Private by architecture, not by policy.
We build AI that runs entirely on your phone’s neural processor. No cloud. No subscriptions. No data leaves the device. Ever. Works offline, immune to breaches — because there’s no server to hack.
The Opportunity
Every modern smartphone ships with a dedicated AI chip — yet every major AI product still routes your most intimate data through someone else’s servers. After Dobbs, after T-Mobile and 23andMe breaches, after the hearings, 81% of Americans say they’re concerned about data collection (Pew). Meanwhile, cloud assistants fail silently the moment you lose signal. The demand is here. The hardware is ready. No one is building for it.
Three Products, One Architecture
SafeType Keyboard
On-device speech-to-text. $4.99 one-time on the App Store. Zero network access — works in airplane mode, off-grid. Presale live — ships May 17, 2026. Apple-verified “Data Not Collected.”
Cara
Free menstrual tracker. Nothing to subpoena, nothing to breach. Post-Dobbs demand surged 3,000%. Available on the App Store.
Private Assistant
On-device voice AI with open API. $15 individual, $50 household, $40/100 enterprise seats — one-time, no subscription. Platform play — Q4 2026 beta.
Structural moat: Apple, Google, and Samsung depend on data and ad revenue. They can’t fully commit to on-device AI without cannibalizing their core business. We have no such conflict — privacy is our product. Our architecture is also structurally immune to data breaches and works without any internet connection.
Market
$103B
Edge AI TAM by 2030 (29% CAGR)
~100%
Gross margin — zero cloud compute costs
$0
Per-user infrastructure cost (vs. 60–80¢/$ for cloud AI)
Traction
- JiminAI Lie Detector won AI Safety at Funding the Commons hackathon — patent-pending real-time AI deception detection (AUROC 0.983)
- SafeType Keyboard presale live on App Store — App Store listing with Apple-verified “Data Not Collected” privacy label · $4.99 one-time · ships May 17, 2026
- Published research: 18,000 inference runs across 7 AI models — original research
- Self-hosted infrastructure: 4× RTX 3090 (96 GB VRAM), production servers — zero cloud dependency
- Full team assembled: 70+ combined years across cybersecurity, AI/ML, hardware, and finance
Team
Jamal Porter, CEO — 20+ yrs enterprise systems & technical leadership
Cassidy Barton, CTO — 20+ yrs cybersecurity architecture, pen-testing, red-team ops
Eric Basham, PhD — Neuromorphic engineer; Intel, HP. AI hardware specialist
Heather Gorr, PhD — 10 yrs at MathWorks (MATLAB); ML, computer vision, data analytics
Kristine Socall, MBA — 25+ yrs finance across $1M–$300M orgs; SEC-regulated industries
JC Gonzalez-Ramirez — 10 yrs enterprise BD; $5M+ pipeline from zero across SaaS, immersive tech & civic tech
$2.5M Seed Round
Pre-Series A · 18-month runway · Equity
Revenue Model — Moderate Case
| Year 1 | Year 2 | Year 3 |
|---|
| SafeType Keyboard (consumer) | $1.75M | $2.79M | $3.49M |
| SafeType Keyboard (enterprise) | — | $150K | $749K |
| Private Assistant | — | $250K | $1.5M |
| Total Revenue | $1.75M | $3.19M | $5.74M |
Conservative case: $698K / $1.55M / $2.85M. All projections net of Apple’s 30% App Store fee. Near-100% gross margin — zero cloud compute costs.
Competitive Landscape
| Capability | Digital Disconnections | Apple Intelligence | Google / Samsung |
|---|
| 100% on-device inference | Yes | Partial (cloud fallback) | Partial |
| Zero network permissions | Yes | No | No |
| No data collection | Yes | No | No |
| Open API / open source | Yes | No | No |
| Cross-platform (iOS + Android) | Yes | iOS only | Fragmented |
| Works fully offline | Yes | Partial (cloud fallback) | No |
| Breach-immune architecture | Yes | No (cloud storage) | No (cloud storage) |
Unit Economics
SafeType Keyboard
$4.99 retail → $3.49 net (after Apple 30%). Zero marginal cost — no server, no compute, no per-user infrastructure. Near-100% software margin after acquisition.
Zero COGS Advantage
Cloud AI companies spend 60–80¢ per dollar on compute. We spend zero — the model runs on hardware the customer already owns. Every additional user is pure margin.
Use of Funds — $2.5M
- 45% — Engineering & Product Development
- 25% — Go-to-Market & User Acquisition
- 15% — Research Infrastructure
- 10% — Operations & Legal
- 5% — Reserve
Milestones
Q2 2026
SafeType Keyboard & Cara on App Store. Enterprise pipeline activated. Revenue generating.
Q3 2026
Cara user growth. Organic acquisition via post-Dobbs demand. Scale conversion funnel to Private Assistant.
Q4 2026
Private Assistant open beta. Open API for third-party integrations. Platform revenue begins.
Why Now
- Hardware inflection: NPUs now standard in premium smartphones — local inference is no longer theoretical
- Regulatory tailwind: GDPR, state privacy laws, post-Dobbs data concerns are driving consumer action, not just sentiment
- Structural lock-in for incumbents: Apple, Google, and Samsung depend on data/ad revenue. They can’t fully commit to on-device without cannibalizing their core business
- Category is unclaimed: No dominant privacy-first AI platform exists. First-mover advantage is available today — it won’t be tomorrow
