Hello Codeforces.

Recently, I have been building a project called SMS AI. It is a simple but fast multimodal workspace to chat, generate code snippets, and parse documents.

I noticed that a lot of general web tools have quite high latency when streaming code blocks or rendering formatting properly. I wanted to see if I could optimize the processing structure to bring the time-to-first-token down under 100 milliseconds. The current implementation is live and can be tested here: SMS AI.

Core implementation details

The platform is lightweight and split into a few basic utilities that might be helpful if you are working on problems or analyzing long reference texts:

• Syntax Highlighting & Code Generation: It outputs and formats raw scripts in standard languages (C++, Python, etc.) with clean syntax rendering.

• Context Memory: The conversation uses a clean markdown rendering engine that maintains state variables and long-term memory across chats.

• Document Analysis: You can upload .pdf, .docx, or .txt templates to extract insights or query specific data points within large text inputs.

• Web Retrieval: It includes a simple web-aware search layer to pull and cite references directly from live web data. Benchmarks & Architecture

I focused mostly on response speed and edge execution: • Time-to-first-token: < 100ms. • Uptime: Currently stable at 99.99%.

There is a basic free layer available immediately (10 messages per day, no registration required to test). For those interested in heavy processing or API usage, higher tiers exist to cover infrastructure scaling costs.

The system was fully written and deployed by M_Sameer in Pakistan. If you find any edge cases, latency drops, or parsing bugs, please leave a comment below or mention them. I will try my best to optimize them.

The platform has currently compiled around 10M+ message strings across various standard testing suites and supports multi-language query layouts natively.

Thanks for reading.