
Give your courses a voice – AI narration and presenters, powered by ElevenLabs
Video lessons can now narrate themselves. Write the script, pick a voice, and choose what learners see while it plays – a presenter, or a mark in your brand colour. Your first five generations are on us.
Recording narration is the part of a course everyone postpones. You need a quiet room, a decent microphone, and eleven takes. Then one slide changes next quarter and you do the whole thing again.
Today we are releasing AI narration for video lessons, powered by ElevenLabs. Write the script – or let Konstantly draft it from your slides – pick a voice, and generate. No recording session, and no re-recording when the content moves.
Write, Listen, Then Apply
Open a video lesson and click Generate with AI. The assistant reads the pages you have already built and writes a first-pass script. It is a draft, not a decision – edit it, or ignore it and write your own.
Under the script sits a line telling you how long it is and roughly how much audio that becomes. Then you pick a voice from your ElevenLabs library and press one button.
You hear the result before it touches the lesson. Applying it is a separate, deliberate click. A lesson is never overwritten by something you have not listened to.
What Learners See While It Plays
An audio-only lesson used to show learners nothing at all – a voice from somewhere, a static slide, no sign anything was even running.
Now narration can carry a presenter, and there are two kinds.
A mark in your brand colour. A waveform or a soft pulse, drawn from the narration's own amplitude. When the voice pauses, it goes still. It is not decoration pretending to be in sync; the motion is the audio.
Or a person. Twelve presenters, lip-synced to your narration. They are generated people – not actors, not stock photography, and not anyone real. That matters more than it sounds: a stock model release written before AI animation does not cover making that face say new words, and standard photography licences do not either. A person who does not exist has no likeness to license.
Pick either from the same grid. The mark costs nothing beyond the narration. A person is a video render, and the tile says so before you choose it.
Try It Free, Then Bring Your Own Key
Every workspace gets five narration generations on us. No key, no card, no setup – open a video lesson and generate. It is enough to narrate a short course and find out whether the result is something you would actually publish.
After that, you connect your own ElevenLabs key to keep going. That is a deliberate choice rather than a paywall, and it is worth explaining.
Text-to-speech is genuinely metered – every character of script costs real money at the vendor. The usual move is to hide that: bundle "AI minutes" into plan tiers, add a margin, cap you when the margin gets uncomfortable, then re-price once you depend on it.
We wired your ElevenLabs account instead.
Your rates. ElevenLabs bills you at whatever your plan there costs. We add no markup and take no cut – the same reasoning as our zero-commission Stripe integration.
Your voices. Once your key is connected, the list is your own library rather than a fixed menu of ours. If you have built a brand voice on ElevenLabs, your courses can speak in it.
Your ceiling. Your ElevenLabs plan is the only ceiling. There is no bundled quota of ours to run into, and nothing to re-price later.
On-screen presenters are different, and always need your own key. A person is a video render rather than speech – roughly 30 to 100 times the cost – and it needs an ElevenLabs plan of Pro or above. We will not put a render at that price on a shared account, so the twelve people unlock when your key does. The abstract marks stay free forever.
One thing runs the other way. Drafting the script costs you nothing extra: that step uses Konstantly's own AI, like the rest of the assistant. Only the voice itself touches your ElevenLabs account.
Guardrails Before Anything Is Spent
An author with a text box and someone else's API key is a budget incident waiting to happen, so the controls came first.
- The feature is off by default and switched on per account.
- Your admin holds the key and the limits: minutes per author per day, minutes per account per day, and a maximum script length: all in Settings → AI.
- The character count and duration estimate sit above the Generate button, which is the only control that spends anything.
- Failures say what actually happened. An ElevenLabs account out of credits and a plan that lacks a capability are different problems with different fixes, and the message tells you which one you have.
Getting Started
- Open any video lesson and click Generate with AI. Your first five generations need no setup at all.
- To keep going, get an ElevenLabs API key from your ElevenLabs account.
- In Konstantly, open Settings → AI → AI Presenter, paste the key, and set your daily limits. On-screen presenters appear once a key is connected.
AI narration is available on Business and Enterprise plans. If it is not switched on for your account yet, get in touch.
