A privacy-first Android assistant that handles the basics, with limits
Inc reported on Aug 5, 2026, that Google will begin migrating users from Google Assistant to Gemini on Sept 4, 2026. That matters because the move shifts more conversational features into Google’s cloud and paid tiers, leaving privacy-minded users and people who don’t want another subscription asking: what now?
Meet Dicio. As the project’s GitHub puts it, “Dicio is a free and open-source voice assistant running on Android. It supports many different skills and input/output methods, and it provides both speech and graphical feedback to a question. It interprets user input and (when possible) generates user output entirely on-device, providing privacy by design.” ZDNET’s hands-on review tested Dicio from the Play Store and found it a useful, practical alternative for a clear subset of users, with important caveats.
Quick setup and what to expect
Installation is simple: install from Google Play, grant microphone permission, download the speech-to-text model when prompted during onboarding, and enable the wake-word model. The reviewer noted the wake word is pronounced roughly “dis‑ee‑oh.” Because Dicio emphasizes on-device processing, a model download during setup is expected.
What worked well in hands-on testing
- Timers and alarms, reported to set, cancel, and check reliably in normal use.
- Hands-free calling, initiating calls by voice behaved as expected in the reviewer’s tests.
- Media control, play/pause and track skipping worked; controlling Spotify was reported as reliable.
- Navigation, hands-free directions handed off to Google Maps cleanly.
- Weather, quick searches, and notification reading, simple queries returned speech and on-screen results; notifications could be read aloud when granted permission.
Those are the everyday tasks many people actually use assistants for: quick, utility-level interactions while cooking, walking, or driving. For those chores, Dicio behaves like a practical, privacy-forward assistant.
Where Dicio still runs into real limits
- Android Auto (high impact), ZDNET’s hands-on test found Dicio cannot be used within Android Auto because it isn’t recognized as a supported system assistant. That blocks in-car voice workflows for now.
- Messaging and email composition (high impact), voice-driven drafting and sending of messages or emails is not supported in the builds reviewed.
- Note-taking and audio memos (medium impact), there’s no reliable voice note workflow in the tested version.
- Calculations and edge bugs (medium), the reviewer encountered a calculation bug during testing. The review did not list frequency or exact expressions, but the issue was significant enough to mention.
- App-launch inconsistencies (low to medium), Dicio sometimes announces that it opened a non-default app (for example, Spotify), yet the app UI did not always appear. Media controls still worked even when the app failed to surface.
In short: Dicio covers the basics reliably, but it isn’t a drop-in replacement for users who depend on in-car integration, voice-first messaging, or advanced generative features that live in the cloud.
Settings, customization, and the developer angle
Dicio exposes many controls: language, theme, skills, input method, wake-word recognition method, an option to import a custom wake word, search output method, and listening behavior such as playing a sound when starting to listen, maximum silence thresholds, and whether to send speech-to-text results directly. These options let power users tune privacy and behavior.
The “import custom wake word” flow seems aimed at developers. The reviewer noted it appears to require command-line tooling like Python/pip, so non-technical users should treat custom wake-word imports as an advanced option.
Why platform rules and on-device trade-offs matter
Two ecosystem realities shape whether an open assistant can replace a platform native one:
- Platform certification and gatekeeping, features such as Android Auto require platform recognition or certification. That is a policy-level barrier rather than a purely technical one, which explains why a capable on-device assistant can still be blocked from in-car use.
- Cloud vs. on-device trade-offs, on-device processing reduces cloud exposure and improves privacy, but it usually means smaller models, narrower knowledge, and fewer generative features than cloud-based systems like Gemini.
The ZDNET reviewer also flagged the commercial pressure: Google’s push to fold conversational features into Gemini, plus paid tiers, makes an on-device, no-cost alternative attractive for people unwilling to trade privacy or add another subscription. The reviewer referenced Gemini AI Plus at $7.99/month.
Who should try Dicio, and how to get started
- Try Dicio if you prioritize local processing and privacy, and your assistant use is mostly timers, calls, music control, navigation, and quick lookups.
- Keep a fallback if you rely on voice messaging, voice email composition, Android Auto, or fully seamless app-launch behavior across every app.
Practical starter tips:
- Allow the speech-to-text model download during onboarding and test the wake word in a quiet room first.
- Keep your platform assistant configured as a fallback so Android Auto and other certified experiences remain available.
- Monitor battery and CPU use for a few days. On-device models can be heavier on older hardware.
- Track Dicio’s GitHub issue tracker for fixes. The project is actively developed and responsive to user reports.
- If you’re not comfortable with command-line tools, don’t expect to use the custom wake-word import out of the box. Treat that as a developer-only feature today.
Reproducibility note
ZDNET’s conclusions are based on a hands-on test of the Play Store build. The review does not list device model or Android/Dicio version details. Results will vary by device, Android build, and app version. If you test Dicio, validate your core routines (timers, calls, navigation, and music control) for 48 hours before making it your primary assistant.
Key takeaways, short questions, straight answers (as of Aug 2026)
- Can Dicio replace Google Assistant for most people?
For privacy-minded users whose needs are limited to timers, hands-free calls, media controls, navigation, and simple queries, Dicio can serve as a practical replacement. It doesn’t match Google’s cloud-based generative features or certified integrations like Android Auto.
- Does Dicio keep voice data private?
Dicio is designed to interpret input and, when possible, generate output on-device, reducing cloud exposure (per the project’s GitHub description). Some lookups or integrations may still use external services, check the project docs for exact details.
- Will it work with Android Auto?
No, Android Auto did not recognize Dicio as a supported system assistant in the hands-on test, so in-car voice workflows are blocked for now.
- Can it draft emails or send messages by voice?
Not in the reviewed build, voice‑driven message and email composition aren’t supported yet.
- Is installation and configuration hard?
The basic Play Store setup is straightforward: grant microphone access, download the speech-to-text model, and enable the wake-word. Advanced features like importing a custom wake word appear to require developer tooling.
If you try Dicio: keep your system assistant as a fallback, test timers and calls for at least 48 hours, and log any bugs on Dicio’s GitHub so maintainers can prioritize fixes.
Verdict: Dicio is a compelling, usable on-device assistant for a clearly defined audience, privacy-focused users who value hands-free basics over deep platform integrations. For everyone else, it’s a promising alternative that still needs a few missing pieces before it can replace the full Google Assistant experience everywhere.