Image & PDF input
Support for JPEG, PNG, WebP, and PDF. Handwritten or printed; single page or multi-page. Optional folder structure (e.g. Topic/Grade/Month) for organizing uploads and jobs.
From worksheet to verified solution in minutes
Upload a math sheet—image or PDF—and receive structured questions with step-by-step solutions. Multi-agent AI and a coordinator ensure consistent, explainable results you can rely on.
What is Math Solver?
Math Solver is an AI-powered pipeline for education and content teams. It takes raw input—photos of handwritten or printed math, or PDF question papers—and returns:
- Extracted questions in structured form (LaTeX, multiple choice, labels)
- Step-by-step solutions for each question, with a final answer
- Consensus-based results from multiple AI agents, compared by a coordinator
- Similar questions generated on demand for practice or assessment
All calculations go through a dedicated math utility—no “mental math” from the model—so numeric answers are precise and auditable.
Features
Built for reliability, transparency, and integration into your workflow.
Structured extraction
Questions are stored as JSON with LaTeX, optional choices and choice labels, and metadata. Same structure whether you upload from the Solver page or the Dashboard.
Multi-agent solver
Configurable number of agents per question. Each agent uses the same math tool; the coordinator only accepts consensus, reducing single-model errors.
Step-by-step solutions
Every solution includes numbered steps and a final answer. Steps are parsed and displayed in the UI and in exports (PDF/DOCX). Human-readable fractions and symbols, not raw LaTeX.
Similar questions
Generate similar questions for any extracted item. Useful for practice sets, alternate forms, or item banks. Stored per job and viewable in the app.
Dashboard & jobs
Dashboard for uploads, jobs (questions), and steps. Filter jobs by source file, rename files/folders with automatic relinking of jobs. Delete jobs or files with cascade cleanup.
Export
Download results as PDF (from rendered UI) or as editable DOCX. LaTeX in content is converted to readable text (e.g. 1/3, ×, √) in exports.
Configurable models
Use different models for extraction vs. solving (e.g. Opus for vision extraction, Sonnet for solving). Prompts are externalized in YAML for tuning without code changes.
Use cases
From classroom materials to item banks and platform integration.
Educators & tutors
Turn existing worksheets or past papers into solved keys with steps. Share solutions or similar questions for practice without reworking content by hand.
Publishers & content teams
Scale solution authoring for textbooks and digital products. Keep LaTeX and structure; generate solutions and variants from scans or PDFs.
EdTech platforms
Integrate via API: upload → extract → solve. Use job IDs and stored results to show solutions in your own UI or to power adaptive practice.
Assessment & item banks
Maintain question banks with structured data. Generate similar items and solution keys; export to DOCX or PDF for print or LMS.
Technology
Designed for accuracy, auditability, and deployment.