Triskell Software has expanded the artificial intelligence capabilities of its project and strategic portfolio management platform with the release of Triskell 2026 SP2. The update introduces a Model Context Protocol (MCP) Server for connecting external AI assistants to portfolio data, alongside an AI Knowledge Base, configurable AI Skills and new controls for restricting sensitive information shared with external AI providers. The changes are aimed at helping organizations bring conversational AI into project operations, resource planning and strategic decision-making.
Triskell Brings Conversational AI Into Portfolio Management
Project and strategic portfolio management platforms are increasingly incorporating AI to help organizations move from tracking project activity toward making faster, more informed decisions about investment priorities, delivery risks and resource capacity.
Triskell’s latest release targets several parts of that workflow. Its AI Assistant can analyze portfolio information, generate summaries and charts, identify resource constraints, and perform selected operational tasks through natural-language instructions. Users can also request actions such as creating project objects, updating timesheets, generating SQL queries, configuring business rules, and translating or summarizing content.
The objective is to reduce the time employees spend navigating complex portfolio data and handling repetitive administration. For portfolio managers, the potential benefit is easier access to information needed to assess project progress, identify bottlenecks and evaluate whether resources are aligned with strategic priorities.
However, the value of these capabilities will depend on how reliably the assistant interprets enterprise data and whether organizations can maintain appropriate controls over actions that affect project records.
MCP Server Connects Triskell to External AI Assistants
The most notable addition in Triskell 2026 SP2 is its MCP Server, which allows compatible external AI assistants to interact with the company’s portfolio management environment.
The Model Context Protocol (MCP) is an open standard for connecting AI applications with external systems. It provides a structured way for AI assistants to access information and invoke tools exposed by connected services. GitHub+1
In Triskell’s implementation, users can ask an external assistant to retrieve project information, examine attributes, create project objects or record timesheet entries. This extends access to portfolio data beyond the platform’s traditional user interface and potentially allows employees to incorporate project operations into existing AI-powered workflows.
The approach reflects a broader shift in enterprise software: business applications are increasingly expected to make their data and functions accessible to AI assistants, rather than requiring users to work exclusively within individual application interfaces.
For portfolio management, this could be useful when executives need quick summaries of project status, program managers want to investigate resource constraints, or team members need to update records without moving between multiple screens.
The introduction of an MCP connection also raises questions about permissions and oversight. Access to project records and the ability to execute actions require careful control, particularly when an AI assistant can modify operational data. The MCP specification emphasizes transparency around exposed tools and human oversight of tool invocations.
Triskell’s announcement does not provide detailed information about the MCP Server’s authentication model, approval mechanisms or audit capabilities. Those details will be important for organizations evaluating the feature for sensitive portfolio workflows.
RAG and Custom AI Skills Add Enterprise Context
The release also introduces an AI Knowledge Base that uses Retrieval-Augmented Generation (RAG) to provide responses grounded in uploaded corporate documentation.
RAG connects a generative AI model to an external knowledge source. Instead of relying exclusively on information contained in its training data, the system retrieves relevant material from an organization’s documentation and uses it to formulate a response.
For portfolio teams, this could make internal project methodologies, business policies, process documentation and organizational guidance more accessible through conversational queries. It may also help users interpret portfolio information in the context of company-specific rules and terminology.
Custom AI Skills provide another layer of configuration. Administrators can define instructions for different business contexts, shaping how the assistant interprets and responds to project and portfolio information.
This is relevant because portfolio management practices vary between organizations. A product development team, an IT transformation office and a corporate strategy function may use different approval processes, prioritization criteria and performance indicators. Configurable instructions could help adapt AI interactions to those requirements.
The quality of RAG-based responses will still depend on the relevance, accuracy and currency of the source documentation. Organizations will need to maintain their knowledge bases and ensure that users receive information appropriate to their access rights.
Data Security and Model Choice Address Governance Concerns
Triskell is also adding a Data Security module that allows administrators to exclude sensitive attributes from information shared with external AI providers.
The control is significant for enterprises that manage confidential project information, financial data, resource records or strategic plans. Sending information to an external model provider can introduce data handling and compliance considerations, particularly when multiple teams use AI features.
The company says customers can choose between its built-in OpenAI provider and supported alternatives from Anthropic, Google Gemini and Mistral. This provider flexibility gives organizations options when evaluating model capabilities, vendor arrangements and internal AI policies.
However, excluding selected attributes is only one part of enterprise AI governance. Companies also need to consider user permissions, data retention, model-provider terms, activity monitoring and the potential for AI-generated errors. Triskell has not specified the full scope of its security controls in the release announcement.
AI Is Reshaping the PPM Market
The update arrives as project and portfolio management leaders reassess how AI should fit into planning, execution and governance. Gartner’s 2026 research on PPM operating models identifies AI-enabled planning and execution as part of the changing technology environment for portfolio management leaders. Its guidance also highlights the need to manage adoption risks while modernizing the PPM function.
The Project Management Institute has also published a global standard for applying AI in portfolio, program and project management, reflecting the growing need for structured practices as organizations introduce AI into project work. Project Management Institute
Against that backdrop, Triskell’s strategy combines embedded assistance, external AI connectivity, company-specific knowledge and data controls. Rather than positioning AI solely as a reporting feature, the company is extending it into operational workflows and the wider enterprise AI environment.
Competition in this space will increasingly depend on more than the ability to generate summaries or answer questions about project status. Organizations will assess whether AI can improve portfolio visibility, support resource decisions, preserve governance standards and connect reliably with other business systems.
For Triskell, the MCP Server is a move toward making portfolio management part of broader AI-assisted work. Its success will depend on the depth of its integrations, the safeguards around AI-driven actions and the accuracy of insights generated from enterprise information.
Market Landscape
Project and Strategic Portfolio Management software is moving toward AI-assisted planning, analytics and execution. Vendors are adding natural-language interfaces, knowledge retrieval and integrations that let external assistants access business application data. Triskell’s 2026 SP2 release combines these capabilities with configurable AI Skills, provider choice and controls over sensitive attributes. As enterprises expand AI use across project operations, governance, data permissions and reliable execution will be important differentiators.
Top Insights
- Triskell 2026 SP2 adds an MCP Server that connects compatible external AI assistants with project and portfolio information.
- The AI Assistant can summarize data, generate charts, identify resource constraints and support selected project administration tasks.
- A RAG-powered AI Knowledge Base uses corporate documentation to provide responses grounded in organizational information.
- Custom AI Skills let administrators configure instructions for different business contexts and portfolio management requirements.
- A new Data Security module allows administrators to exclude sensitive attributes from information shared with external AI providers.
Power Tomorrow’s Intelligence — Build It with TechEdgeAI
