What We're Building
A transparent look at what has shipped, what's actively in development, and what's coming next. We build in the open.
Askimo started as a command-line tool, a focused utility for automating AI workflows from the terminal. It was fast and scriptable, but it couldn't cover the interactive, exploratory side of working with AI. That gap led to the Askimo desktop app in late 2025: a native, privacy-first client that lets you switch between OpenAI, Claude, Gemini, and Ollama without losing your context or your data. Since then, each release has layered on what users actually needed: project-based RAG to bring in your own documents, a built-in terminal, MCP tool integrations, and a script runner. The roadmap below shows where that journey has taken us, and where it's heading next.
Read the full storyv1.0: Desktop App Launch
Shipped November 2025First public release of the native desktop app.
Native desktop app (macOS, Windows, Linux)
CoreCross-platform Kotlin/Compose application.
Multi-provider support
ProvidersOpenAI, Claude, Gemini, Ollama in one place.
Persistent local chat history
Full-text search across all conversations.
Custom directives
Save system prompts to shape AI behavior.
Theme & font customization
Light/dark themes and font size controls.
Keyboard shortcuts
Fast navigation without touching the mouse.
Export conversations
Export chat history to Markdown.
Folder & star organization
Group and favourite important chats.
v1.2: Project RAG & File Attachments
Shipped December 2025Answers grounded in your own documents.
Project-based RAG
RAGIndex your documents and code for context-aware answers.
jvector + Apache Lucene search
RAGHybrid full-text and semantic similarity search.
File attachments in chat
Attach PDFs, spreadsheets, images, or code directly.
Docker AI support
ProvidersContainerised models alongside OpenAI, Claude, Ollama.
Project sessions
Organised knowledge-based chat sessions per project.
Responsive design
Adapts across all screen sizes.
Notarized macOS installer
Signed & notarized for secure installation.
v1.2.x: Terminal, MCP & Image Support
Shipped February 2026Built-in terminal, MCP tools, and multimodal inputs.
Built-in terminal emulator
TerminalRun commands without leaving the app.
MCP integration (stdio)
MCPConnect MCP servers for tool-enabled AI workflows.
RAG sources tree
RAGInspect exactly which documents the AI used.
Image model support
Multimodal inputs for vision-capable models.
Grok provider
ProvidersxAI Grok added alongside existing providers.
Performance & UX improvements
Faster rendering, smoother scrolling, better reliability.
v1.2.2x: Script Runner
Shipped March 2026Run AI-generated scripts directly inside the app.
Script Runner (Python, Bash, Node.js)
AutomationExecute AI-generated code without copy-pasting to a terminal.
Auto-install Python packages
AutomationRequired packages are installed automatically before running.
Inline script output
See real stdout output so AI can iterate and fix errors.
Next Release: Agent & Workflow Enhancements
In Progress Q2 2026Deeper agentic capabilities and smoother everyday UX.
Expanded MCP transport support
MCPSSE and HTTP transports in addition to stdio.
Multi-step agentic workflows
AgentLet AI plan and execute multi-step tasks autonomously.
Improved RAG chunking & re-ranking
RAGBetter retrieval quality for large knowledge bases.
UX polish & accessibility
Keyboard accessibility, focus management, and polish.
Planned: Team Edition
Planned Team Edition H2 2026A professional edition built for teams. These features will not be part of the open source release.
These features are planned exclusively for the Team Edition — a professional paid edition for teams. They will not be included in the open source release.
Shared knowledge bases
RAGTeams share a common RAG knowledge base, one source of truth for everyone.
Shared prompt & directive library
Publish and sync directive packs across the whole team.
Role-based access control
Control who can read, edit, or manage shared knowledge and configurations.
Audit & activity log
Track AI usage and knowledge changes across team members.
Centralised provider configuration
ProvidersAdmins set API keys and model defaults once for the entire team.
Cloud sync (opt-in)
Encrypted sync of knowledge bases and settings across team members. Privacy-first, never stored in plaintext.
Considering: Community Ideas
ConsideringFeatures we're evaluating based on community feedback.
Mobile companion app
Access your local Askimo instance from a mobile device.
Plugin system for RAG sources
RAGLet developers write custom plugins to pull knowledge from any source: databases, APIs, internal tools, and more.
Have an Idea or Request?
Open a GitHub issue to suggest a feature, report a bug, or vote on what matters most to you. Every release is shaped by real user feedback.
Free & open source · Community-driven · Your feedback shapes what ships next