Only the topic survives until you reach the computer
You remember having a solution but lose the condition, sequence and decisive edge case.
Voice notes for vibe coding
The solution arrived away from the code —
keep the whole chain of logic
Describe the behavior, constraint and next technical step in normal words. When you open the editor, you will have a clear entry point into the code instead of a fragment like “fix auth.”
The engineering thought is waiting in text
The behavior, reason and first file or test are preserved for the next coding session.
A familiar moment
A good code solution often appears after you stop staring at the code
During a walk, the race condition finally makes sense. In the shower, a simpler state model appears. On the train, you remember the missing test case. If you save only “fix loading,” the problem name reaches the laptop but the solution does not. Voice preserves the causal chain while it is still whole.
Three simple steps
Say it now — sort it out later
-
1
Name the observable behavior
What happens now, under which condition and why is that behavior wrong?
-
2
Explain the implementation hypothesis
Preserve the constraint, likely change, edge cases and behavior that must not break.
-
3
Leave an entry point into the work
Add the first test, file or check you will open when you return to the computer.
Instead of one big pile
Each technical idea stays inside its feature
Keep the bug, solution, tests and documentation together. Product hypotheses and content ideas live elsewhere instead of mixing with implementation.
Offline upload: reproduce retry · inspect idempotency · add a connection-loss test
What changes
Remember less — finish more
Long explanations are welcome
A causal chain does not need to fit inside a short command.
You do not have to dictate source code
Capture the intent and constraints, then write the exact implementation in the editor.
Your original wording remains
Replay the recording when the transcript loses an important qualification.
The task gets technical context
The idea sits beside other actions for the same feature or fix.
AI helps only when requested
The beta can suggest structure, but architecture and verification remain the developer's job.
AI assistant · beta
Say it however it comes — AI can help make sense of it
AI can suggest a title, a due date and the right context, or help turn a larger idea into a clear plan. The beta is currently available to some Premium users. We are checking quality and will gradually open it to everyone.
The essential part already works without the beta: voice capture saves immediately and does not disappear when you are offline
A dictated implementation is a hypothesis, not a ready patch
- Check the idea against the code, tests and current contracts before changing the system.
- Do not speak secrets, tokens, personal data or details from private logs.
- If the thought describes a user problem, capture it as a product hypothesis before jumping into code.
What people usually ask
What is useful to dictate for a future code change?
Capture current behavior, the failure condition, likely cause, constraint, edge case and the first test that can verify the hypothesis.
Will Unload write the code for me?
No. This use case is about preserving engineering thought. Implement and verify the change in the project's working environment.
How is this different from a product idea?
A product idea describes a user, problem and value test. Here the intended outcome is a change in software behavior and its technical implementation.
Can I add the task without internet?
Yes. The recording stays on the phone and is sent when connectivity returns.
Should I dictate source-code snippets?
Usually it is more useful to name the file, function, condition and expected behavior. Exact code is safer to write and verify in the editor.
Try it with your next thought
Say the task, then get on with your day
Unload keeps it on your phone and brings it back when you need it. The app is free.
iPhone · Web · Telegram