ParseMeSpreadsheet Intelligence
Your spreadsheets,
understood.
ParseMe reads Excel workbooks like an expert analyst. It traces formulas, maps dependencies, and extracts verified answers that cite the exact cells they came from. No more fragile scripts or copy-paste workflows.
The problem
Spreadsheets aren't flat files
Every workbook is a program. Formulas reference other formulas across sheets, styles carry meaning, and one changed cell can cascade through hundreds of others. Dumping it to CSV loses all of that. ParseMe keeps the full picture.
How it works
From messy workbook to structured data
Upload your workbook
Send any .xlsx or .xlsm file to the API. Financial models, reporting packs, data exports: we handle them all.
Parse every layer
We extract formulas, styles, merged cells, named ranges, and hidden sheets into a typed Workbook IR. Nothing gets lost.
Map the structure
Semantic detectors identify headers, tables, formula families, and data regions. A dependency graph traces every calculation.
Extract with proof
Ask questions, extract schemas, or audit formulas. Every answer comes with cell references and validation against the source.
What you get
Built for real spreadsheets
Not another "convert to CSV and hope for the best" tool. ParseMe understands workbook semantics the way you do.
Typed Workbook IR
Every cell, formula, style, and structure parsed into a clean, typed intermediate representation. Merged cells, hidden rows, named ranges: all preserved.
Dependency Graph
Trace formula lineage across sheets. See what feeds into any cell, find circular references, and understand calculation flow at a glance.
Semantic Detection
Automatic identification of headers, data tables, formula families, input cells, calculation blocks, and output regions.
Hybrid Retrieval
Find the right cells fast. Lexical search, semantic embeddings, value matching, and graph-aware signals fused together for precise lookups.
Validated Extraction
Every answer is checked against the workbook. Citation validators and formula validators ensure you get facts, not hallucinations.
AI-Native
Purpose-built context for LLMs. Structured workbook representations that models can actually reason about, not walls of serialized text.
Use cases
Wherever spreadsheets are critical
Financial Model Ingestion
Extract schemas, trace assumptions, and validate calculations in LBO models, DCFs, and budgets, with full formula lineage.
Data Pipeline Normalization
Turn messy reporting packs and KPI exports into clean, structured data. No more per-client CSV scripts.
Spreadsheet Auditing
Find hardcoded values masquerading as formulas, detect broken links, flag circular references, and spot hidden sheets.
Workbook Q&A
Ask natural language questions about any workbook and get verified answers that cite specific cells, not guesses.
Coming soon
A simple API for complex workbooks
Upload a workbook, ask a question, get a verified answer. A few lines of code is all it takes.
import parseme
# Upload and parse a workbook
wb = parseme.parse("financial_model.xlsx")
# Trace formula dependencies
deps = wb.graph.dependents("Sheet1!B12")
# Ask a question with verified answer
answer = wb.ask(
"What is the projected revenue for Q3?",
validate=True
)
print(answer.value) # "$4.2M"
print(answer.cell_ref) # "Summary!D15"
print(answer.confidence) # 0.97Proven on real benchmarks
We test ParseMe against public spreadsheet benchmarks to make sure it actually works on the kind of messy workbooks you deal with every day, not just toy examples.
900+
benchmark tasks
8
eval suites integrated
45%+
Pass@1 accuracy
Early access
Be first in line
We're building the API now. Drop your email and we'll invite you as soon as it's ready. No spam, just spreadsheet intelligence.
Or just say hi at hello@parseme.dev