Backlog
15Under consideration
Native watchOS build for AI interactions on Apple Watch
Introduce a dedicated watchOS app for 120 AI Chat, enabling users to interact with the assistant directly from their Apple Watch. This lightweight, glanceable experience would focus on quick prompts, voice-to-text interactions, and notifications, ensuring the AI is always accessible without reaching for a phone or computer.
Native iOS build for 120 AI Chat app
Deliver a native iOS build of the 120 AI Chat app, optimized for iPhone and iPad devices, ensuring the same high-performance, 120 FPS experience users enjoy on desktop platforms. This would provide a seamless, touch-optimized interface with full support for system gestures, haptics, and accessibility features.
Voice-to-Text input for prompting with speech
Add Voice-to-Text (Speech-to-Text) functionality that allows users to speak their prompts instead of typing them. Microphone button to record voice and convert it to a text prompt Fast, accurate transcription with support for punctuation and natural phrasing Review before sending Multilingual support for global users
Cloud sync: real-time data synchronization across devices
Enable seamless, bidirectional synchronization of user data and settings across all devices in real time. Leverages our cloud‐sync infrastructure to persist conversation state, preferences, and drafts so users experience continuity whether on Windows, macOS, or Linux.
Native Linux build
Provide a fully packaged native Linux installer (.deb, .rpm, and AppImage) optimized for performance and theming. Ensures that Linux users benefit from the same responsiveness and customizable UI as on other platforms.
Retrieval Augmented Generation (RAG)
Integrate RAG workflows that fetch and cite external documents or knowledge bases in responses. Users can train and connect custom data sources for on‐device retrieval, preserving privacy and speed.
Model Context Protocol (MCP) Integration
Implement native support for the Model Context Protocol (MCP), a standard open protocol that enables the AI to connect with and utilize external data servers, local tools, and third-party APIs. Goal: Transform the application from a passive chat interface into an extensible agentic platform, allowing users to plug in any MCP-compliant tool (such as file system access, database connectors, or browser automation) to significantly expand the AI's real-world capabilities without custom integration logic.
Project Workspaces
Implement a "Projects" system that allows users to create dedicated workspaces, each configured with its own persistent custom instructions, uploaded knowledge files, and specific tool sets. Goal: Facilitate complex, multi-step workflows by scoping AI context to specific projects (e.g., "React App V2", "Marketing Campaign"), ensuring the model automatically accesses relevant background data and constraints without repetitive prompting or context window clutter.
Cross-Platform History Import (JSON files)
Develop a robust import engine capable of parsing and ingesting JSON data exports from major AI providers (OpenAI/ChatGPT, Anthropic/Claude, Google/Gemini) directly into the application's local database. Goal: Remove the barrier to entry for new users and eliminate vendor lock-in by enabling a seamless migration of existing conversation history, effectively turning 120 AI Chat into a centralized, searchable archive for all past and future AI interactions.
Long-Term Memory Management
Integrate a persistent memory layer that allows users to manually add, edit, and delete specific facts or instructions that the AI will reference across all conversation threads. Goal: Achieve true cross-session continuity and personalization, eliminating the need for users to constantly repeat their preferences, background details, or coding standards in every new chat interaction.
Next up
3Committed and queued
HTML/Web Preview for Code Blocks
Implement a native rendering engine within the chat interface that allows users to preview HTML, CSS, and SVG code blocks instantly without leaving the application. Goal: Reduce context switching (copy-paste to browser) and accelerate the UI iteration loop for developers using 120 AI Chat.
Adding conversation samples
This will help users quick start and understand 120 AI Chat easier with pre-filled samples in conversations
Video generation
Integrate OpenAI's Sora API into 120 AI Chat, enabling users to generate, preview, and remix high-definition videos directly within the chat interface. Goal: Expand the app's multimodal capabilities beyond text and image generation, offering a seamless workflow for video content creation without needing a separate browser or command-line tool.
In Progress
4Actively being built
Add search / favorites in model selection windows
Make favorites and search available in the model selection drop down menu, to speed up workflow.
Live chat with screen capture and contextual screenshot support
Add the ability to capture screenshots or screen recordings directly within the chat interface and send them to the AI model as contextual input. Key capabilities: Capture full screen Attach screenshots as part of a message AI processes visual content alongside the text prompt
Glass effect UI
A new glass-inspired interface with subtle transparency and blur, adding depth and a modern look
App themes
120 App themes Select theme for light mode Select theme for dark mode Convert 10 themes from Internet
Done
39Recently shipped
Multiple AI models within the same conversation thread
This feature allows users to compare outputs, blend strengths across models (e.g., GPT-4o for reasoning, Claude for writing, Mistral for speed), and conduct multi-perspective workflows. Inline model switcher for each message input Model displayed under each response for clarity Maintains full conversational context across model switches
Multi-threaded conversations for parallel contexts
Enable multi-threaded chats within a single conversation to allow users to branch off into sub-topics without losing context or cluttering the main dialogue.
Prompt library
Built-in Prompt Library Add custom prompts to library Select prompt inline the chat
Edit user's previous prompt
Enable users to edit their previous message or prompt in a conversation and regenerate the AI’s response based on the updated input. This feature brings more flexibility to iterative workflows, allowing users to refine questions, fix typos, or rephrase prompts while preserving the conversation flow.
Edit LLMs' message (model's message)
Allow users to edit the AI’s previous response directly within the conversation thread. This feature supports scenarios where users want to correct minor errors, annotate the output, translate, redact sensitive content, or personalize the response before sharing or exporting it.
Markdown text selection
Enable users to select and copy messages from the 120 AI Chat app.
Improve Markdown rendering
Enhance the Markdown rendering engine in 120 AI Chat to fully support advanced markdown elements like tables, multi-line code blocks, nested lists, and block-level formatting. Full table support with proper column alignment, borders, and scrollable overflow handling on small screens Syntax-highlighted code blocks with language detection and optional copy button Improved list rendering (support for nested bullet points and checklists with correct indentation) Better blockquote styling with visual distinction and nesting support
Image generation: DALL.E
Native support for AI-powered image generation, allowing users to generate visual content from text prompts directly in the conversation interface. This feature would support DALL·E and would integrate seamlessly into the chat workflow.
Chat configuration: Settings panel
Introduce an advanced settings interface allowing users to customize model parameters, default themes, and prompt templates per conversation. Settings are applied instantly without reloads, preserving state across sessions.
Attachment support: Inline file uploads
Allow users to attach images, PDFs, and other documents directly into the chat stream. Files are previewed inline and can be referenced by the AI for context-aware responses, all handled locally for privacy.