From Document to Podcast, Quiz, and Briefing: How Docent's Transformers Work

Every firm has the same graveyard: the excellent 40-page report nobody read, the policy document that changed and nobody noticed, the methodology that lives in a deck one person made two years ago. The knowledge is there. It’s just trapped in a format that competes for time it will never win.

The problem isn’t that the knowledge is missing. It’s that people consume information in different ways, at different moments — and a dense PDF serves almost none of them.

The insight: one source, many formats

Document transformers start from a simple observation: the same knowledge is useful in very different shapes depending on who needs it and when.

• A partner heading into a meeting needs a one-page briefing.
• A new hire learning the methodology needs a study guide and a quiz to check they’ve got it.
• Someone commuting would absorb it as a podcast-style audio summary.
• A team reconstructing a project’s history needs a timeline.

A transformer takes one source — a report, a policy, a set of documents — and reshapes it into whichever of these fits, automatically. It’s the difference between summarising (same format, shorter) and transforming (a new format built for how the knowledge will actually be used).

The best format for a piece of knowledge is whatever gets it into someone’s head. For most documents, that format isn’t the document.

Grounded, so it stays true

The obvious risk with AI-generated content is drift — output that sounds right but wanders from the source. Docent’s transformers are built on the same principle as its search: they’re grounded in the original document, not generating freely from a model’s memory. That’s the RAG discipline applied to creation instead of question-answering.

And as with grounded answers, the source stays available. The posture is “trusted assistant, verifiable source”: the transformer does the tedious reshaping; the human keeps the original to check against. It’s assistance, not automation you have to blindly trust.

Why this changes what a knowledge base is for

Most knowledge systems are passive: a place to put documents. Transformers make the knowledge base active — a place that produces what people need from what you already have.

That’s what makes the onboarding story in From 8 Weeks to 2 practical. The firm doesn’t run a separate content-authoring project to build training; the training is generated from live documents, and updates when they do. A single methodology document becomes an onboarding path, a quiz, and a briefing — without anyone writing them by hand.

Inside Docent

In Docent, transformers sit on top of your indexed documents. Point them at a source — or a set of sources — and generate the format you need: podcast, slides, study guide, briefing, or timeline. Because the whole pipeline runs EU-native and GDPR-first (Where Your Data Lives Matters), even sensitive internal material can be reshaped without leaving European jurisdiction. The current status of our security and compliance posture is public on the Trust page.

The takeaway

A document only creates value when someone absorbs it. Transformers meet people where they are — audio, quiz, briefing, timeline — and turn a firm’s static archive into knowledge people actually use. Combined with grounded search and structured onboarding, it’s how the documents you already have stop being storage and start being leverage.

That closes this series. Follow Docent on LinkedIn — and see it for yourself with a demo.