Mail triage
Bring Gmail, Outlook, Yahoo, iCloud, or any custom IMAP/SMTP mailbox. Connect over IMAP with an app password — never your real password. The AI reads, summarises, drafts. Anthropic-grade triage, on your terms.
Privacy-first AI assistant
Your AI.
Your device.
Your data.
One app. Every model. Zero compromises. The intelligence runs on your phone — and stays there.
No account. No tracking. No servers in between.
7+Providers
100%On-device option
0Servers in between
∞Conversations, locally
On-device intelligence
The model lives in your pocket.
A quantized Llama 3.2 1B model ships separately on first launch — kept off the install size, kept on your device. From then on, it answers without a single network call.
Offline by default. By design.
Open the app on a flight, in a tunnel, in a country your provider doesn't reach — the local model is already there. No hand-off. No fallback to a server. The conversation simply happens.
- Quantized Llama 3.2 — fast on modern phones
- Delivered through Play Asset Delivery, not bundled at install
- Auto-deletes everything but your last 5 conversations
Network
Offline
Inference
On-device
Data leaving phone
0 B
Local Llama 3.2 · 1B · q4_K_M
When you want the cloud
Every model. One key. Direct.
Your API key talks to the provider. Nothing routes through us — there's nothing to route through. Bring whichever brain you trust today, swap it for another tomorrow.
OpenAIChatGPT · GPT-4o
AnthropicClaude
GoogleGemini
GroqLlama · Mixtral
MistralLarge · Codestral
DeepSeekR1 · V3
Qwen2.5 · Coder
CustomAny OpenAI-compatible endpoint
CoWork suite
Your AI gets things done.
Mail, notes, voice — all in the same app, all working off the same private context. Credentials stay on your device.
Notes & To-Do
Capture quickly, let the AI rewrite, structure, and translate. To-dos that understand priority. Notes that understand what you actually meant.
Transcribe
Voice in, structured text out. Meetings, voice memos, ideas at 3am — captured, cleaned, summarised. Audio never leaves the device.
Privacy architecture
There is no middle.
Most assistants route through their own servers — to bill, to log, to "improve". We don't have any. Your phone talks to your provider. That's the whole diagram.
Your phone
App + local model + keys
direct, encrypted
Provider you chose
OpenAI, Anthropic, Google…
×
No server in between
Nothing to log. Nothing to leak. Nothing to subpoena.
More than chat
A full assistant. Not a chat box.
Vision
Point at anything. ML Kit handles labels and OCR locally; vision-capable models do the deeper read.
Voice
Press to talk. The transcript hits the model — your voice never leaves the device.
PDFs
Drop in a paper, a contract, a manual. It gets parsed locally, then summarised, queried, drafted against.
Web search
Optional. When you need the world, the assistant pulls in only what's relevant — filtered, citation-aware.
Side by side
Your AI. Or theirs.
Most "AI apps" are a thin shell over someone else's cloud — your conversations live on their servers, indexed and retained at their discretion. DigitalOman runs on yours.
DigitalOman.aiOn your device
- Inference: on-device or your own API key
- Conversations: stored locally, auto-pruned
- Account: none required
- Network at rest: none
- Provider keys: stay on the device, never seen by us
- Offline: full app, full local model
- Telemetry: none on your conversations
Generic cloud chat appOn their servers
- Inference: their cloud, their pricing, their terms
- Conversations: retained server-side, sometimes indefinitely
- Account: required, often phone-verified
- Network at rest: always-on telemetry
- Provider keys: uploaded, decrypted server-side
- Offline: nothing works
- Telemetry: conversations may train future models
“Privacy isn't a Feature. It's the Foundation.”
Get the assistant
that doesn't get you.
Free. Offline-capable. No account. No tracking.