shadow-planner

Product Introduction

1. Definition: Shadow-planner is a desktop-based, AI-assisted project planning and Gantt chart software application. It is a technical tool for project management that runs natively on Windows, macOS, and Linux operating systems, utilizing a local SQLite database for data storage.
2. Core Value Proposition: It exists to automate the initial drafting and ongoing management of complex project plans through a conversational AI interface, while prioritizing data privacy and ownership by operating primarily offline with optional cloud AI integration. Its primary keywords are: AI project planner, local Gantt software, desktop project management, and offline project planning.

Main Features

1. AI-Powered Plan Drafting: The software uses a Large Language Model (LLM) agent to transform a natural language project brief into a structured project plan complete with epics, tasks, assigned resources, and estimated timelines. How it works: The user pastes a Markdown brief; the AI parses goals, constraints, and streams to generate a draft work breakdown structure (WBS). This draft remains in a review state until the user explicitly "materializes" it, ensuring no automatic changes are made to the live plan.
2. Chat-Controlled Project Management: The integrated chat interface allows users to query and modify the project plan using natural language commands. How it works: An AI agent with tool-calling capabilities has read/write access to the active project scenario. Users can ask analytical questions ("Who is overbooked?") or issue change commands ("Push launch by two weeks"). The agent executes these by calling internal functions (like resource allocation calculators or schedule dependency engines) and always requests user approval for destructive changes, presenting a diff preview.
3. Multi-Scenario Planning with Local-First AI: Users can create parallel "scenarios" (e.g., Baseline, Accelerated) for what-if analysis. The AI can draft unique plans for each. Critically, the AI agent is provider-agnostic. It defaults to running on a local LLM (like qwen2.5:7b) via Ollama, ensuring complete offline functionality and data privacy. Alternatively, users can optionally connect cloud providers (Anthropic Claude, OpenAI, OpenRouter) via API keys, with all AI requests scoped strictly to the active scenario.
4. Capacity-Aware Resource Utilization View: This is a dynamic table that calculates and visualizes planned versus actual person-days per resource per week. How it works: The software uses resource capacity settings (e.g., weekly available days) and project allocations to compute utilization percentages. It visually highlights over-allocations (e.g., over 110%) and allows drilling down to the task level. This feature prevents overallocation before it occurs by providing forecast data.
5. Interactive Gantt Chart with Dependency Management: A fully interactive timeline view that respects task dependencies. How it works: Users can drag-and-drop tasks to reschedule, and the software previews the cascading impact on dependent tasks before applying changes. It supports a hierarchy (Project > Epic > Task) and allows zooming from weekly to yearly views. All changes are recorded in an audit log and are undoable.

Problems Solved

1. Pain Point: The time-consuming and manual process of translating a project concept into a structured, resourced, and timed Gantt chart. Shadow-planner automates this initial "blank page" phase using AI.
2. Pain Point: The difficulty of performing dynamic what-if analysis and re-planning without manually adjusting hundreds of task dependencies and resource assignments. The chat interface allows rapid, conversational exploration of alternatives.
3. Target Audience: Solo Project Managers & Consultants who need professional-grade planning without enterprise software costs. Technical Leads & Engineering Managers in small to mid-size teams who require capacity planning and scenario modeling. Privacy-Conscious Organizations in regulated industries (legal, finance, R&D) that cannot store sensitive roadmap data in the cloud.
4. Use Cases: Rapid Project Proposal Drafting: Turning an RFP or product brief into a detailed plan in minutes. Agile Re-planning: Using chat to instantly assess the impact of a scope change or delay. Resource Conflict Resolution: Identifying overbooked team members weeks in advance and using AI to suggest rebalancing options. Portfolio Comparison: Running baseline and accelerated scenarios side-by-side to present options to stakeholders.

Unique Advantages

1. Differentiation: Unlike cloud-based SaaS tools (Asana, Monday) with per-user subscriptions, shadow-planner is a one-time purchase desktop application. Unlike traditional desktop software like Microsoft Project, it integrates a conversational AI assistant natively, rather than as a separate Copilot add-on with ongoing fees. Unlike pure Kanban tools (Trello), it offers a true, dependency-aware Gantt chart and resource utilization analytics.
2. Key Innovation: The local-first, switchable AI architecture. The core innovation is decoupling the planning intelligence from a mandatory cloud service. By using Ollama and local models as the default, it guarantees data sovereignty and offline use. The same agent logic can seamlessly switch to more powerful cloud models when needed, all while maintaining a consistent user experience and keeping the project database (a single SQLite file) entirely under the user's control.

Frequently Asked Questions (FAQ)

1. How does shadow-planner's AI work offline? Shadow-planner uses Ollama to run open-source Large Language Models (like Qwen 2.5) directly on your computer's hardware. No data is sent to the internet; all processing for drafting plans and responding to chat commands happens locally on your machine.
2. Can I use shadow-planner with ChatGPT or Claude? Yes. While it defaults to a local model, you can optionally connect to Anthropic Claude, OpenAI ChatGPT, or OpenRouter by entering your own API key in the settings. Only the prompts for the active scenario are sent to your chosen cloud provider.
3. What are the limitations of the free version? The free version, available via the Microsoft Store or Snap Store, is fully functional but limits you to 1 planning scenario, 3 projects, and 3 resources. All features, including the AI assistant, Gantt, and utilization views, are accessible.
4. How do I share plans or collaborate with my team? Shadow-planner is primarily a desktop planner for individual use. Collaboration is achieved through exporting: you can share Gantt charts as PNG/JPEG images, export the schedule to Mermaid Markdown for documentation, or export to Microsoft Project (.xml) format for sharing with stakeholders using that tool.
5. Is my project data secure and private? With the default local Ollama setup, all data—your project plans and AI interactions—remains on your device in a single SQLite database file. No telemetry or analytics are collected. Even with a cloud AI provider, only the specific prompts necessary to process your chat commands are transmitted.