Bolchai (বলছি — "I am speaking") is an AI agent that lives on your desktop. Ask it anything — it writes code, runs it, and gets things done.
import matplotlib.pyplot as plt
years = [2015, 2016, 2017, 2018, 2019, 2020, 2021, 2022, 2023, 2024]
gdp_growth = [6.6, 7.1, 7.3, 7.9, 8.2, 3.4, 6.9, 7.1, 5.8, 6.0]
plt.figure(figsize=(10, 5))
plt.plot(years, gdp_growth, marker='o', color='#e94560')
plt.title("Bangladesh GDP Growth Rate")
plt.ylabel("Growth %")
plt.grid(True, alpha=0.3)
plt.savefig("gdp_growth.png")
print("Chart saved!")Describe what you want in plain language. Bolchai handles the rest.
Analyze spreadsheets, create charts, and process CSV files. Just describe the analysis you need.
Organize files, batch rename, convert formats, and automate repetitive file operations.
Runs Python and PowerShell code natively on your machine. No sandboxes, no limits.
Install software, configure settings, manage processes — all through conversation.
Extract data from websites, download files, and automate web tasks with natural language.
Works with any LLM provider — OpenAI, Anthropic, Fireworks, local models via Ollama. You own your API key.
Download and run the installer. No Python or dependencies needed — everything is bundled.
Enter your API key in Settings. Pick any LLM model supported by LiteLLM.
Type what you want done. Bolchai generates code, asks for your approval, then executes it.
Yes. Bolchai is free and open source under the MIT license. You need your own LLM API key, which may have costs depending on the provider.
Any provider supported by LiteLLM — OpenAI, Anthropic, Google, Fireworks AI, Together AI, Groq, Ollama (local), and many more. Just enter the model name and API key.
Bolchai asks for your permission before executing any code. You see exactly what will run and can approve or skip each execution. You're always in control.
If you use a local model through Ollama, yes. Otherwise, it needs internet to reach the LLM API.
বলছি (Bolchai) means "I am speaking" in Bangla. It's your computer, speaking back to you.
Currently Windows only. Mac and Linux builds are planned.