
**The framing: the innovation is the speed, not only the app.** The point made in the room is that any idea can now be built quickly and cheaply enough to try — so the app should be treated as something that can absorb their feedback, not a finished object to approve or reject.
**The app is built from modular engines** that slot into a common foundation:
- **Speak-Choose-Speak** — lets the child start a conversation rather than only answer one.
- **Sentence engine** — predicts the next words as the user makes selections.
- **Buddy engine** — tracks which words a child actually uses and brings them forward.
- **Custom vocabulary** — lets a family add their own photos and words. Untested so far, and flagged as such.
**It is effectively two products, split by developmental stage.** At around 12 months it is a modelling tool: words only, no genuine back-and-forth. At the next tier up it becomes interactive — pressing, swiping, real vocabulary and real choices.
**Voice is the known weak point, and it was named as one.** Apple has no good child voice, and the AI voice companies prohibit child voice synthesis outright. Apple's family voice blending is being explored as a workaround. Alyse offered her four-year-old's voice as a source; the idea is welcome in principle, but the rights and consent questions around recording a child are not trivial and would need proper handling.
**How to give feedback, and where.** Read the beta website before opening the app — it sets expectations and carries the clinical traceability matrix as the detailed reasoning layer. Then: **bugs and crashes go in-app; ideas, impressions and innovation go on the website**, where three prompts ask what you like, what you don't, and where to push further.
**The rough edges are deliberate.** Polish stopped on purpose to avoid over-building before outside eyes had seen it. Known rough areas were named up front: grid layout, visual aesthetics, and lock/settings access (press and hold the shield icon). The explicit ask was for brutal feedback — the work has been happening in a bubble and needs fresh eyes.

Action items 1 / 4 complete

Review the beta website first and submit feedback — judging whether it sets the right expectations for the wider AbleNet team, not only for yourselves

OWNERAlyse & Nicole

—

Install via TestFlight and test the app — log bugs and crashes in-app, log ideas and impressions on the website

OWNERAlyse & Nicole

—

Email Ari when the team is ready to add more testers — each new tester has to be added by hand before they can get access

OWNERAlyse

—

✓

Keep exploring the voice problem, including Apple's family voice blending, since no acceptable child voice exists off the shelf

OWNERAri

M2-005

07 Jul 2026

Beta Readiness — The Crash, the TestFlight Pipeline & a Narrow SLP Rollout

With AriAdam Wing

M2BetaProduct

**The app was crashing on Adam's iPad, and we found out why.** Tapping anything on the screen after launch dropped him straight back to the home screen. The cause is a **text-to-speech bug** — confirmed from the crash reports submitted through the in-app feedback form. Three crashes are known in total. His iPad meets the iOS 18 minimum, so this is a solvable defect, not a device-compatibility wall.
**He was also testing an old build.** Adam was on v0.0.1.0 while the current build was v0.0.1.02. Automatic updates are switched on for him, but the pipeline that pushes builds into TestFlight isn't connected properly, so new versions never reached him. Fixing that connection matters as much as fixing the crash — testers silently reviewing a stale build produce feedback about problems already solved.
**The beta site is the front door, and it's passcode-protected.** Adam has the beta access page and the passcode. The page is deliberately scoped to beta testers rather than the full product story, and it carries the clinical traceability document — the section laying out the clinical reasoning behind each design choice, for reviewers who want the receipts.
**Deliberately narrow first, not wide.** The plan is roughly **three SLPs in utilization, not forty**. Adam will set clear expectations with them: dedicated hours in the week, structured feedback delivered back. A small group who actually use it beats a large group who install it and forget.
**Internal alignment is in place.** The AI collaboration was announced internally at the Rising Stars meeting, and Adam's team has been briefed on what's being delivered, so the work is visible across the business rather than running quietly on the side.
**An audio-briefing idea for clinicians.** Speech pathologists drive a great deal and consume audio while they do. Adam wants access to the audio-briefing generator to show his team — a possible channel for reaching clinicians in the time they already have, rather than asking for time they don't.
**Target: ** app stable and in SLPs' hands by the end of that week or the following Monday.

Action items 1 / 4 complete

✓

Fix the text-to-speech crash — the app exits to the home screen when anything on the post-launch screen is tapped

OWNERAri

M2-005

Reconnect the TestFlight distribution pipeline so new builds reach testers automatically instead of leaving them on an old version

OWNERAri

—

Keep the first rollout to roughly three SLPs, not forty — Adam sets the expectation of dedicated hours and structured feedback in return

OWNERAdam

—

Give Adam access to the audio-briefing generator so he can show it to his team

OWNERAri

—

22 Jun 2026

Product Strategy — SLP-First, Ease Beats Delight & the Kitchen Demo

With AriAdam Wing

M2M3ProductDesignStrategy

**The journey is for the SLP, not only the child.** The reframe of the session: "the journey isn't even for the child — it's for the SLP. You're taking them on a journey." The clinician's confidence is what carries the family and the child, so the experience is designed for the SLP first. Success rests on a **three-legged stool — SLP, child, and family buy-in** — with the **SLP as the priority leg**, because a comfortable clinician brings the family and child along. The child and family largely follow the clinician's lead.
**The data behind it.** AbleNet's updated survey of **38,000 SLPs found no correlation between the specific app and child-success outcomes** — the app is essentially app-agnostic. Clinicians tend to prescribe whatever they were trained on (neighboring clinics prescribe entirely different apps based on which vendor ran a nearby training), and outcomes come out equally effective. The implication: the win isn't a "better app" in the abstract — it's earning clinician confidence. The aim isn't to claim perfection; it's to introduce a genuinely new approach and spark the field to do better.
**Ease beats delight.** The design north star named in the room. It echoes AbleNet's guiding principle — **"least possible equals best outcomes"** — and points one way: make the app so simple it needs no manual to explain it. The instinct is to **simplify and hand the user more control**, not to add surface. (A useful counterweight to the earlier "invest in fun/delight" thread — delight stays an accent; ease is the spine.)
**"Swell" — let ranked vocabulary do the work.** The concrete mechanic: rank the vocabulary and let that ranking drive **density and "swell"** (tile size). Pick a scene and the words that matter most grow — choose **snack, breakfast, or lunch** and the relevant items swell up while the rest recede, with a standard fallback always available ("I'm hungry / eat"). Choosing gets faster because what a person reaches for most is the easiest thing to find — not a flat grid of equal-sized tiles. Builds directly on the 2026-06-11 work on reducing busyness and tuning per-tier density.
**World-view landing before the kitchen.** The app should open on a "world" view — a home screen that orients the user — before navigating into the kitchen. Kitchen/food is the only navigable area for now, but without that guiding entry point the experience "is a bust." The world view is what frames it.
**Deployment plan.** Target the food/kitchen demo (with the world-view landing) for **Thursday, June 25**, with that day as the checkpoint. Deploy to internal SLPs on **iPads** with feedback built into the app — the kind of thing they'll reach for at night, not nine-to-five — scaling **10 → 20 → 30** reviewers. The one firm gate is readiness: a half-finished build draws "this isn't good enough" noise instead of useful signal, so it must be polished enough first. The week of June 29 (incl. July 4) is an unusually long AbleNet break with many staff out, so if it isn't ready on the 25th the next stage gate is **July 6** rather than deploying into the quiet week.

Action items 2 / 4 complete

✓

Finish the food/kitchen demo — including the world-view landing screen so the user is oriented before entering the kitchen — by Thursday, June 25

OWNERAri

M3-001

Checkpoint on June 25; if it isn't polished enough, slide the next stage gate to July 6 rather than deploy into the long AbleNet holiday week (week of June 29, incl. July 4)

OWNERAdam & Ari

—

Deploy to internal SLPs on iPads (10, then 20, then 30) with feedback built into the app — once it's polished enough to generate signal, not noise

OWNERAdam

—

✓

Implement 'swell' — rank vocabulary so tile size and density scale with relevance within a scene

OWNERAri

M2-002

11 Jun 2026

In-Person Design Review & Milestone Check-In

With AriAdam Wing

M1M2M3DesignBrandingProduct

**Screen busyness is the #1 fix before any external SLP review.** The kitchen background currently competes with the vocabulary items. Plan: fade the background to ~10% opacity (implied context, not dominant) with a user-adjustable slider; selectable objects (fridge, cabinet, oven) stay at full intensity, everything else fades. Guiding principle: reduce anxiety for SLPs, who are the primary gatekeepers.
**Tile layout — images are too small.** Target ~85% of each tile as image with minimal white space. The updated illustrated (non-emoji) imagery was confirmed as a significant improvement.
**Pete the Cat as a north-star visual reference** — simple, bold, child-friendly art with a clear, consistent visual language. Simplicity matters more than realism or detail for a child who is engaged but linguistically early.
**Scene interaction — solving the "objects in a realistic scene" problem.** A realistic kitchen creates an unsolvable proportionality problem ("where does the chicken go?"). Ari raised a meal-contextual food platter as one possible direction (contents changing by meal, items expanding on press, environmental objects faded back). This was an idea explored in the room, not a settled decision — it has since been revisited in favor of a tiered grid approach, where the same objects organize and the scene recedes as the child's tier grows (see Documents → "Tier Interaction — One World, Three Stages"). The durable principle stands: action and context matter more than realistic set dressing, and reducing busyness is the priority before external SLP review.
**Engagement is a game to motivate, not a quiz.** "Do you like chicken?" builds engagement; "Does milk go in the fridge?" trains vocational reasoning, not expressive communication — kept available but used deliberately, not as a core navigation layer.
**Fun and delight need more investment.** Example: selecting ice cream could trigger a short comedic animation. Emotion/feelings display preferred as a rainbow/circular swipeable rather than a flat grid — joy is as important as function.
**Sentence engine — clarify built vs. suggested.** The engine works (pronoun "I" anchored on the right, builds left-to-right into "I want pancakes," surfaces context-appropriate vocabulary), but selected words and anticipated words look too similar. Fix: give confirmed selections a distinct visual treatment. Key principle — build on language, don't restart it: after "I want pancakes," the child should extend ("more," "again"), not rebuild from scratch. This continuity is a meaningful differentiator vs. Proloquo2Go.
**Three tiers built in parallel** (different variable sets, not sequential phases). Density per tier (high/med/low items on screen): Tier 1 = 96/86/76, Tier 2 transitional, Tier 3 = 55/65/75. Tier 1 = modeling & receptive language (no sentence building expected); Tier 2 introduces the sentence engine and vocabulary building; Tier 3 adds the dialogue/chat layer for full expressive communication.
**Verb animation — strong developmental case.** Children map action images (walking, eating, jumping) to universal, immediate meaning, whereas static noun images are more arbitrary. Plan: pressing a verb plays a short animation of the action. This is where visual investment pays off developmentally.
**Target user clarified: ** chronologically ~36 months, cognitively ~36 months, linguistically ~8 months. The design bar is whatever keeps a 3-year-old engaged — but it cannot be busy.
**Pedagogy is itself an asset.** The developmental rationale behind each design decision is part of the marketing and trust-building story for SLPs — transparency about methodology builds the trust that is the central challenge in AAC adoption. Art-style research: stick figures vs. illustrated does not meaningfully change outcomes; children are drawn to colorful imagery, which aligns with the current direction.
**On-device intelligence, no AI in-app — confirmed the right call.** Fully on-device, no network dependency. Framed positively as "using technology to determine appropriate words." Future possibility flagged: generating vocabulary symbols on the fly rather than serving static assets, and exploring more of what the Apple chip offers on-device.
**Branding needs to be "pediatricized."** The current "M with a leaf" logo reads as a different product category and no longer matches the app experience. Direction: logo, name, and app should feel like one coherent product for ages 1-4; the "Meadow" name itself was flagged for revisiting. Website design praised as strong and iterating well.
**Milestone status.** M1 deliverables largely signed off. M2/M3 are being built together (scene + engine), not sequentially: navigation, communication board, sentence engine, text-to-speech, and core word bar are functional; avatar/buddy in progress; parent gate is live (age, voice, speech rate, tier, speech-bubble timeout, vocabulary by scene, signing toggle, celebration intensity, routine glow, routine times); routine-aware timing works (knows 7 AM is breakfast and surfaces relevant vocabulary). Camera input (custom photos) confirmed as needed; event logging for usage insights in progress.

Action items 3 / 9 complete

Reduce screen busyness — fade the scene background to low opacity (~10%) with a user-adjustable slider; keep selectable objects at full intensity

OWNERAri

—

✓

Resolve the kitchen scene-interaction model so vocabulary objects aren't placed into a realistic scene (the proportionality problem) — organize the objects clearly instead

OWNERAri

M3-001

✓

Clarify the sentence engine visually — give built/confirmed words a distinct treatment (outlined/colored) so they read differently from suggested words

OWNERAri

M2-003

Explore and test verb animation (pressing an action word plays a short animation of the action); send a few options for review

OWNERAri

—

Review GoTalk and Proloquo2Go on a real unlocked device for competitive reference

OWNERAri

—

Investigate what on-chip Apple processing is available for future on-device intelligent features

OWNERAri

—

✓

Revisit the logo and the 'Meadow' name to fit a pediatric (ages 1-4) audience

OWNERAri

M1-008

Raise the app-innovation question with Harper Collins (Pete the Cat) at the upcoming NY/NJ meeting

OWNERAnne Wolf

—

Send the milestone invoice for payment processing

OWNERAri

—

28 May 2026

Product Strategy & 30-Day Delivery Target

With AriAdam Wing

M1M2StrategyProductVoiceTimeline

SLP feedback incorporated — all five of Alyse's requests addressed: grid dimensions are flexible (any size to match language milestones and chronological age), five new T1 words added (me, play, cold, hi, bye — total now 22), celebration toggle added (Full/Subtle/Off), quick chat scaffolding toggle added, art direction confirmed as Bright & Bubbly.
42 customizable slots confirmed — 18 people, 5 things, 4 places, 15 wildcards. Camera input for custom additions. Gives SLPs and parents flexibility to personalize vocabulary for each child.
Apple TTS child voice confirmed as requirement — ElevenLabs dropped for child speech output. Mixing AI-generated voice with Apple TTS for custom words would feel inconsistent. ElevenLabs retained only for companion character. V2 plan: server-side voice pack generation downloaded to device.
Dialog is the goal — not just single-word communication. The app should drive conversation. Adam emphasized this as a core product principle: a child saying "Hi, my name is Adam. What's your name?" starts a dialogue, which is what Meadow should enable.
Three-tier system validated — Tier 1 (heavy scaffolding, engagement focus), Tier 2 (reduced support), Tier 3 (transition preparation). "Less is more" at early stages, with more vocabulary unlocking as the child develops. Bounded by design — intentionally transitional.
30-day delivery target set — functional kitchen scene on TestFlight by ~June 27 for 6 internal reviewers. Apple approval process is the bottleneck, not development. Kitchen scene prioritized because it's the hardest complexity gate.
14% stat shared — 14% of AbleNet users don't need the device after one year. This is unprecedented in the field and reinforces Meadow's transitional design philosophy.
Future app development acknowledged — children who still need AAC after 48 months will likely need a different app for the 4-8 year and 8-12+ year age groups. This is expected and validates the bounded scope. Potential for a second app project.
AbleNet expansion interest — multiple departments exploring AI-powered projects. Customer experience team and medical funding team both interested in connecting with Ari for potential agent development work.
Visual style locked — Bright & Bubbly confirmed. Adam noted Storybook feels "kinda old." Art-agnostic architecture means styles can be swapped, but Bright & Bubbly is the launch direction.

Action items 2 / 5 complete

✓

Deliver TestFlight build (kitchen scene) within 30 days (~June 27)

OWNERAri

M3-001

Identify 6 internal TestFlight reviewers

OWNERAdam Wing

—

Connect Ari with customer experience and medical funding teams for potential AI agent projects

OWNERAdam Wing

—

✓

Complete product branding and website

OWNERAri

M1-008

Schedule direct meetings with other department heads

OWNERAdam Wing

—

28 May 2026

SLP Design Review — Feedback Integration & Tier 3 Design

With AriAlyse Royba

M1M1-003M1-006SLP-ReviewProduct

All M1 feedback items addressed and confirmed — grid dimensions flexible (any size), five words added to T1 (me, play, cold, hi, bye), celebration toggle (Full/Subtle/Off), scaffolding toggle, Bright & Bubbly art direction. Alyse confirmed each change resolves her original feedback.
Celebration toggle rationale deepened — immediate reinforcement is associated with ABA therapy approaches that aren't always neurodiverse-affirming. SLPs want communication to drive meaning through environmental response, not just app rewards. The toggle lets SLPs choose between environmental-only and app-assisted celebration. Fading system prevents reward dependency.
Carrier phrase scaffolding scoped to Tier 2 only — "I want ___ please" fill-in templates risk replacing core word learning if always available. Families may rely on carrier phrases and children don't learn to combine words independently. Agreed: available in Tier 2, removed in Tier 3, toggleable on/off within Tier 2.
Prediction vs. comprehension tension explored — the core challenge for 36+ month users. Predictive follow-ups speed communication, but risk putting words in the child's mouth. Agreed that core words must always remain accessible outside of predictions so the child can express anything, not just what the app suggests.
Intent bubble concept discussed — for Tier 3, child selects communicative intent (request, comment, question, share) before the app generates follow-ups. Prevents assumptions about why a child tapped a word. Core words still available for full open expression regardless.
Buddy backpack concept validated — companion character grows with child, wears a backpack storing frequently used words for quick access. Customizable by SLP and parent. Similar to "my phrases" in other apps but integrated into the companion character.
Three-click navigation rule confirmed — never more than 3 taps to reach any word. Alyse confirmed this is exactly what SLPs look for. Current motor planning apps achieve this but in ways parents find confusing. Meadow's scene-based approach bridges that gap.
Cross-scene navigation validated — child can switch scenes (playground to restaurant) to talk about anything from anywhere. Left sidebar expands to places/routines in Tier 3. Addresses the "off-topic conversation" concern from initial feedback.
Innovation confirmed by SLP perspective — Alyse and colleague Nicole both validated the approach as genuinely innovative for early intervention. Features like scene-aware vocabulary and the tiered interaction model are not available in any current app.
SLP as ongoing champion — Alyse will continue reviewing as deliverables publish and will pull in additional SLPs from her team if needed. Iterative review approach given the fast development timeline.

Action items 1 / 3 complete

Review previous SLP app consultation notes and share any relevant insights

OWNERAlyse Royba

—

✓

Incorporate Tier 3 design feedback: intent bubble, carrier phrase limits, backpack concept

OWNERAri

M1-003

Continue reviewing M2 deliverables as they publish

OWNERAlyse Royba

—

26 May 2026

Design Review Kickoff & Usage Statistics Requirement

With AriAdam Wing

M1M1-004M1-005StrategyProductInsuranceCompetitive

Design specs validated at first glance — Adam and Jen reviewed initial pages together, called it "completely incredible" and "nothing anybody's ever seen." Said it's exactly what he expected from last week's conversation and can review via comments without a giant dialogue on each section.
SLP clinical review in progress — Adam is pulling in Elise from the utilization team to review specs from an SLP perspective. Goal: validate that nothing causes anxiety for clinicians, not to redesign. If she flags an area, we pull it or adjust. Reviews expected within 24-48 hours before Thursday follow-up.
Project management framework praised as innovative — Adam highlighted the portal workflow itself as a differentiator, nobody in his world delivers project visibility this way. His Minneapolis team still doesn't understand the speed or capability.
New requirement: on-device usage statistics. Insurance funding process has three stages where data matters: (1) benefit check, (2) SLP eval requiring utilization evidence, (3) ongoing usage insight for parents and SLPs. Key metrics: how often the device is used, what words are selected, which scenes/contexts. All data stays on-device, no third-party server access. Framed as "usage statistics, not data collection" to stay clean with App Store and privacy posture.
Insurance context unpacked — majority of insurances now require data during the eval phase to justify funding. After funding, insurance relationship is over (except 5-year warranty), but SLPs continue billing insurance and recording efficacy. Parents want to see if their child is actually using the device daily.
Proloquo2Go comparison — Proloquo sends anonymous utilization data to their servers for research purposes (Australian usage studies, etc.). Our approach is the opposite, data stays with the triad (SLP, parent, child), never the developer.
Contextual awareness confirmed as key differentiator — setting the kitchen scene means highest-probability kitchen items surface automatically. Reduces cognitive and physical load. This is functional communication, not paragraph building.
Competitor limitations reinforced — Proloquo2Go is a 15-year-old interface with no swiping despite universal child familiarity with the gesture.
Swiping validated as interaction pattern — swiping between scenes is natural for this age group. Every child knows the gesture from birth. Built into the conversation tier already, Adam hadn't noticed it in the spec yet and was enthusiastic.
Device size is irrelevant to architecture — grid layout adapts to fine motor skill level, not screen size. A 12-month-old gets bigger targets and simpler grids; a 48-month-old gets denser layouts. The tiered progression already handles this naturally.
Post-48 months is AI territory — Adam and Ari aligned that lifelong AAC users will eventually just use AI directly. Meadow doesn't need to build for that; children graduate off the device.
Adam pushing to accelerate Ari's business positioning — asked about other app ideas, other clients. Acknowledged this is a "critical time for the bifurcation" and wants to help Ari grow.

Action items 0 / 5 complete

Add on-device usage statistics to development roadmap (frequency, word selection, scene usage)

OWNERAri

—

Review all design spec milestones and add comments via portal feedback

OWNERAdam Wing

—

Have Elise (SLP, utilization team) review design specs for clinical validation

OWNERAdam Wing

—

Send sample insurance data collection worksheet to Ari for reference

OWNERAdam Wing

—

Follow-up meeting Thursday 11AM to discuss review feedback

OWNERBoth

—

18 May 2026

Product Scope Reframe & Core Requirements

With AriAdam Wing

M1M1-001M1-002M1-003StrategyProductVocabulary

Age range reframed: Drop 0-12 months entirely. Target 12-48 months chronological age — children with missed language milestones, not a specific cognitive band. Supersedes the earlier 0-36 month decision.
Milestone-driven tiers: Adam requested a matrix from 12 to 48 months showing expected receptive and expressive vocabulary at each milestone (12mo: mama/dada, 50 receptive; 18mo: about 50 words, 2-3 word combinations; 36mo: about 2,000 receptive, 3-4 word sentences).
Communication-first device: This is a speech generating device (E2510), not a teaching tool. The child must be the one selecting words — otherwise it's language-learning software, not an SGD. Teaching features like Learn mode stay supportive and below the fold.
14% of kids graduate off devices to functional verbal communication — unprecedented outcomes. The device supplements weekly SLP therapy sessions.
Core word board alignment: Audit current vocabulary against established core word board lists familiar to SLPs. Avoid proprietary standards that require SLP education on a new framework.
"I love you" is permanent and global: It must be accessible from every screen, in every context. Based on family testimonial data, hearing "I love you, mama" through the device is consistently the most meaningful moment for families.
Text labels as an optional toggle: Some SLPs require text for literacy development, others prefer image-only. Make it a setting. FlexSpeak was cited as precedent.
Scope narrowed to "My House": Kitchen, bathroom, and bedroom as initial contexts. School was removed for the 12-48 month target age. Daycare remains an acceptable secondary community reference.
Routines over scenes: "It's not the kitchen, it's time to eat." Organization shifts from location-based to activity-based — getting dressed, bath time, bedtime.
Imagery is the differentiator: "Everybody's pictures suck." Animated imagery for all vocabulary items remains a major product value, with a possible tiered approach where animation is funded and static assets are free.
"What do you want?" is the core use case: Are you hungry? Hot? Tired? Want orange juice? Want your brother? Want to watch TV? Want to go to bed? This is the communication path that must be efficient.
Build the app around Taylor's testimonial: specificity needs like what color popsicle, routine communication like what TV show, emotional expression like "I love you, mama," and tantrum prevention through clear communication.
Learn mode validated: sunflower button, spotlight-in-scene, "say it with me" prompting, 5-second auto-advance, and big or small celebrations. Designed for 2-3 minute facilitated sessions and acceptable as a supplementary feature.
Not a diagnostic tool: Users arrive with deficits already identified through the pediatrician-to-SLP referral path. Meadow adapts to what the SLP already knows — it does not assess or determine the child's level.
SLP-parent-child trinity: SLP leads, parent enables, child uses. All three are inseparable. AbleNet's funded model keeps the SLP connected, which is a business differentiator.
Engagement problem is already solved: iPads proved kids with disabilities can use touchscreens. Don't over-engineer basic engagement scaffolding. This population is mostly language-delayed, not necessarily cognitively impaired.
Modeling drives verbal outcomes: 14% of kids graduate off devices to functional verbal speech. The device is the modeling tool, and repeated use can lead to expressive speech emerging.
Family video research directive: Watch Taylor, Kim, Megan, and other family stories on AbleNet's site. Extract common themes for specification validation. Adam: "This is a treasure trove. Have your AI watch this and suggest the iteration."

Action items 8 / 8 complete

✓

Build 12-48 month expressive/receptive milestone matrix using ASHA-standard benchmarks

OWNERAri

M1-001

✓

Audit current 64-word vocabulary against established core word board lists

OWNERAri

M1-002

✓

Add missing vocabulary essentials: colors, 'I love you' expression

OWNERAri

M1-002

✓

Define permanent 'I love you' global accessibility in navigation architecture

OWNERAri

M1-003

✓

Develop detailed Learn mode interaction spec (timer logic, celebrations, randomization)

OWNERAri

M1-003

✓

Narrow initial content scope to 'My House' routine framework

OWNERAri

M1-002

✓

Analyze AbleNet family testimonial videos (Taylor, Kim, Megan) for feature validation

OWNERAri

link

✓

Update comprehensive spec document incorporating all discussed changes

OWNERAri

—

15 May 2026

Project Kickoff & Product Strategy

With AriAdam Wing

M1ContractStrategyProduct

SOW approved and ready for signature — Adam delegating vendor setup to Joe
Project team confirmed: Adam, Ari, Jen, and Joe (4 people)
Monday noon recurring check-ins established
Deliverable portal shared — transparent tracking of all artifacts and decisions
Starting with kitchen scene deep dive, vocabulary architecture as first deliverable (needs SLP input)
Hybrid approach confirmed: scene-based navigation (differentiator) + flashcard mode (SLP-familiar, reference: SuperDuper)
Target vocabulary: ~1,000 words vs competitors' broader catalogs — intentionally focused
Potential Disney/Harper Collins character licensing — Adam updating contracts for "free digital assets"
Work-for-hire IP structure confirmed: AbleNet owns everything
Vertical slice methodology: food and bathroom first (essential 0-3 needs)
Visual design system with flexibility for future character integration

Action items 4 / 4 complete

✓

Call Joe for vendor account setup and contract signing

OWNERAdam Wing

—

✓

Continue vocabulary architecture work

OWNERAri

—

✓

Send deliverable portal link to Adam

OWNERAri

—

✓

Update contracts for free digital assets (character licensing)

OWNERAdam Wing

—

14 May 2026

SOW v3 Review & Strategic Alignment

With AriAdam Wing

M1SOWStrategy

SOW v3 agreed: work-for-hire IP to AbleNet, Ari publishes app to maintain liability separation
AbleNet stays anonymous as publisher — maintains agnostic positioning in AAC market so they can support any app through existing service model
Plan to load app on all clients classified as 0-3 years via self-service module
Coming out strategy at ASHA conference this fall — intentionally under radar from competitors until then
Target age confirmed: 0-36 months, language milestones up to 2-3 word combinations max
Design direction: storybook aesthetic over cartoon style
Core features: multiple scene generation, communication board always available, story-based interaction as primary interface
Potential high-profile SLP partnership for credibility (from 9,000+ benefit check database)
Ari needs unlocked QTF device for development
Support structure leverages AbleNet's existing SLP success team — scheduled meetings through hub, routes app questions to service team

Action items 4 / 4 complete

✓

Review updated SOW with Adam's edits merged

OWNERAri

—

✓

Arrange QTF device shipment through Joe

OWNERAdam Wing

—

✓

Legal review for publishing liability structure

OWNERAri

—

✓

Develop support routing strategy for app questions

OWNERAdam Wing

—

14 Apr 2026

SLP Team Design Review — Interaction Design Walkthrough

With AriAbleNet SLP team

M1M1-001ProductSLP ReviewVocabulary

Presented HTML interaction design mockup to AbleNet SLP team for clinical feedback.
Robust vocabulary endorsed — SLPs want full access, not a stripped-down set. "Don't shy away from robust." When a 2-year-old isn't speaking, give full access to understand what they're capable of.
Multiple access paths needed — vocabulary should be reachable through different navigation routes, not locked to one path.
Core words must not depend on the scene — build faster, easier access to words no matter the environment. Core vocabulary travels across all contexts.
Core vs. fringe distinction matters — SLPs think in these categories. Core words are high-frequency across settings; fringe words are context-specific.
Start with a strong default, then personalize it with the child's name, family members, pets, and familiar people.
Offer focus filters at routines like dinner time, but keep the system open enough to follow natural topic changes.
Real photos were requested — families want to see their own pictures in the app, not just illustrated symbols.
Development support features are still wanted: grammatical support, language-building tools, multimedia, parent and caregiver support resources, word finder, and vocabulary filtering.
Parent modeling support is critical — Meadow should help families model language use, both inside and outside the app.
In-device coaching is needed — the app itself should help explain "how can we help," alongside the companion-app idea for parent and SLP support outside the device.
Sound bridge and feel pulse concepts were well received as sensory reinforcement options.
The halo concept for the avatar resonated as a way to surface more word options.
No communication sabotage — this is a known AAC danger area. Meadow must not inadvertently restrict or frustrate communication attempts.
Early intervention urgency matters — "They need lift before muscle memory takes over." The language-intervention window is time-sensitive.
Context-aware vocabulary surfacing was discussed as a way to help children and partners find the right words faster.
More flexibility and extensibility are needed over time, with tools that feel inviting instead of intimidating.
The reading tool received positive feedback from the SLP team.
TouchChat and Word Power were used as comparison points for expectations.
AI personalization on the backend is acceptable if it stays helpful and not creepy.
Meadow should scaffold just enough without trapping the child in a control loop.
Gamification is welcome when it increases engagement in service of communication rather than distracting from it.

Action items 5 / 5 complete

✓

Incorporate SLP vocabulary architecture feedback into interaction design spec

OWNERAri

M1-002

✓

Design core-word navigation that works across all scenes and contexts

OWNERAri

M1-003

✓

Explore real-photo integration for personalized vocabulary items

OWNERAri

—

✓

Design parent and caregiver coaching experience in-app and for the companion-app concept

OWNERAri

—

✓

Research sensory feedback options for communication reinforcement

OWNERAri

—

03 Apr 2026

Innovation Proposal Presentation

With AriAbleNet team

Proposal

Presented Meadow concept to AbleNet team — scene-based AAC for 0-5 pre-verbal children
Strong interest in "innovation" angle — they want something the market doesn't have
Agreed to proceed with SOW development for phased engagement
App Store distribution model discussed — independent release under non-profit sponsor

Action items 2 / 2 complete

✓

Draft SOW for review

OWNERAri

M1-001

✓

Confirm internal budget approval

OWNERAbleNet

—