Tool Calling

Function/tool-calling inside the reasoning engine is on the roadmap. Here is what to use today.

#Status

Coming soon LLM function/tool-calling inside the reasoning engine is not available today.

Coming soon

Vedika does not currently support LLM function/tool-calling inside the AI Query reasoning engine — there is no way to hand the model a function schema and have it decide when to invoke it mid-answer. This is on the roadmap; it has not shipped.

#What you can do today

Vedika already exposes 700+ deterministic operations — chart math, dashas, doshas, panchang, matching and more — that behave like callable tools: fixed inputs, fixed outputs, no ambiguity. Orchestrate them yourself from your own code, or, on Business and above, point an MCP-compatible client at the Vedika MCP server and let it call them directly.

All EndpointsEvery deterministic operation family, grouped and counted.MCP ServerBusiness+Point an MCP-compatible client at Vedika's operations.

#What we'd build

The natural next step is a tool-calling layer that lets the reasoning engine invoke Vedika operations mid-answer — for example, pulling a live dasha timeline partway through a synthesis, rather than relying only on the pre-computed chart context. That layer does not exist yet; this section describes a direction, not a shipped feature.