PlanVector AI today announced the availability of its first project-domain foundation model, PWM-1F, a specialized project world model designed to act as the base intelligence layer for project agents and analytics. PWM-1F runs on top of leading general-purpose foundation models, but its behavior is shaped by project-specific post-training.

The engine is built for integration into project management platforms, ERP systems and vertical SaaS products that need project-aware AI. Most current “AI for project management” efforts connect standard large language models to project files, reports and data warehouses. This can help with summarization, but it rarely answers control questions such as where risk is building, why margin is drifting, or how a project is likely to behave next.

PlanVector takes a different approach. PWM-1F is a project-domain foundation model: a specialized world model created by adapting a general foundation model to the project domain and training it to read project telemetry rather than just documents. Instead of learning from live customer projects, the engine is post-trained on a broad range of synthetic project scenarios expressed as metric histories. From these, it learns to recognize patterns in project metrics over time and to turn those trajectories into structured judgments of health and key positive and negative drivers. This capability is what PlanVector refers to as Temporal Causal Inference (TCI).

Because the engine is trained on project-domain metrics rather than one platform’s schema, it can adapt to different environments and metric sets. When a host platform sends recent project metrics, PWM-1F reads the trajectory, infers current health and identifies the drivers that matter most. Platforms can use this output as the basis for explanations, early warnings and “what if” discussions. Using metrics as its primary signal, the engine can also integrate with the host platform’s tools and APIs to pull in additional context such as tasks, risks or phases when deeper root cause analysis is required.

PlanAutomate, which serves project-based businesses on Microsoft Dynamics 365 Finance (D365F), is the first platform to deploy PWM-1F in production. Within PlanAutomate’s Enterprise Project Management system, the engine ingests metric snapshots from D365F and PlanAutomate’s project data model, identifies projects whose trajectories resemble emerging risk patterns, explains variances in terms of specific drivers, and provides assessments into PlanAutomate’s analytics and conversational tools.

PlanVector AI’s PWM-1F, is available now for project management platforms, ERP and PSA vendors, and other enterprise systems that manage projects. PWM-1F is built on top of Google Gemini as a domain-specific foundation model, functioning as a multi-modal project world model and temporal causal analysis layer on a general foundation backbone.

About PlanVector AI

PlanVector AI builds and fine-tunes specialized AI models for the project management industry, with a focus on project world models and temporal analysis that support robust, explainable project assessment.

Read more about PlanVector PWM-1F here: https://www.planvector.ai/next-wave-of-ai-for-project-management-pwm-1f-a-project-domain-foundation-model/

View source version on businesswire.com: https://www.businesswire.com/news/home/20251209960896/en/

PlanVector AI’s PWM-1F, is available now for project management platforms, ERP and PSA vendors, and other enterprise systems that manage projects.