Next-generation AI assistant

GPT Chat

Ask questions and get answers using your documents and knowledge.

Explain the difference between SQL JOIN and LEFT JOIN in simple terms
RAG: 3 sources found
JOIN returns only matching rows from both tables. LEFT JOIN returns all rows from the left table and matching rows from the right (NULL when there is no match)…
Show me an example with code
SELECT u.name, o.total
FROM users u
LEFT JOIN orders o ON u.id = o.user_id;
Markdown • Code • RAG sources

Everything you need for productive AI dialogue

Personal RAG

Upload documents (PDF, TXT, DOCX) and let AI use them in its answers. Hybrid search combines semantic vector search with BM25 full-text search and reranking.

AI agents

Create agents for specific tasks: a Python interviewer, code reviewer or SQL tutor. Each has its own prompt, emoji and RAG settings.

Multiple providers

Switch between OpenAI, Claude, DeepSeek and Ollama in the interface. Choose a model for the task — from fast to most powerful.

Markdown and code

Full Markdown in chat: headings, lists, tables and quotes. Code blocks with highlighting and a toolbar for quick formatting.

Session history

All conversations are saved in the sidebar. Quickly clear outdated sessions and manage your history without losing what matters.

Chat export

Save conversations as JSON or Markdown, or extract only the code. Convenient for documenting and reusing solutions.

Image generation

Create images directly in chat: enter a prompt and get the result without switching to other tools.

Fine-tuned RAG

Control the similarity threshold, enable or disable hybrid search and reranking directly in the chat interface.

Security and privacy

Your documents and agents are private. Publish agents only when you decide — visibility is under your control.

How it works

1
Upload your documents

PDF, TXT or Markdown — your knowledge base enters RAG and becomes available to AI.

2
Create an agent (or use the shared one)

Describe the task — AI will generate the system prompt and agent settings automatically.

3
Ask a question

AI will find relevant fragments from your documents and use them for an accurate answer.

4
Save the result

Export to Markdown or JSON, or copy the code — history stays in your sessions.

Supported providers
🟢 OpenAI GPT-5.4^ 🟢 OpenAI GPT-4.1 🟣 Claude Sonnet 5^ 🟣 Claude Opus 4.8 🔵 DeepSeek 🔴 Gemini 🦙 Ollama (optional)
RAG under the hood
  • Vector search (pgvector) — semantic similarity
  • BM25 full-text search (tsvector)
  • Hybrid Reciprocal Rank Fusion (RRF)
  • Reranking results
  • Adjustable similarity threshold directly in chat

Ready to start?

Create your first session, upload a document and ask something — the result will surprise you.