Changelog
What shipped, newest first.
A real front end, and pages that were missing
- Every interface rebuilt on one component library and one set of design tokens, across all four sites.
- Published pricing: packs, what each operation costs, and the guarantees, generated from the billing code so the page cannot drift from the charge.
- An API reference with the error codes and limits the server actually enforces.
- A setup page for the MCP server, and an index that searches every free tool.
- Privacy and terms pages.
The last five products
- Document conversion across the pairs the converter was measured doing, each with a working converter on its page.
- Tables out of PDFs into CSV, XLSX or JSON, charged per table-bearing page rather than per page of the file.
- Evidence-grade web capture on its own brand, with a public verifier anyone can check a capture against.
- Chat and email exports turned into a readable, searchable archive — no account, paid per job.
- Messy spreadsheet cleaning with a diff you approve before anything is accepted.
- Web page to clean markdown, with adverts and cookie banners blocked before the page renders.
The platform
- Markup to PDF: Typst, HTML and Markdown, with stored templates.
- Eight PDF operations, most with a free browser tool that never uploads the file.
- True redaction that removes the content stream rather than drawing a black rectangle, plus a metadata scrubber.
- Any file to LLM-ready markdown or JSON, with a classifier that decides whether OCR is needed before charging for it.
- Credit ledger, API keys, one-time packs, and an MCP server where every tool reports the credits it used.
- Sign-in by Google or a six-digit emailed code. No passwords, either way.