AI that runs on your device.
In the woods. On a plane.
No signal required.
We disconnect you from the systems that profit from watching you. On-device AI — models that run entirely on the hardware you already own. Private by architecture, not by promise. No cloud. No subscriptions. No data leaves your phone.
How most AI works
Your words leave your device, travel to a server, get processed, return as a response. That server logs what you said. It builds a profile. It trains the next model.
How we work
Your words go to the neural processor in your phone. Apple calls it the Neural Engine. Qualcomm calls it the Hexagon NPU. The response comes back in milliseconds — and nothing leaves. There is no server to breach. No database to subpoena. No company incentivized to keep what you typed.
Private by architecture means it cannot betray you. Even if we wanted to.
Two models. One structural difference.
SafeType Keyboard
The iOS keyboard that physically cannot spy on you. Every keystroke stays on your device — by architecture, not policy.
Pre-Order on the App Store →- ✓ $4.99 — one-time purchase, no subscription
- ✓ Releasing May 17, 2026
- ✓ Data Not Collected — verified by Apple
- ✓ 100% on-device — zero network access
- ✓ On-device speech-to-text (no server)
- ✓ Requires iOS 17.0+
One-time pricing is not a marketing tactic. It is a business model statement: we make money when you buy the software. Not while you use it. That means your data is structurally worthless to us — and that is the point.
We ran a live on-device inference demo for engineers and researchers at Frontier Tower. No slides. No pre-recorded video. Real-time, on hardware. The result: a system that predicts whether an AI model is hallucinating before it finishes responding. The room had not seen that before.
See our hackathon submission →What if your software couldn’t betray you?
Each product answers that question at a different scale. Your words. Your body. Your life. Three products that prove AI doesn’t need your data to work — it needs your hardware. You already own it.
Private Assistant
More than a voice assistant — it sees, reads, organizes, and remembers. Point your camera at an invoice and it’s OCR’d into a clean PDF. Photograph inventory at a warehouse and it builds a searchable catalog. Say “show me invoices from job 400” three months later and they appear. Voice, vision, document processing, and contextual recall — all running on your device’s neural processor. No cloud. No uploads. No one listening.
- Not just a voice assistant — a complete on-device AI
- Any app can connect — no approval gates, no walled garden
Cara
Answers the question for your body. Your cycle, your symptoms, your most intimate health data — encrypted on your device and nowhere else. In a world where health data can be subpoenaed, Cara has nothing to hand over.
- Post-Dobbs demand — period tracker adoption grew 20% (Neiman et al., Contraception, 2025; PMID 39800282)
- 50M+ TAM — women of reproductive age in the US alone
SafeType Keyboard
Answers the question for your words. Every keystroke stays on your device. No network connection, no server, no record of what you typed. Your thoughts stay yours.
- Enterprise adoption — organizations already in pipeline
- Nonprofit program — free licenses via GiftedDreamers.org
- Zero network access — physically cannot transmit keystrokes
These aren’t products. They’re answers.
Every product we build started with a real moment — someone we love running into a system that wasn’t built to protect them.
The moment that built Cara.
Last year, Cassidy’s wife came to show him something on her phone — she was getting ads targeted to her menstrual cycle. Ads that knew things she hadn’t told anyone.
They realized her cycle tracking app was selling her most personal health data. “But it’s just a calendar!” Cassidy said. It wasn’t. It was a pipeline — from her body to an ad-tech server to the highest bidder.
After Dobbs, that pipeline became something far more dangerous. Law enforcement agencies began subpoenaing menstrual data from the very apps millions of women trusted. The “private” trackers had nothing private about them — just databases of intimate health data, sitting on servers, waiting.
So we built Cara. A period tracker with no server. No account. No cloud. Your cycle data lives on your phone and nowhere else. There is nothing to subpoena because there is nothing to hand over.
Privacy isn’t a setting in Cara. It’s the architecture.
The conversation that started Private Assistant.
Around the same time, Jamal’s mother-in-law came to us with a concern. Her employer was requiring her to train AI on every aspect of her position — a role she’d held for more than twenty years. She wasn’t just experienced. She was irreplaceable.
The company wanted her to hand over a lifetime of expertise so a model could learn to do what she does. For free. With no ownership. No credit. No control over how it would be used.
We had a different idea: what if she could build a tool around her expertise that she owned? An AI trained on her knowledge, running on her device, under her control — not her employer’s server.
That’s the vision behind Private Assistant. AI that runs entirely on your hardware. Your knowledge stays yours. Your data never leaves your device. No one else gets to profit from what you know.
Your expertise. Your device. Your AI.
Edge AI is the fastest-growing segment in tech.
Every major chipmaker is shipping neural processors. The hardware is deployed. The software opportunity is wide open.
“Our moat isn’t a feature — it’s a conflict of interest that our competitors can’t resolve. They make money from data. We make money from software.”
Engineers who believe technology should serve people.
A team united by a shared conviction that software should empower, not extract.
Jamal Porter
Cassidy Barton
Eric Basham, PhD
Heather Gorr, PhD
Kristine Socall, MBA
JC Gonzalez-Ramirez
Let's Talk.
Whether you're a potential partner, curious about our products, press, or just believe in what we're building — we'd love to hear from you.
Phone
(877) 674-8874For Investors
We are building the infrastructure layer for on-device AI — a category growing at 29% CAGR with the hardware already in 1.4 billion hands. If you want to discuss our roadmap, traction, or cap structure, reach us directly:
info@digitaldisconnections.comOr use the form and select Investor Inquiry.
Send Us a Message
Partnerships, press, product inquiries, or anything else — we read everything.