For project managers and business leaders
Resource and capacity planning with AI
- MCP Server integrated
- Local MCP server
- Undo for plan changes
Who it is for
Who Rillsoft AI is made for
- Project-driven industrial and mid-sized companies running several projects in parallel that share people, machines and material.
- Planners, PMOs and project managers who work with or evaluate Rillsoft Project 10 on Windows and want to query or edit their plan with an AI assistant.
- Integrators and administrators who read or write planning data by script — with or without AI.
For management
Which projects need attention now?
See project traffic lights for schedule, progress, resources and costs. For each critical assessment, the report shows the relevant metric and the tasks or roles involved. Choose verified chapters for your report or explore the example from a test portfolio.
Three steps to your report
- Choose chaptersSet the scope, period and verified questions.
- Copy the promptReview and copy the assembled text.
- Run it in your AI clientYour connected AI client creates the report from Rillsoft Project.
How it works
One question. The right project data. A reliable answer.
Where will capacity become a bottleneck in the next three months?
- Capacity view
- Skills
- Absences
- Demand & supply
- Monthly scale October – May
In the first three months – October to December 2026 – electrical design and technical purchasing exceed their available capacity – on 33 and 20 of 65 working days – because demand and base load coincide. The assembly skills only run short around Christmas and in January, and only just; the others stay within their capacity.
This example compares demand and supply per job role. Rillsoft Project supplies the capacity figures; the AI sets the view by role and reads the shortfall. The source is the shipped synthetic mechanical engineering sample package, Rillsoft Project 10, read from the view.
Setup
Connect your AI client
http://127.0.0.1:3928/mcpChoose your AI client
Claude Code
claude mcp add --transport http rillsoft-project http://127.0.0.1:3928/mcpClaude Desktop
{
"mcpServers": {
"rillsoft-project": {
"command": "npx",
"args": ["-y", "[email protected]", "http://127.0.0.1:3928/mcp"]
}
}
}Codex
codex mcp add rillsoft-project --url http://127.0.0.1:3928/mcpVS Code
{
"servers": {
"rillsoft-project": {
"type": "http",
"url": "http://127.0.0.1:3928/mcp"
}
}
}Cursor
{
"mcpServers": {
"rillsoft-project": {
"url": "http://127.0.0.1:3928/mcp"
}
}
}Other clients
{
"mcpServers": {
"rillsoft-project": {
"type": "http",
"url": "http://127.0.0.1:3928/mcp"
}
}
}Your benefits
What can you do with Rillsoft AI?
Spot the bottleneck Compare demand and supply per job role. See on which days capacity is short. See example → Find the right resource Who is free, qualified and without a schedule conflict? Also outside the preferred team when the site is fully booked. See example → Read the plan status Delayed tasks, missing float, variance from the baseline in dates, effort and cost. See example → Analyse the project portfolio Review project status, delivery dates at risk and bottlenecks together. Choose verified questions for your report. Choose management questions →Your AI. Your projects.
An open connection.
- Projects
- Tasks
- Resources
- Capacities
- Portfolios
Security
Bound to your machine, no overpromise
- The server binds to
127.0.0.1only; requests carrying a browser origin are rejected. - Optional API key for multi-user and terminal-server machines – without a key there is no additional authentication.
- Optional read-only mode: ask, understand, change nothing. It does not replace organisational access control.
- Every mutating call is one undo step; file and session operations are not.
- What your AI client sends to its provider is governed by that provider’s privacy terms – not by Rillsoft.
- The trade-off: an MCP call carries no user identity and reaches exactly one Rillsoft Project instance.
There is no hosted Rillsoft MCP server. The server runs inside your installation.
Without AI
Scripts and integrations
The same endpoint works without an AI assistant: with curl, PowerShell or any language that speaks HTTP and JSON. Mirror absences from the leave system, pull reports on a schedule, generate projects from templates – every tool describes input and output as a JSON schema, errors carry stable codes.Handshake, tools/list and a first read-only call are on the MCP page.
Frequently asked
What to know before the first prompt
Does the AI recompute my schedule?
Rillsoft Project calculates dates, workload and the critical path. The AI reads and analyses these results. Plan changes made through the corresponding MCP tools can be undone; file and session operations are excluded.
Does my project data leave the machine?
The MCP server runs locally in the open Rillsoft Project and provides project data to the connected AI client. The client may transmit this data to its AI provider. What is transmitted depends on the client’s settings and terms.
How is this different from connecting an AI to a task tool?
The AI works with calculated dates, float and capacity data from Rillsoft Project. The verified management analyses compare demand and supply per job role and account for shortfalls on individual days. For staffing, the candidate list provides qualifications, availability, costs and schedule conflicts. The comparison overview explains how Rillsoft Project compares with other planning tools.