FireM3UAsk in your own words
Explore push-to-talk on supported devices for questions, daily plans, and follow-ups. Keep a remote or keyboard alternative so voice is a choice.
OUR SOCIAL PURPOSE · PRODUCT VISION
Your AI.
Heard as well
as seen.
Our goal is to help blind and low-vision people use their AI assistant independently at home. Ask a question, hear a useful answer, and control the experience without needing to read the TV.
See the accessibility milestonesThese are proposed features. The current development APK does not yet offer this voice-first AI experience. Device support will be confirmed through testing.
DESIGNING FOR INDEPENDENCE
From the first question.
To the last control.
A spoken answer is only one part of the experience. Setup, navigation, and privacy need the same attention. These are the priorities we propose for FireM3U.
Hear the whole answer
Read responses aloud and explain the important values and comparisons in AI-generated charts. Add repeat, pause, and speech-speed controls.
Know where you are
Announce menu options, the selected control, loading states, and errors. Make remote focus predictable and test compatibility with supported device screen readers.
Set up independently
Design spoken setup instructions and an accessible connection flow. Avoid image-only steps or a pairing code that can only be read on the TV.
Stay in control
Make microphone use explicit, give spoken confirmation when listening stops, and provide a clear way to pause speech or disconnect a bot. Explore private headphone output where supported.
Choose how to follow along
Keep text alongside speech. Explore larger type, strong contrast, and uncluttered layouts for people with low vision and households with different preferences.
BUILD WITH PEOPLE
Let lived experience
shape the product.
We propose a co-design pilot with blind and low-vision participants before a wider AI release. Participants would help choose useful tasks, review prototypes, and identify barriers. Recruitment and pilot details are not yet announced.
Our proposed release test is practical: can someone connect a bot, ask a question, understand the answer, recover from an error, and stop or disconnect without sighted assistance?
We would report task completion, barriers, and device limitations from that pilot before describing the experience as ready. Until then, this page sets out our intended direction.
EVERYDAY VALUE · PROPOSED USE CASES
Useful help.
On your terms.
Hear your day
Ask your connected bot to read a plan you provide, then repeat one item or explain it in more detail.
Understand a visual answer
Ask for the values, comparisons, and main takeaway from a chart in an AI response, without relying on its colors or layout.
Share a conversation
Explore an idea together, with spoken responses and optional on-screen text for others in the room.
A PRACTICAL FIRST USE CASE
Hear the answer.
Hear its source.
Our proposed Claude document companion would let you ask about a text document you choose and hear the supporting passage. The pilot would measure independent use, answer support, and control over the session.
Explore the document companion and more ideasAI YOU CAN HEARWritten example · Proposed experience
- You ask
- “Explain my budget without the chart.”
- Your assistant responds
- “In this example, your monthly budget is $3,000. Housing is $1,200, essentials are $900, savings are $600, and flexible spending is $300. Housing is the largest category.”
Independence belongs in the roadmap.
Follow the proposed steps from the current app to a voice-first AI experience.