Product Roadmap

From proof to scale,
in three gated phases.

An outcome-driven, staged release. Each phase tests an assumption before the next one commits resources, features ordered by RICE, scope held by MoSCoW, and privacy built in from Month 1 rather than bolted on for compliance.

gated phases over 6 months

405

top RICE score, Zoom API

< 15%

false-positive gate before pilot

2 of 3

pilots must renew to scale

The plan

Three phases, each with an exit gate.

Phase 1 · Now

MVP Build

Months 1–2

Prove the core mechanism works on-device, without surveillance.

›Two-Witness Rule engine, an alert fires only when facial and behavioral signals agree
›Zoom API integration (top RICE feature), behavioral signals from poll, chat & dwell timing
›On-device facial processing, no raw video ever stored or transmitted
›Per-student baseline calibration inside the first session
›3-state engagement classifier (Engaged / Drifting / Disengaged)

Gate

Exit gate (end M2): engine runs on-device with zero raw video stored; per-student baseline calibrates within one session; internal false-positive rate < 15% on test data.

Phase 2 · Next

Pilot

Months 3–6

Put it in real classrooms and prove alerts turn into instructor action.

›Closed beta at 3 design-partner universities
›Real-time instructor dashboard + Private Ping
›Cameras-off behavioral fallback mode
›AI Instructor Copilot v1, one-click recovery actions
›Live Student Pulse validation + dismissal-reason capture
›Pilot debrief: instructor NPS, false-positive rate, friction points

Gate

Exit gate (end of pilot): 3 universities live; ≥ 70% of pilot instructors return in consecutive weeks; NPS baseline captured; friction points documented for Copilot v1.

Phase 3 · Later

Scale

Months 7+

Convert evidence into paying institutions, only after the gates clear.

›Aggregate curriculum analytics live (Month 5), class-level only, no individual records
›Engagement ↔ quiz/comprehension correlation analysis (the go/no-go gate)
›Expansion to 10 Tier-3 universities
›First institutional SaaS contract signed
›Renewal / expansion intent from 2 of 3 original pilots

Gate

Exit gate (scale): all scale thresholds confirmed, FP < 15% overall / < 25% subgroup, actionable-alert rate > 80%, consent > 70%; first contract signed; 2 of 3 pilots renew.

How we ordered the build

RICE prioritization.

Score = (Reach × Impact × Confidence) ÷ Effort. Reach is estimated pilot users per quarter, Impact on a 0.25–3 scale, Confidence a percentage, Effort in person-months. Zoom integration and the Two-Witness engine score highest and anchor the MVP; curriculum analytics scores lowest and is deliberately deferred to Scale.

Feature	Reach	Impact	Conf.	Effort	RICE score	Phase
Zoom API integration	300	3	90%	2	405	MVPMust
Two-Witness Rule engine	300	3	80%	4	180	MVPMust
Real-time dashboard	300	2	85%	3	170	PilotMust
Cameras-off fallback	200	2	80%	2	160	PilotShould
AI Instructor Copilot	300	2	70%	4	105	PilotShould
Curriculum analytics	100	2	75%	3	50	ScaleCould

MoSCoW held Must-have features to ≤ 60% of total effort, leaving room to absorb pilot findings.

Cross-check

Kano classification.

We separated features that merely satisfy from those that differentiate, then secured the must-be and performance features before leaning on the delighter as the wedge.

Must-beCameras-off fallback

Its absence breaks the product for students who disable cameras; its presence earns no praise. Table stakes.

PerformanceReal-time dashboard · Curriculum analytics

More accuracy and coverage yield proportionally more satisfaction. Worth investing to make better.

DelighterTwo-Witness Rule

Instructors don't expect a system that refuses to flag on a single misread face. This restraint is the positioning wedge.

Are we there yet?

Product-market fit.

Following Andreessen, real PMF shows in pull behavior, not vanity numbers. Instructor behavior and institutional conversion are the signals.

> 70%

Repeat instructor use

Pilot instructors return in consecutive weeks

> 80%

Flag agreement rate

Share of alerts acted on rather than dismissed

> 40

Instructor NPS

By end of pilot debrief

2 of 3

Pilot renewal intent

Universities intend to continue or expand

Go / No-Go, end of Month 4. No scaling is committed until the pilot proves engagement signals correlate with real quiz & comprehension outcomes.

From proof to scale,in three gated phases.

Three phases, each with an exit gate.

MVP Build

Pilot

Scale

RICE prioritization.

Kano classification.

Product-market fit.

From proof to scale,
in three gated phases.