Microphone is one product: a voice agent. It generates ads from a spoken description of your idea to build conviction, then deploys social media subagents to seed channels autonomously. The subagents are a handoff, not a second product — the conversation remains the source of every claim they repeat, and the hardware they run on is yours.
Key takeaways
- The conversation is the source. Problem, hypothesis, audience, and creative all derive from it.
- Subagents execute; they do not invent the company.
- Seeding follows validation. A claim gets amplified only after it survives a cold audience.
- The social work runs on a Raspberry Pi you control, not a cloud social product.
Microphone is a voice agent that learns from paid ads, then puts that conviction to work seeding organic channels through social media subagents. That sentence contains a design decision worth making explicit, because most tools in this space are two products stapled together and pretending otherwise.
Why that split matters
The conversation is the source. Problem, hypothesis, audience, creative — they all come from how you say the idea, including the parts you say sideways when you are being honest about what does not work yet. Nothing downstream can be more specific than that recording.
The subagents do not invent the company. They take a claim that already survived an ad test and put it where people are already talking. That ordering is the whole architecture: judgment upstream, repetition downstream, and no pretending the second can substitute for the first.
An agent that decides what to say and an agent that says it a hundred times are not the same kind of thing. Confusing them is how automation becomes spam.
Where the subagents live
On a Raspberry Pi you control. Comments, posts, and likes stay on hardware in your house. That is not a cloud social product with your credentials in a multi-tenant browser farm. It is work the voice agent can hand off after the test, running somewhere you can physically unplug.
What that buys you
- Your logged-in sessions stay on a machine you own.
- The cadence continues whether or not you are paying attention that week.
- Cost stays flat instead of scaling with rented browser instances.
- The boundary between deciding and executing stays visible.
One product, two jobs
Talk. Test. Then seed what survived. The voice agent covers the first two and decides what enters the third. Plus is the validation loop; social media subagents and routines arrive on Pro and Ultra, at the point where you have something worth repeating.
Microphone is a voice agent that learns from paid ads, then puts that conviction to work seeding organic channels through social media subagents. The agent that listens to you and the agents that go out into the world are the same product, in that order. Talk the idea through, run the test, and let the response tell you whether the next month is worth spending.
Related reading: how to run social media from a Raspberry Pi, how to find your first audience on social, your social agent should live in your house. Each one covers a different part of the same loop — say the idea out loud, put it in front of strangers, and let the result decide what you build next.