For genuinely low-stakes documents, a signature flow is a bounded, achievable build — upload a PDF, capture a signature, flatten it onto the page. The audit-trail and legal-recognition weight of a real e-signature product is the deeper, deliberately out-of-scope part.
Build it
Effort: a weekendRunning cost: $0/mo to start
Why CanICodeThis says Build it
Uploading a PDF and letting a recipient draw or type a signature (canvas input) is straightforward
Flattening the signature onto the document at a marked position using an existing PDF library (pdf-lib) is well-documented
A signing-status tracker (sent, viewed, signed) is a small state machine on a document table
The parts that'll cause problems
A legally defensible audit trail (IP, timestamp, consent language) needs to be genuinely correct if it matters — the same caution as this register's DocuSign entry
Multi-party sequential signing (party A signs, then party B) needs a small workflow layer beyond a single signer
Template documents with fill-in fields add real, separate templating scope
How I'd build this
frontend
Next.js
pdf
pdf-lib
database
Supabase
hosting
Vercel
MVP scope
Upload a PDF and mark a signature position
Recipient draws or types a signature via canvas input
Flatten the signature onto the document and email both parties the result
Postpone to v2
Multi-party sequential signing
Template documents with fill-in fields
A tamper-evident audit trail
What it actually costs to build
AI agent — with a subscription (Claude Pro/Max, Cursor, etc.)
$0 marginal
AI agent — pay-per-use API, no subscription
$43–$86 one-time
Hosting, once it's running
$0/mo (free tier)
Domain name, if you want your own
~$12/yr
Existing tools solving a similar problem typically run $15–$35/mo — see how they compare below.
The build prompt
Build me a document-signing tool for genuinely low-stakes agreements:
1. Upload a PDF and mark where a signature should go.
2. The recipient draws or types a signature via canvas input.
3. Flatten the signature onto the document using pdf-lib — burn it into the page, not an overlay.
4. Email both parties the signed result with a basic log (who signed, when).
Single signer only, and label it clearly as informal, not a legal audit trail. Out of scope: multi-party sequential signing, which needs a real workflow layer beyond a single signer, and a tamper-evident audit trail suitable for real disputes, which carries real liability weight if it doesn't hold up — that's what a dedicated e-signature service exists to get right.
