Skip to content

Proprietary tool · Video presentations

Slides: a presentation that narrates itself

The tool turns a presentation into a video. Each slide element appears as the voiceover speaks about it. The voice comes from ElevenLabs, and the language layer runs through OpenRouter.

Technical layers
ElevenLabs and OpenRouter
Access
slides.hexart.pl, after logging in
Application
Training and materials for the client

01Challenge

The recording is the second round of work

Training, client materials, or a process guide for a new team member: the content is created as slides, then someone talks through them. Recording means doing the work twice.

A voice-over artist reads the script while an editor assembles the track and matches each cue to the sentence it belongs to. This takes the longest and is the most fragile part: changing one sentence shifts everything that follows.

02Solution

Narration drives the animation

Slides handles this part automatically. The app is available at slides.hexart.pl after logging in. Description: “Create animated slide videos with voice-synced reveals. Powered by ElevenLabs and OpenRouter.”

Text is routed through OpenRouter, a single interface for multiple models. The voice comes from ElevenLabs. A slide element appears when the narrator says the sentence it relates to.

  • The reveal follows the narrative

    No manually set keyframes. Each point appears based on what was said and when.

  • Text correction instead of editing

    Changing a paragraph in the script does not require going through the entire timeline or starting a new recording session.

  • Interchangeable model layer

    With OpenRouter, model selection is a setting, not an assumption built into the architecture.

We do not specify length limits, pricing, or a list of languages here. We run the tool ourselves and define the scope for each assignment.

03Application

Where does this work

For content that is read aloud anyway, when recording and editing take longer than preparing the content: internal training, client materials, process documentation.

It delivers the most value for series of short videos with the same structure, because updates only require replacing a sentence in the script. The difference is smaller for a one-off video.

04Control

Listen and label before publishing

The tool shortens editing time, but does not replace the author. The person preparing the material decides on the script and the order of arguments. We listen to the narration before assembling the film and identify sections that need to be recorded again.

A film with a machine-generated voice must be labelled, as required by Article 50 of Regulation (EU) 2024/1689. We agree on how to label it with the client before the material reaches its audience. Our policy is explained on the AI use page.

Voice alone, including voice cloning and speech synthesis, is a separate service: voice cloning and TTS voiceover.

A presentation that needs to be turned into a film?

Send us an outline or finished slides. We’ll divide the material into what Slides can assemble automatically and what needs standard editing.

Schedule a workshop with Paul

Choose thirty minutes or a full workshop. Book a time slot in the calendar, receive immediate confirmation.