What should I import for Verbasil to be useful?

Verbasil invents nothing. It reads what you give it, extracts what looks like a decision, a commitment, a recurrence or a learning, and submits it to you. If the material contains none of that, none will come out.

This is by far the first sticking point: someone uploads two slide decks, sees nothing appear, and concludes the tool does not work. The engine worked. The material was silent.

What produces material

1. Recent meeting notes. These are the densest in decisions: someone settles a question, someone commits, a date lands.
2. Exchanges with customers or prospects. Emails, feedback, call notes. This is where the same requests repeat, and repetition is what Verbasil is built to see.
3. Your own notes. Whatever you wrote after a judgment call, however rough, even three lines.
4. What failed. An abandoned test, a lead that went nowhere. That is learning material, and nobody ever files it.

What produces nothing

A reference document, a sales deck, a brochure, a standard contract. These are worth keeping and worth being able to look up, but they contain no judgment calls: nobody decides anything in them. Uploading only these gives you a memory that is searchable and silent.

How much, and over what period

Three to five different sources, from the past six months. Variety matters more than volume: two meetings and one customer thread give you more than ten documents of the same kind, because recurrences only appear when sources cross.

Didn't work?

• You imported, and nothing appears. Check validation first: nothing enters the memory without your approval, and an unvalidated source has produced nothing yet. See Why my decision is missing.
• Your file was rejected. The format may not be supported. See Which file formats are supported.
• You only have reference documents. Start by writing a ten-line quick note on the last three judgment calls you made. It is the highest-yield material at the start, and it costs five minutes.

FAQ

Should I import everything at once?

No, and it is actively counterproductive. Three to five well-chosen sources give a readable result you can review and correct. A bulk upload produces a validation queue nobody reads, and a memory nobody has reviewed does not deserve trust.

Are old documents worth anything?

Yes, but later. A decision from two years ago is context, not an action. Start with the past six months, where decisions are still live, then work backwards once the mechanics are clear.

Can Verbasil work with badly written notes?

Yes. Raw notes, in shorthand, with typos, are still usable material. Content is what matters, not form.