Can you call multiple LLM plugins in a single Plum input?

​

How Xoxoday Plum handles LLM plugin invocation

Xoxoday Plum enforces a single-plugin-per-input rule across its AI layer. When a user submits a request, the system identifies and triggers exactly one LLM feature or plugin to process that input. This design choice is intentional and directly supports the accuracy and reliability that enterprise workflows demand. Attempting to invoke multiple LLM plugins simultaneously within a single input does not result in parallel execution. Xoxoday Plum resolves the most contextually relevant plugin for that request and processes it exclusively. This prevents ambiguity in how the instruction is interpreted and ensures the response maps cleanly to the user’s intent.

​

Why concurrent plugin calls are restricted

When multiple LLM plugins operate on the same input at once, competing interpretations can produce conflicting outputs or degrade the quality of context passed downstream. Xoxoday Plum avoids this by serialising plugin execution at the input level. This architecture also protects system performance under load. In enterprise environments where Xoxoday Plum is integrated with HRIS platforms like Workday, SAP SuccessFactors, or Darwinbox, a single reward or recognition workflow can trigger cascading downstream actions. Isolating plugin execution per input keeps each step predictable and auditable.

​

Performing multi-step AI interactions

Sequential interactions are fully supported. If a workflow requires multiple LLM capabilities — for example, first generating a personalised reward message and then translating it for a regional team — each step is submitted as a separate input. Xoxoday Plum processes them in order, maintaining context across chained prompts through its session layer. A practical example: a manager using Xoxoday Plum inside Microsoft Teams or Slack can ask the AI assistant to draft a recognition note, then follow up with a request to adjust the tone for a formal annual award. Each prompt is handled as a discrete input, and the conversation history informs each successive response without requiring parallel plugin calls.

​

What this means for integration and compliance

For IT and security teams, this single-invocation model simplifies audit trails. Each AI interaction is logged as a discrete, traceable event, which aligns with the requirements of compliance frameworks such as ISO 27001 and SOC 2 Type II. There is no ambiguity about which plugin acted on which input, making incident review and access logging straightforward. Xoxoday Plum’s AI architecture is built to scale across large organisations without sacrificing interpretability. The one-plugin-per-input constraint is a deliberate quality guarantee, not a capability ceiling. Learn more: Xoxoday Plum Help Centre — AI LLM