SynergySign
Signatures that land on the line.
E-signature the way it should work: upload a contract, let real AI read it and place the fields on the document's actual signature lines, send secure links your counterparties can sign without an account — and get back a tamper-evident signed PDF.
Upload, detect, invite, sign, done
The full document lifecycle: upload a PDF, run AI detection, invite signers one after another or all at once, and finish with a stamped, tamper-evident, certified document — the whole loop tested end to end in the running product.
🔗One-click signing links
Outside parties open a unique signing link — no account, no login wall, no app to install. The link itself is the key, and it works for exactly one signer on exactly one document.
🖊️DocuSign-grade signing UI
Signature, date, text, and dropdown fields rendered on the real PDF; sequential signing order fires "your turn" notifications between parties.
🧾Tamper-evident finish
Completion produces a real signed PDF with signatures embedded and dates burned in — plus a stored digital fingerprint, so any later change to the file is immediately detectable.
Field detection that reads like a lawyer, places like a typesetter
One pass by the platform's built-in AI reads the uploaded document and returns its type, a summary, the parties and their signing order, compliance flags — and suggested fields with real page positions. The work is split cleanly: the AI understands the words, precise measurement places the fields.
- The document's layout is preserved during reading, so the AI knows where things actually sit on the page
- The system finds the document's real fill-in lines and snaps each field's bottom edge onto the line exactly — worst placement error measured at zero
- Fields bind to the right signer automatically from the detected parties
- Documents with no ruled lines fall back to the model's estimate, honestly badged "estimated — check these" in the UI
Live proof
On a real mutual NDA: it identified the document type, spotted New York governing law, pulled both parties by name, and placed four fields on the actual signature lines with 97% confidence — in a 19-second live run. It even flagged a signature block that was missing its line.
Honest AI is a feature. An earlier iteration of this product shipped fake AI — fixed field positions dressed up as detection. It was ripped out and replaced with the platform's built-in AI reading each live document — one that reports failure rather than making things up — and every claim on this page was verified against the running system.
Two doors, on purpose
Senders and signers are different populations, so SynergySign runs a deliberate two-identity model instead of hiding everything behind one login wall.
🏢Staff via company login
Team members sign in with SynergyAuth — the same seamless sign-in as Drive, Boards, and Meet — landing straight in the dashboard.
🌍Signers via secure link
Signing links stay open by design: an outside counterparty should never hit a corporate login page. The link admits them to exactly one signing session and nothing else.
How it's put together
- Database updates never lose data and are safe to re-run — a fresh production install comes up correct on its own
- Every completed document carries an audit log, a stored digital fingerprint, and emailed completion copies
- Finalization is a single source of truth that fails loudly and stays retryable — it never marks a document complete without producing the signed PDF
Next.js + PrismaPostgreSQLBuilt-in AI document readingPDF display + stampingOne company loginnginx + TLSDocker Compose