The job: turn real work into a recordable interview
Many people have useful material but no clean way to talk about it on camera. The source may be a product decision, a research finding, a project update, a draft essay, customer evidence, or a set of notes that has not yet become a public argument.
REC turns that material into a research-guided solo video interview. It is built for founders, operators, creators, researchers, and subject-matter experts who would rather answer a good question than perform a generated monologue.
The product's role is preparation and structure. The speaker's role is authorship: choosing the examples, making the claims, showing uncertainty, and deciding what is worth sharing.
Before recording: source research and tailored questions
A session starts with a name, a conversation profile, and any link or notes the speaker wants REC to examine. Better source material usually produces more specific questions, but a polished brief is not required.
REC researches relevant public sources and prepares a question path that typically contains 12–15 prompts. The questions are designed to surface evidence, examples, tradeoffs, frameworks, disagreements, and details that are missing from the public record.
Research improves the starting point; it does not guarantee factual correctness. REC is designed to turn unknowns into questions instead of treating them as facts, and the speaker still needs to check claims before publishing them.
During recording: one prompt at a time
REC is a guided solo recording session, not a live or adaptive AI interviewer. The question path is prepared before the recording begins, then presented one prompt at a time with an optional answer anchor nearby.
The speaker answers in their own words. There is no generated answer script to read, no synthetic avatar, and no attempt to imitate the speaker's judgment. Hesitation, conviction, examples, and caveats remain part of the source material.
The recording itself is the primary output. It preserves the full answer behind any shorter moment that may later be selected as a clip.
After recording: review first, render second
When compatible live browser transcription is available, REC can use those transcript segments to recommend timestamped moments. The selector aims for several concise highlights and explains why each candidate may be useful.
If live transcription is unavailable or empty, automatic highlight selection is skipped. The full recording still exists, but REC does not currently export a separate transcript or pretend that automatic recommendations were produced.
For a recommended moment, the user can choose a clip style, render the selected section in the browser, download it, edit or copy a suggested caption, and open the browser share sheet where supported. REC does not schedule or publish to social channels.
What source-backed means here
A source-backed REC clip can be traced to a real recorded answer and its position in the session. The label is about provenance: a viewer or editor can return to the person and the fuller context behind the excerpt.
Source-backed does not mean automatically true, unbiased, or endorsed by the sources used during research. A speaker can still misremember, overstate, or change their mind. Review remains part of the workflow.
That boundary is the value proposition in one line: REC uses AI to prepare and inspect the material while keeping the human being responsible for the meaning.