VerbatimCV

An Affinda alternative for AI agents

Choose VerbatimCV when an AI agent consumes the result: every preference carries the quote it came from, hidden prompt injection is stripped and reported, and agents can connect over MCP. Choose Affinda if you need image or legacy Word inputs, self-hosting, or a long production track record. Facts below come from Affinda's own pages, read on 17 November 2026.

VerbatimCVAffinda resume parser
Free start50 parses a month, no card1,000 free documents in the first 14 days
Self-serve pricePaid plans not yet publishedUS$0.10 per document
InputsPDF, DOCX (scans OCR'd)PDF, DOC, DOCX, TXT, RTF, HTML, images, ODT, XLS/XLSX
Self-hostingNoYes (annual licence)
Quote for each extracted preferenceYes (evidence)Not mentioned
Hidden text and prompt-injection handlingRemoved and reportedNot mentioned
"What to confirm" list for agentsYes (confirm_before_acting)Not mentioned
MCP serverYes (uvx verbatimcv-mcp)Not mentioned
Training on customer dataNo. CVs aren't stored by VerbatimCV; Anthropic, which runs the extraction model, deletes API data within 30 days and doesn't train on it"No training on customer data"

Where Affinda is stronger

More input formats, self-hosting, a claimed 50+ languages and a 50 ms median parse time, and years of production use. VerbatimCV uses a language model, so a parse typically takes 5–10 seconds.

Sources

"Not mentioned" means not stated on those pages on that date; the feature may exist elsewhere.

Try it

Get a free API key (50 parses a month, no card), then:

curl https://api.verbatimcv.com/v1/parse \
  -H "Authorization: Bearer $VERBATIMCV_API_KEY" \
  -F file=@cv.pdf

Agents can use the parse_resume tool instead: uvx verbatimcv-mcp. See the quickstart for Python, JavaScript and MCP setup.