AC.
07 / BRIGHTCANARY · DESIGN CONSULTANT · 2023–2025

Enough for a parent to act on. Not enough to make them a spy.

I designed an AI product that helps parents understand their kids' online lives while resisting the easy, invasive version of that idea.

Role
Design Consultant
Scope
Product · Brand · Research
Team
Founders + CTO
Focus
AI · Research · 0→1
Read the case ↓
70%Using iMessage
monitoring
3–4×Sessions
per day
8–10mPer session
<2%Churn
MEDIA 01 · HEROBrightCanary's parent briefingShow summary, emotion context, and one thoughtful action without raw private-message content · 16:9 (2400×1350px).

01 — THE SETUP

Parents wanted context. They did not need another stream of content to police.

BrightCanary helps parents understand what is happening in their kids' digital lives. I worked as the sole designer with the founders and CTO across product, UX, brand, AI, research, and paywall.

Research moved the product toward private messaging, and iMessage became the core experience.

That made the value much clearer. It also raised the privacy stakes immediately.

Raw
messages→AI
summary→Useful
context→Parent
action
MEDIA 02 · JOURNEYFrom message stream to parent briefingShow hierarchy and progressive disclosure using redacted or fabricated message content · 16:9 (2400×1350px).

02 — THE CHALLENGE

Too shallow was useless. Too detailed turned the parent into the monitor.

Summarize enough to support a conversation. Add friction before revealing the thread.

If summaries were vague, parents still had to read everything. If they were exhaustive, the product became surveillance with better typography.

We kept raw threads behind a confirmation that named the privacy consequence. The extra step was intentional, even though removing it might have increased engagement.

SUMMARYPRIVATE THREAD
MEDIA 03 · GUARDRAILDeliberate friction before detailShow the summary-to-confirmation-to-thread sequence so the decision is self-evident · 4:3 (2400×1800px).

03 — WHAT I DID

AI can label a pattern. Parents still need context.

  1. Designed emotion analysis with a real framework.Structured five primary emotions with three secondary states each, plus APA-informed categories and 24-hour curves, so the experience had more grounding than a happy/sad label.
  2. Explained the language around the signal.Added slang explanations and context so parents could understand what a phrase meant without pretending the model understood the whole relationship.
  3. Designed an early LLM assistant.Created Canary in 2023 as a guardrailed assistant for questions, summaries, and possible next steps.
  4. Tested onboarding instead of assuming shorter was better.Marketing proposed eight questions. I expected fewer to win. The eight-question version converted better, improved engagement, and gave us useful segmentation.

04 — WHAT HAPPENED

The private-message use case became the product people came back for.

70%Of users adopted iMessage monitoring

3–4Sessions per day

8–10 min.Average session length

<2%Churn

MEDIA 04 · PRODUCTEmotion, language, and CanaryUse three focused mobile sequences with parent decisions legible and AI supporting the story · 16:9 (2400×1350px).
TESTIMONIAL PLACEHOLDER

BrightCanary founder or CTO quote

Best quote: privacy judgment, product range, or translating sensitive AI behavior into something families could use.

05 — LOOKING BACK

The restraint was part of the product.

We could have made private information easier to open and harder to stop reading.
We chose to help parents know when to have a conversation instead.