Talk to your documents
Load one PDF or a whole folder. NativeLab chunks, indexes, and feeds relevant passages into the model automatically — no manual copy-pasting. Ask questions naturally and it finds the right sections.
Multi-PDF mode summarises each document separately, then synthesises a final answer across all of them. Ideal for literature reviews, legal discovery, or comparing research papers.
PDF ingestion
Multi-document synthesis
Context-aware chunking
Visual Pipeline Builder
A full drag-and-drop canvas where you build AI workflows by connecting blocks. Drop in an Input, wire it through Reference or Knowledge context, pass it to a Model block, carry results through an Intermediate block, and collect everything in the Output.
Every connection is a smooth curve you draw yourself. Build loops that repeat automatically, watch each stage stream live in its own tab, and save any pipeline to reload with one click. Run saved pipelines directly from the chat window — no tab-switching needed.
Drag-and-drop canvas
7 block types
Save & reload
Loop support
Cloud API Models
Connect to OpenAI, Anthropic, Groq, Mistral, Together AI, OpenRouter, Ollama, or any custom endpoint from the new API Models tab. Pick a provider, paste your key, and click Test & Load. A 1-token check confirms it works before any real request is sent.
Once loaded, the cloud model behaves exactly like a local one — it appears in chat, works inside pipelines, and shows in the status bar. Keys are stored only on your machine. Save configs to reload instantly next session.
8 providers built-in
1-token verification
Custom endpoints
Saved configs
Reasoning into Code
Assign a reasoning model and a coding model. When you submit a coding task, the reasoning model thinks through intent and architecture first, then passes a concise brief to the coding model to generate the implementation.
You see the full chain of thought in the chat — nothing is hidden. Better code, less back-and-forth. Build this flow visually in the Pipeline Builder or let it trigger automatically from a single prompt.
Two-model chain
Visible reasoning trace
Auto-detects coding tasks
Reference Library
Build a persistent knowledge base for any chat session. Attach PDFs, notes, or text files as references — they're indexed once and retrieved semantically every time you ask something relevant.
The reference panel slides in smoothly from the right edge when you need it and stays out of the way when you don't. Think of it as giving your AI a bookshelf separate from the active conversation.
Semantic retrieval
Persistent index
Per-session libraries
Pause & Resume
Summarising a 200-page document? You don't have to sit and wait. NativeLab pauses mid-job, saves exactly which chunks have been processed and what's been generated, and lets you pick up precisely where you left off — even after restarting the app.
State serialization
Auto-pause threshold
Persistent job queue
Sessions & Exports
Every conversation is a session with its own history, title, and context window usage indicator. Export to JSON for programmatic use, Markdown for notes and documentation, or plain text for anything else.
Nothing is locked in. Your conversations are yours.
JSON · Markdown · TXT
Context usage meter
Sidebar history
