01
Language setup
The learner sets their native language, target language and level, shaping every scene generated after.
AI Products
Zazu Beak AI Learning Platform
Multi-language learning app that generates real-world scenario stories with target-language audio, target-language text and phonetic text side by side.
Context
What needed to be made clearer
A Framework7 app built for learners who pick up language through phonetics and real-world scenarios rather than static lesson trees. The learner sets a native language, a target language and a scenario, and the platform generates a playable scene with narrated audio and matching text.
What was built
The useful version came from a few deliberate parts
The work was scoped around what needed to move, connect or become easier to understand.
- Framework7 app build
- Multi-language story generation pipeline
- Phonetic and audio scene production
- Language and scenario configuration
- LLM story generation
- Phonetic transcription
- ElevenLabs voice generation
- Inngest event orchestration
Production flow
One prompt becomes a playable scene
The work is easier to scan as a chain of generation steps rather than one large AI pipeline.
02
Scenario prompting
A prompt describes the real-world scenario the learner wants to practise, from ordering coffee to a job interview.
03
Story generation
An LLM turns the scenario into a short dialogue-driven story matched to the learner's level.
04
Phonetic and text output
Each line is rendered as target-language text alongside a phonetic version, so learners can read while they listen.
05
Voice generation
ElevenLabs narrates every line in the target language, one audio track per scene.
06
Scene delivery
Inngest coordinates each generation step in the background before the finished scenes are ready to play in the app.
Key features
What it does
01
Learner-configured native and target language pairs
02
Real-world scenario story generation
03
Target-language text and phonetic text shown side by side
04
ElevenLabs narration per scene
05
Inngest-orchestrated generation pipeline
Stack / tools involved
The pieces behind it
The important part is not the tool list. It is what each piece is responsible for in the workflow.
01
App and configuration
Framework7 powered the app shell. Learners configure their native language, target language and a real-world scenario before generation starts.
02
Generation services
An LLM generated the scenario's story and phonetic transcription while ElevenLabs narrated each line in the target language.
03
Orchestration and delivery
Inngest coordinated each background generation step on Vercel, so scenes were ready to play in the app once every piece had finished.
Useful for
Where this pattern fits
Product builds needing personalised, multi-language conversational content for phonetics-first or real-world-scenario learners rather than fixed lesson trees.
Primary capability
- Multi-language AI content generation
- Phonetic learning support
- Scenario-based practice
If this looks close to your own workflow, bring the messy version and we will map the first useful build.
Book a CallTalk through a similar project
Book a time to talk through what you're trying to build.