Inspiration With the rise of LLMs, I wondered:
“Can I create a new programming language using only my imagination and prompts, without writing traditional code?”
That’s how Srinjan was born — a symbolic mini-language, built entirely using Bolt AI and prompt chaining, designed to process logic and simple instructions in a way that’s readable, teachable, and extendable
What it does
Srinjan lets users write logic in a custom format like:
start set mood = happy if mood is happy then speak "Joy!" end
It is:
A custom prompt-based language interpreter
Built fully using Bolt AI
Capable of handling user-defined variables, basic if/then logic, and print commands
Designed for education, creativity, and LLM prompting research
How we built it
No traditional code was used
🌐 All logic is built using Bolt’s prompt-based flows
🧱 I designed the syntax, rules, and handling instructions using my own chained prompts
⚡ Real-time response to Srinjan inputs using Bolt’s processing model
Challenges we ran into
Designing consistent, predictable prompts that behave like an interpreter
Managing variable storage, flow control, and string handling through plain text logic
Ensuring Srinjan felt like a real language despite being built only with natural language prompts
Accomplishments that we're proud of
Showcased the power of AI prompting as a full development medium
Created a live, working demo of Srinjan usable by anyone
Invented a unique way to teach logic and programming concepts without code
What we learned
We learned that prompt engineering can act as real programming when used creatively. Designing Srinjan taught us how to build a custom logic language using only Bolt AI and structured prompts — no code. We discovered the power of AI to interpret symbolic instructions, and how constraints can spark innovation in language design.
What's next for Srinjan Coding Language with Compiler
Build an proper output terminal Add support for loops, lists, and math
Create Bolt AI-powered Srinjan lessons for schools
Launch a Srinjan x LLM experiment: teaching LLMs with a symbolic interface
Built With
- ai
- bolt
- github
- json
- typescript
Log in or sign up for Devpost to join the conversation.