What happens when you upload a video to lernaura.

The upload itself

You drop a video file into a lesson, an episode draft, a newsletter post, a community thread, or a coaching session record. It doesn't matter which surface — the pipeline is the same. The file lands on api.video EU, gets transcoded for adaptive streaming, and gets a permanent URL inside your content catalog.

So far this is what every other platform does. The interesting part starts now.

Step 1 — transcription

The video's audio goes to Gladia, our France-hosted speech-to-text provider. Within a few minutes you have a full transcript in the source language, with timestamps preserved at sentence level.

The transcript is editable. Most creators skim it for the one place the AI confused a brand name or a technical term, fix that line, and ship. Some creators don't open it at all and let it serve as-is — the accuracy is high enough that it usually works.

A contractual no-training clause with Gladia means none of this audio or transcript ever ends up training somebody else's model. The transcript stays yours.

Step 2 — translation

If you've configured target languages for your audience, the transcript gets translated automatically into each one. A creator with a German, French and Spanish audience uploads one English video and ends up with four transcripts attached to the same artefact.

You can override or edit any translation. For technical content where literal translation is risky, the override controls let you ship a human-verified version per language while the others run automatically.

Step 3 — multi-language subtitles

The translated transcripts become subtitle tracks attached to the video. Viewers in their browser see a language picker; mobile viewers see the same. No track management on your side — the platform pairs the right subtitle with the right viewer based on their account language preference.

Step 4 — local-currency pricing

If the content is paid — a lesson in a course, a paid newsletter post, a podcast subscriber-only episode — prices get set across every market you sell into. We use purchasing-power-parity adjustment from your reference price as the default. A €100 course in Germany might land at €60 in Brazil and ₹4,500 in India without you mapping a single market manually.

If PPP isn't right for your offer — say you're targeting a corporate buyer audience where local pricing would underprice you — the override controls let you fix specific markets to a flat global price. Defaults are sensible. Overrides are easy.

Step 5 — catalog entry and semantic search indexing

The video plus transcripts plus translations join your reusable content catalog. The text content is embedded into the semantic search index (Mistral, France, no training on your content). Within minutes of upload, a student can ask in natural language — "where did Sarah explain the difference between churn and contraction?" — and the search returns the exact 90-second clip with surrounding context.

The catalog matters more than the indexing. Every artefact you upload becomes a permanent, addressable, reusable element. Drop the same lesson into multiple courses without re-uploading. Reuse the slide deck from module 6 in a paid newsletter issue. Surface a community video answer inside a coaching session by linking to it. After a year you've built a searchable knowledge surface, not a folder of forgotten files.

This is what we mean by "never do anything twice." The upload is what you do. Everything else compounds.

Step 6 — the AI marketing layer sees the new content

If you've turned on the AI marketing intelligence layer, the new video becomes available for campaign drafting immediately. The AI now knows you have a new asset, can pull lines from the transcript for ad copy, can generate audiograms for podcast clips or social cards for video lessons, and can draft a launch campaign across your scheduled platforms.

You don't have to use it. The drafts wait in a queue for your approval. But if you do — which most creators end up doing for the parts of marketing they were previously neglecting (LinkedIn posts, X threads, Instagram clips) — the gap between upload and audience reaches collapses from days to minutes.

What it adds up to

The video that took twenty minutes to record now lands as:

• A streamable lesson in your course
• A full transcript searchable by your students in natural language
• Subtitles in every language your audience speaks
• Priced correctly in every market you sell into
• A permanent catalog entry you can reuse across surfaces
• Ready-to-schedule social promotion drafts across your channels

On a different platform that workflow takes a half-day and four tool subscriptions. On lernaura the upload is the half-day equivalent — minus the four subscriptions, minus the half-day.

This is the platform working for you while you do something else.

What we don't do automatically

In the spirit of being honest about what the platform doesn't try to do:

• We don't auto-publish anything. Translations, subtitles, marketing drafts, regional prices — all of it sits ready for you to review before it ships. The default is human-in-the-loop.
• We don't make decisions about your students. Search and recommendations surface content; they don't grade, certify, or gate access. (More on the Article 22 reasoning here.)
• We don't pretend the AI is your voice. The drafts are starting points. Most creators edit before sending; the system learns from what you edit and what you reject, and gets closer to your voice over time.

The AI does the boring work. The judgment work stays with you. That's the model.