Math Agent.
Agentic AI tutor that decomposes problems, calls solver tools and shows its work
Planner LLM dispatches to symbolic + numeric solvers
Step-by-step working reconstructed from tool calls
Built-in plotter and unit-converter tools
Every answer is auditable — no one-shot hallucinations
An agentic calculus / algebra tutor wired together with LangChain tool-use. A planner LLM parses the natural-language problem, decides which solver tools to call (SymPy for symbolic, NumPy/SciPy for numeric, a step-explainer, a plotter, a unit-converter) and stitches a structured solution back together. The student never gets a one-shot model guess — they get a transparent chain of tool invocations and intermediate states, which is exactly the difference between a chatbot and a tutor.
