Skip to main content
We build Noesis in public and share our development timeline openly so you always know what’s shipped, what’s actively being worked on, and what’s coming next. This page reflects the current state of the product across three phases — Released, In Development, and Planned — and is updated as priorities shift. If you have a feature suggestion or a workflow you’d love to see supported, we want to hear from you.

Released

These features are live and available to all users on qualifying plans today.

Thread Branching & Forking

Fork any conversation at any message with a single click, exploring alternative ideas side-by-side without losing your original context. Noesis maintains a complete visual history tree of every path you’ve taken, and you can switch context seamlessly between branches at any time.
  • Single-click branching from any chat message
  • Visual history tree of all conversation paths
  • Seamless context switching between branches

2D Neural Mind-Map Visualization

Auto-generate an interactive 2D node canvas from your conversation graph as it grows. Your thoughts and AI branches appear as connected nodes in real time, laid out automatically using intelligent layout algorithms. Double-click any node to jump directly back to that point in your conversation history.
  • Real-time interactive node canvas
  • Auto-layout algorithms for complex trees
  • Double-click nodes to jump back to specific chat history

Multi-Node Knowledge Synthesis

Select two or more conversation branches and synthesise them into a unified summary that resolves contradictions and connects key concepts. A built-in source switcher lets you paginate between the original nodes and shift focus back to the source concepts that informed the synthesis.
  • Select 2 or more nodes for synthesis
  • Synthesised source switcher and pagination
  • Focal shift back to original source concepts

Multi-Model Toggle Engine

Switch between the world’s leading large language models mid-conversation to leverage the unique strengths of different architectures. Compare outputs from multiple models on the same branch, or assign different models to different branches to see how each one reasons through the same problem.
  • Gemini 3.5 Flash, Claude Haiku 4.5, GPT 5.4 Mini, Grok 4.3, Claude Opus 4.8, GPT 5.5, and Gemini 3.1 Pro support
  • Compare model outputs on the same branch
  • Optimised token consumption mapping

Custom System Prompts & Presets

Define custom personas, writing guidelines, and logic rules that shape how the AI handles new conversation branches. Save and reuse prompts across your Spaces, or apply per-branch overrides when a specific thread needs a different voice or set of constraints. Curated presets for coding, writing, research, and analysis are available out of the box.
  • System prompt editor with dynamic variable insertion
  • Curated presets for coding, writing, research, and analysis
  • Per-branch prompt overriding

Real-Time Collaborative Spaces

Invite teammates into your conversation tree for live, cursor-synced collaboration. Co-explore branches, review logs together, and co-write summary documents on a shared spatial canvas — complete with audio and video bubbles for in-context communication. Granular permission levels give you precise control over who can view, comment, or edit.
  • Real-time cursors and collaborative tree branching
  • Multiplayer spatial canvas with audio/video bubbles
  • Granular permission levels (viewer, commenter, editor)

High-Fidelity Document Export

Export your research branches and discussion threads into beautifully formatted output files for sharing or archiving. Choose from interactive Markdown that preserves thread hierarchy, SVG vector graphics for presentations, or structured JSON for programmatic use. Direct integrations with Notion and Obsidian are also available.
  • Interactive Markdown export preserving thread hierarchy
  • SVG vector mind map downloads for presentations
  • Direct export integration to Notion and Obsidian

Core Document Ingestion

Upload academic PDFs, Word documents, or plain text files and use them as grounded context within your conversations. Noesis parses and chunks the source material so the AI can reference specific sections during branching discussions — keeping each document’s context cleanly isolated so it doesn’t bleed across unrelated threads.
  • Support for PDFs, DOCX, and TXT files
  • Fast local document parsing and chunking
  • Context-isolated chat referencing uploaded files

In Development

These features are actively being built and are targeted for release in Q3 2026.

Citation Mapping & Web URLs

Ingest any public web URL with a single click and have Noesis automatically map visual citation links that connect your generated chat nodes directly to the specific source paragraphs that informed them. An interactive source-document viewer embedded in the canvas will let you read the original material without ever leaving your workspace.Target: Q3 2026
  • One-click URL and website ingestion
  • Visual citation links mapping nodes to sources
  • Interactive source-document viewer on canvas

Autonomous Multi-Agent Research Loops

Dispatch autonomous agents that scour the web, synthesise complex topics, and cross-examine competing claims on your behalf. Each agent can spawn sub-tasks to recursively verify details, and a self-correcting web search layer catches contradictions before they surface. Findings are automatically laid out as structured mind maps — with pros, cons, and competing hypotheses clearly separated.Target: Q3 2026
  • Recursive deep-dives: agents spawn sub-tasks to verify details
  • Self-correcting web search with source verification
  • Automatic layout of pros, cons, and competing hypotheses

Planned

These features are on the roadmap for Q4 2026 and beyond.

Local Execution & Ollama Support

Run your Noesis conversations entirely on your own machine using locally hosted models via Ollama — ideal for highly confidential research, air-gapped environments, or simply keeping costs down. You’ll be able to use Llama 3, Mistral, and Phi-3 profile configurations, and a hybrid processing mode will let you use local models for quick draft nodes while routing deeper research queries to cloud models.Target: Q4 2026
  • One-click connection to local Ollama server instances
  • Llama 3, Mistral, and Phi-3 model profile configurations
  • Hybrid processing: local models for draft nodes, cloud for deep research

API Access & Developer Integrations

Access your conversation graphs, Spaces, and synthesis outputs programmatically through a REST API secured by API key management. Webhooks will let you subscribe to node creation and synthesis events in real time, and official SDKs for TypeScript and Python will make it straightforward to build custom plugins, CLI tools, or sync agents that connect directly with your Noesis workspace.Target: Q4 2026
  • REST API endpoints with API key management
  • Webhooks for node creation and synthesis events
  • Developer SDKs (TypeScript and Python)
Have a feature suggestion? Email us at support@thinknoesis.xyz or join our Discord to share your ideas. We build Noesis in alignment with our community.