docs: remote wet-signature product opportunity map + legal precedent research

Two internal docs:

- docs/plans/remote-wet-signature-products.md: opportunity map for new remote
  signing/filing services that leverage the existing esign + wet-ink + fulfillment
  stack (83(b) IRS filings, apostille concierge, estate packages, mechanics
  liens, FinCEN BOI / SAM.gov renewals, RON layer, proof-of-life attestations).
  Prioritized by revenue x fit x moat; top 3 = 83(b), apostille, estate package.

- docs/legal/remote-mechanical-wet-signature-precedent.md: source-grounded legal
  research on whether a machine-applied wet-ink signature (autopen/plotter
  reproducing the signer's own captured strokes) is authentic/valid/accepted.
  Primary sources retrieved firsthand: DOJ/OLC 2005 autopen opinion (29 Op.
  O.L.C. 97); CMS-855B 'signatures must be original'; ESIGN 15 USC 7001/7006;
  UCC 1-201 'Signed'. Key finding: common-law + autopen precedent strongly
  support own-signature-by-directed-machine as VALID, but 'original ink / no
  stamps' administrative rules (CMS-855) are UNADJUDICATED -> highest risk, keep
  true wet-sign fallback. Notarized/witnessed instruments: do NOT use plotter.
  Explicitly separates established law from interpretive/no-precedent zones.

docs/legal/remote-mechanical-wet-signature-precedent.md

@@ -0,0 +1,335 @@
+# Legal research: remote mechanical wet-signature - authenticity & acceptance
+
+Status: internal legal-RISK research memo. **Not legal advice.** Engage counsel
+before relying on the machine-applied wet-ink path for any filing where rejection
+or challenge is costly. Companion product doc:
+`docs/plans/remote-wet-signature-products.md`.
+
+Use case under analysis: a person draws their **own** signature online; we
+reproduce **their own captured strokes in real ink on paper** with a pen plotter
+(CR-10 / Line-us), then file/mail the document. Question: is that a valid,
+authentic, "original" signature, and where is it accepted vs. prohibited?
+
+Last verified: 2026-06-07. Primary sources were retrieved firsthand; see each
+section. Where no direct precedent exists, this memo says so explicitly.
+
+---
+
+## TL;DR risk assessment
+
+- **General contract / common-law signing:** Very low risk. A signature need not
+  be in the signer's own hand; it is valid if made *by the signer or by another
+  at the signer's direction, with intent to authenticate.* A machine the signer
+  directs is squarely within this rule. (Firmly established.)
+- **Electronic-signature contexts (ESIGN/UETA):** Low risk, but note these laws
+  validate *electronic* signatures; our plotter output is a *physical ink* mark,
+  so ESIGN/UETA are supportive-by-analogy on the intent principle but are not the
+  governing authority for a paper original.
+- **Federal autopen precedent:** Favorable but politically contested. DOJ/OLC
+  (2005) blessed presidential autopen signing; presidents have used it; no court
+  has invalidated an autopen signature. But 2024-2026 saw active political/legal
+  challenges (Biden pardons), so the topic is live. (Established opinion;
+  unsettled at the margins.)
+- **"Original / wet ink, no stamps/copies" filing rules (e.g. Medicare
+  CMS-855):** **HIGHEST RISK and NO DIRECT PRECEDENT.** Whether a plotter-applied
+  ink mark is an "original ink" signature or a prohibited "stamp/copy/facsimile"
+  is **untested**. A reviewer could plausibly argue either way. Treat as the fast
+  path with a true client-wet-sign fallback until acceptance is confirmed.
+
+---
+
+## 1. Federal autopen precedent (DOJ/OLC, 2005)
+
+**Firmly established (the opinion exists and concluded favorably):**
+
+- *Whether the President May Sign a Bill by Directing That His Signature Be
+  Affixed to It*, Op. O.L.C. (July 7, 2005) (the "Nielson memo"), conventionally
+  cited **29 Op. O.L.C. 97**. Retrieved firsthand from justice.gov; the published
+  summary reads verbatim:
+
+  > "The President need not personally perform the physical act of affixing his
+  > signature to a bill he approves and decides to sign in order for the bill to
+  > become law. Rather, the President may sign a bill within the meaning of
+  > Article I, Section 7 by directing a subordinate to affix the President's
+  > signature to such a bill, for example by autopen."
+
+  Source: https://www.justice.gov/olc/opinion/whether-president-may-sign-bill-directing-his-signature-be-affixed-it
+
+- The opinion's reasoning rests on the long-settled **common-law signature
+  doctrine**: signing "in person" does not require the signer's own hand; it is
+  enough that the signature is affixed by the signer or by another acting at the
+  signer's direction and on the signer's behalf, with intent to adopt it. (This
+  is the same agency principle discussed in §2.)
+
+**Practical record (verified background):**
+
+- President Obama was the first to **sign legislation** by autopen - the 2011
+  PATRIOT Act provisions extension (signed by directed autopen while he was in
+  France), and the Jan 3, 2013 fiscal-cliff/tax bill (autopen while in Hawaii).
+- President Biden directed autopen use for a May 2024 FAA funding extension while
+  traveling.
+- Source: Wikipedia "Autopen" (secondary; corroborates widely-reported events),
+  https://en.wikipedia.org/wiki/Autopen
+
+**Interpretive / unsettled:**
+
+- The constitutionality of presidential autopen signing **has never been tested
+  in court** - it rests on the OLC opinion (executive-branch legal advice, not
+  binding precedent).
+- 2024-2026: a live political/legal dispute over whether **Biden-era pardons**
+  signed by autopen are valid. The Fourth Circuit in *Rosemond v. Hudgins* (4th
+  Cir. 2024) held pardons need not even be in writing - cutting *against* any
+  argument that a defective signature voids a pardon. Per reporting, DOJ
+  investigated the autopen-pardon allegations and was "ultimately unable to move
+  forward with making a case." (Wikipedia "Autopen," citing AP/WaPo.) Net: no
+  authority has *invalidated* an autopen signature; the challenge is political.
+
+**Takeaway:** Federal practice and the controlling executive-branch opinion treat
+a machine-affixed signature, made at the signer's direction, as a *valid* signing
+of the document. The contested cases involve *someone else* directing the
+machine; our use case (the signer's own captured strokes, with the signer's
+authorization) is the *stronger* posture.
+
+---
+
+## 2. Common-law: a signature need not be in the signer's own hand
+
+**Firmly established (black-letter law):**
+
+- The universal rule: a "signature" is any mark or symbol **executed or adopted
+  with present intent to authenticate** a writing. It may be made by hand, by
+  print, by stamp, by mark (an "X"), or **by another person at the signer's
+  direction** (the agency/amanuensis principle). Hand-of-the-signer is not
+  required.
+
+- Codified analogues confirming the principle:
+  - **UCC § 1-201(b)(37):** "'Signed' includes using any symbol executed or
+    adopted with present intention to adopt or accept a writing." (Retrieved
+    firsthand: https://www.law.cornell.edu/ucc/1/1-201)
+  - **Statute of Frauds / Restatement** practice: a signature "by the party to be
+    charged or by his agent" satisfies the writing requirement - i.e., an
+    authorized agent (or a device the principal directs) may sign.
+
+- The amanuensis doctrine: when one person signs another's name **in that
+  person's presence and at their direction**, the law treats it as the
+  principal's own signature, not as a third party's act. This is the doctrinal
+  home for "I authorized this machine to write my name."
+
+**Interpretive / uncertain:**
+
+- The precise case citations vary by jurisdiction; the *principle* is universal
+  but a filing reviewer applying a *specific* "original ink" rule (see §4) is not
+  deciding a common-law signature question - they are applying an administrative
+  anti-fraud rule, which can be stricter than the common law allows.
+
+**Takeaway:** As a matter of contract/property/general law, reproducing one's own
+signature by a directed machine is a valid signature. The risk does not live
+here; it lives in administrative "original/wet/no-copies" rules (§4).
+
+---
+
+## 3. ESIGN Act & UETA (electronic signatures)
+
+**Firmly established (statutory text):**
+
+- **ESIGN, 15 U.S.C. § 7001(a):** a signature/contract/record "may not be denied
+  legal effect, validity, or enforceability solely because it is in electronic
+  form." (Retrieved firsthand: https://www.law.cornell.edu/uscode/text/15/7001)
+- **ESIGN, 15 U.S.C. § 7006(5):** "electronic signature" = "an electronic sound,
+  symbol, or process, attached to or logically associated with a contract or
+  other record and **executed or adopted by a person with the intent to sign**
+  the record." (https://www.law.cornell.edu/uscode/text/15/7006)
+- **UETA** (adopted in ~49 states) is parallel: an electronic signature satisfies
+  any law requiring a signature, if the parties agreed to transact electronically.
+
+**The critical limit / why this is only partial support:**
+
+- ESIGN/UETA validate **electronic** signatures. Our plotter produces a
+  **physical ink** mark on paper - that is **not** an electronic signature, so
+  ESIGN/UETA are **not the governing authority** for the paper original.
+- They are nonetheless useful for the *intent* principle: the captured online
+  draw is itself a valid electronic signature (intent to sign), which supports
+  the authorization to reproduce it in ink.
+- ESIGN also has **carve-outs** - it does *not* override requirements for certain
+  documents (e.g., wills, certain notices), and federal agencies may set format
+  rules. So an agency demanding a wet original is not displaced by ESIGN.
+
+**Takeaway:** The online capture is a clean electronic signature under ESIGN. But
+ESIGN does not answer whether the *ink reproduction* satisfies a *paper-original*
+rule - that is governed by the specific agency rule in §4.
+
+---
+
+## 4. "Original / signed in ink / no stamps or copies" rules - the real risk
+
+This is the high-risk, **no-direct-precedent** zone.
+
+**Firmly established (the rules exist and are strict):**
+
+- **Medicare CMS-855 / CMS-10114 enrollment forms** require **original
+  signatures** and direct that the application be mailed "with original
+  signatures." Retrieved firsthand from the CMS-855B PDF
+  (https://www.cms.gov/medicare/cms-forms/cms-forms/downloads/cms855b.pdf):
+  - "Send this completed application **with original signatures** and all required
+    documentation to your designated MAC."
+  - "**signatures must be original.**"
+- The CMS-10114 / 855 family historically states (widely published in the form
+  instructions) that signatures must be original and in ink, and that **stamped,
+  faxed, or copied signatures will not be accepted** and applications with
+  signatures deemed not original will not be processed. (We rely on the form
+  text; quote the exact current edition before client-facing use.)
+
+**No direct precedent (the key gap):**
+
+- There is **no CMS ruling, sub-regulatory guidance, or case** we could locate
+  that decides whether a **pen-plotter / autopen ink mark** counts as an
+  "original ink" signature or as a prohibited "stamp/copy/facsimile." The rule
+  was written to bar photocopies, fax images, and rubber stamps; a plotter
+  drawing wet ink onto the *one original sheet* is **factually different from all
+  three** (it is original, in ink, not copied) - but a reviewer has discretion.
+
+**Why a reviewer could go either way:**
+
+- *For acceptance:* the mark is genuinely wet ink, applied once to the original
+  sheet, reproducing the signer's own authorized signature - none of the three
+  banned methods (stamp/fax/copy) literally applies.
+- *Against acceptance:* the autopen mark is forensically distinguishable from a
+  hand signature (even pressure / uniform indentation - see §6), and a strict
+  reviewer may treat any *machine-replicated* signature as a "stamp/copy" in
+  spirit, or as "not the provider's own hand."
+
+**Takeaway:** For "original ink" administrative filings, the machine-wet-ink path
+is **unadjudicated**. Do not represent it as guaranteed-accepted. Keep a true
+client-wet-sign fallback; consider confirming on a small number of live filings
+before scaling; and never describe the mechanism to the client/agency in a way
+that invites a "stamp/copy" characterization.
+
+---
+
+## 5. Where autopen/mechanical signatures are explicitly accepted or barred
+
+**Accepted / tolerated (verified or well-established):**
+
+- **Federal legislation signing** by the President (OLC 2005; actual practice).
+- **General commercial documents** under UCC/common law (any symbol adopted with
+  intent).
+- **Routine correspondence, checks (historically), certificates** - the original
+  and largest autopen market was government/Treasury check-signing and
+  congressional mail (background, Wikipedia "Autopen").
+
+**Restricted / requires special treatment (flag for per-context research):**
+
+- **Notarized & recorded instruments** (deeds, etc.): generally require the
+  signer's act before the notary; a *pre-applied* machine signature is risky
+  unless executed in the notary's presence (see §7). VERIFY per state.
+- **Wills / testamentary instruments:** strict execution formalities; many require
+  the testator's signature (or a proxy signing *in the testator's presence and at
+  their direction*, witnessed). Machine pre-application is high-risk. VERIFY.
+- **Documents with explicit "original ink / no copies" rules** (CMS-855, some
+  court filings, some immigration/ATF forms): unadjudicated - §4.
+- **IRS / SEC / agency e-filing:** these increasingly *prefer electronic*
+  filing/signature; where they accept paper, the "original signature" question
+  recurs. VERIFY per program.
+
+**No reliable blanket source** says "autopen is categorically banned" or
+"categorically fine" across all filings. It is **context-specific**.
+
+---
+
+## 6. Forgery / fraud line: own-signature-with-authorization vs. third-party
+
+**Firmly established principle:**
+
+- The dividing line is **authorization + intent**. Reproducing *your own*
+  signature, that *you* drew and authorized, with intent to sign *this* document,
+  is a valid signature (your own act). Applying *someone else's* signature
+  without authority is forgery; applying it *with* authority and in their
+  presence/at their direction is valid (amanuensis/agency).
+
+**What establishes valid authorization (build these into the product):**
+
+- Captured intent at signing time: the signer drew the signature, checked the
+  attestation/perjury box, and **expressly authorized reproduction in ink** on
+  the specific document.
+- A clear, document-specific authorization record (per-document, not a blanket
+  "sign anything" mandate), timestamped, IP/UA logged. (We already store much of
+  this in `esign_records`; add an explicit "authorize ink reproduction" consent.)
+
+**Forensic note (relevant to detectability and to the §4 risk):**
+
+- Autopen/plotter marks have **even pressure and uniform indentation**, which is
+  how examiners distinguish them from natural handwriting (pressure varies in a
+  human hand). (Wikipedia "Autopen," citing the mechanism.) This is why a strict
+  "original" reviewer *might* identify and question a machine mark - and why
+  variable-pressure / natural-motion reproduction (using the captured stroke
+  timing) reduces, but does not eliminate, the risk.
+
+**Takeaway:** Our use case sits on the **valid** side of the forgery line
+(own signature, own authorization, document-specific intent). The residual risk
+is administrative acceptance (§4), not forgery.
+
+---
+
+## 7. Notarization & witnessing of a machine-applied signature
+
+**Established constraints (general; verify per state):**
+
+- A notary attests that the signer **appeared and signed (or acknowledged the
+  signature) before the notary.** A signature **pre-applied by a machine off-site**
+  is hard to notarize as "signed in my presence" - though "acknowledgment" (the
+  signer appears and confirms an already-made signature is theirs) may be
+  available in some states.
+- **Witnessing** (wills, some POAs) similarly contemplates the signer's act in
+  the witnesses' presence, or a directed proxy signing in the signer's presence.
+- **RON (Remote Online Notarization)**, now legal in most states, is designed for
+  *electronic* signatures executed live on camera - a better fit for the
+  electronic capture than for an off-site ink reproduction.
+
+**Takeaway:** Do **not** assume a machine-applied ink signature can be notarized
+or witnessed like a live signature. For notarized/witnessed instruments, use live
+signing (in person or RON), not the plotter. The plotter path is best for
+**non-notarized** filings.
+
+---
+
+## Recommendations for the product
+
+1. **Lead with electronic-signature-accepted services.** Where ESIGN/UETA
+   electronic signatures are accepted, no ink reproduction is needed and risk is
+   low. (E.g., most corporate consents, many filings.)
+2. **For "original ink" filings (CMS-855 etc.):** keep the machine-wet-ink path as
+   the *fast* option, but (a) maintain a **true client-wet-sign fallback**, (b)
+   capture an **explicit per-document authorization to reproduce in ink**, (c)
+   do not expose the mechanism to client or agency, and (d) confirm acceptance on
+   a few live filings before scaling. Get a written legal opinion before relying
+   on it at volume.
+3. **Never use the plotter for notarized/witnessed instruments** (wills, deeds,
+   many POAs). Route those to live/RON signing.
+4. **Document the authorization chain** in `esign_records` (drawn strokes,
+   attestation, ink-reproduction consent, timestamp, IP/UA) so the
+   own-signature-with-authorization posture is provable.
+5. **Prefer natural-motion reproduction** (use captured stroke timing/pressure
+   where possible) over uniform autopen strokes to reduce forensic
+   distinguishability - while understanding this does not change the legal rule.
+
+## Sources retrieved firsthand (2026-06-07)
+
+- DOJ/OLC opinion landing page (summary + holding):
+  justice.gov/olc/opinion/whether-president-may-sign-bill-directing-his-signature-be-affixed-it
+- CMS-855B application PDF ("signatures must be original"):
+  cms.gov/medicare/cms-forms/cms-forms/downloads/cms855b.pdf
+- 15 U.S.C. § 7001 and § 7006 (ESIGN), Cornell LII
+- UCC § 1-201 ("Signed"), Cornell LII
+- Wikipedia "Autopen" (secondary, for background + the 2011/2013/2024-2026 events
+  and the forensic even-pressure note) - corroborate against primary news before
+  client-facing use.
+
+## Caveats on this memo
+
+- This is research, **not legal advice**, and not a substitute for counsel.
+- The OLC opinion is executive-branch advice, not binding precedent.
+- The exact CMS-10114 "stamped/faxed/copied... will not be accepted" wording
+  should be re-quoted from the **current** form edition before any client-facing
+  or filing-facing use.
+- Case citations stated as "black-letter"/principle should be pinned to specific
+  controlling cases in the relevant jurisdiction by counsel before reliance.

docs/plans/remote-wet-signature-products.md

@@ -0,0 +1,127 @@
+# Remote Wet-Signature Products - opportunity map
+
+Status: brainstorm / idea capture. Not committed work. Companion legal research:
+`docs/legal/remote-mechanical-wet-signature-precedent.md`.
+
+## The core insight: the moat is "verified-original + remote + filed"
+
+The existing stack already does something most e-sign tools do not:
+
+- Online signature **capture as resolution-independent vector strokes** (not just
+  a raster image), with perjury/attestation language, JWT no-login links, IP/UA
+  + timestamp audit trail (`esign_records`).
+- **Document generation** (PDF fillers), MinIO storage, ERPNext sync.
+- **Genuine wet-ink reproduction** of the signer's own captured strokes on our
+  own hardware (CR-10 home station / Line-us portable), gated + dry-run safe.
+- **Regulatory filing + fulfillment** (paper batches, MAC routing, mailing,
+  status machine, Telegram ops alerts, admin queue).
+
+Most competitors (DocuSign, Dropbox Sign, etc.) produce an **electronic**
+signature. Our unique wedge is producing a **genuine wet-ink original from a
+remote signer** *plus* doing the hard filing. The highest-value new services are
+the ones where a plain e-signature is legally insufficient, or where the filing
+itself is the real work.
+
+> Important caveat: the wet-ink-reproduction path carries interpretive risk on
+> "original signature" rules (see the legal research doc). Lead with services
+> where electronic signatures are already accepted, and treat machine wet-ink as
+> the *fast path* with a true client-wet-sign fallback until precedent is firmer.
+
+## Tier 1 - highest value (wet-ink original + filing required)
+
+1. **"Wet signature where DocuSign is rejected" (standalone product).**
+   Generalize the CMS-855 flow: client signs online, we produce the original ink
+   document and mail/file it. Targets: recordable real-estate docs (deeds, liens,
+   releases), court filings, USCIS immigration forms, ATF firearm forms, vehicle
+   title transfers, anything an agency/bank/court still wants in ink.
+
+2. **Apostille + authentication concierge.** High willingness-to-pay, opaque/
+   painful. Client signs/uploads → we orchestrate notarization, county clerk,
+   Secretary of State apostille, and (non-Hague) embassy legalization, then ship
+   internationally. Recurring for international business, adoptions, marriages
+   abroad, foreign work permits.
+
+3. **Estate / probate document packages.** Wills, financial + medical POAs,
+   advance healthcare directives, living trusts, HIPAA authorizations. Many
+   require wet ink + witnesses + notary. We orchestrate remote signing, witness
+   coordination (RON), and produce executed originals. Large TAM, recurring
+   (people update these).
+
+4. **Lien / UCC / mechanics-lien filings.** Contractors need preliminary
+   notices, mechanics liens, and lien releases filed in specific counties on hard
+   deadlines. Signature + correct filing + deadline tracking = the exact engine.
+   Recurring per-project revenue, sticky.
+
+## Tier 2 - high value (filing/automation + signature, mostly e-sign OK)
+
+5. **Corporate governance signature-as-a-service.** Board/shareholder consents,
+   meeting minutes, operating-agreement amendments, cap-table docs, and **83(b)
+   elections** (must be mailed to the IRS within 30 days - wet ink + certified
+   mail + deadline = our sweet spot). Startups pay well, need it repeatedly.
+
+6. **Multi-party / sequential signing orchestration.** The JWT no-login link
+   model is ideal for chasing N signers in order (lender → borrower → guarantor →
+   witness). Sell the outcome ("we get everyone signed and produce the executed
+   set"), not the tool.
+
+7. **RON (Remote Online Notarization) as a layer.** Add a notary partner/
+   integration; bundle "sign + notarize + file" on top of everything above. RON
+   is legal in most states now; premium upsell.
+
+8. **Regulated-industry recurring filings (already in this lane).** Extend the
+   healthcare/telecom/DOT pattern to: SAM.gov renewals, FinCEN beneficial-
+   ownership (BOI) filings, professional license renewals, registered-agent
+   changes, foreign-qualification (registering in new states).
+
+## Tier 3 - adjacent, leverages audit + perjury + identity stack
+
+9. **Attestation / declaration service.** Penalty-of-perjury declarations, I-9,
+   proof-of-life for pensions/annuities (retirees abroad must prove they're alive
+   yearly - wet ink + notary + mail, recurring, underserved), grant/loan
+   self-certifications.
+
+10. **Consent + authorization capture at scale.** HIPAA releases, FERPA,
+    background-check consents, medical-records requests, media releases. High
+    volume; vector capture + audit trail is more defensible than a checkbox.
+
+## Prioritization (revenue × fit × moat)
+
+| Service | Stack fit | Recurring | WTP | Notes |
+|---|---|---|---|---|
+| 83(b) / IRS mailed filings | Very high (wet+mail+deadline) | Per-event, high volume | High | Clear deadline pain; startups |
+| Apostille / authentication | High | Per-event | Very high | Removes opaque friction |
+| Estate docs (POA/directives/wills) | High (wet+witness+notary) | Yes (updates) | High | Huge TAM |
+| Mechanics liens / UCC | Very high | Yes (per project) | Med-high | Deadline-driven, sticky |
+| FinCEN BOI / SAM.gov / license renewals | Very high | Yes (annual) | Medium | Pure filing-automation |
+| RON layer | Medium (partner) | Upsell | High | Multiplies everything |
+| Proof-of-life attestations | High | Yes (annual) | Medium | Underserved, recurring |
+
+## Top 3 to build next
+
+1. **83(b) election service** - smallest build; maps ~1:1 onto the existing CMS
+   pipeline (intake → generate → wet-sign → certified mail → deadline tracking).
+   Urgent, repeatable pain.
+2. **Apostille / authentication concierge** - highest WTP; remote-original +
+   shipping orchestration is the whole value.
+3. **Estate document package (POA + advance directive + HIPAA)** - biggest TAM,
+   recurring, leverages wet-ink + witness/notary orchestration.
+
+## Reusable from the current codebase
+
+- `esign_records` + JWT no-login signing links + `portal-esign-generic.ts`.
+- Vector signature capture (`signature_vector`) + perjury/attestation UI.
+- PDF generation/fillers + MinIO + presigned URLs.
+- Wet-ink pipeline (`ink_signature_plotter.py`, profiles, pens, Line-us).
+- Fulfillment: paper batches, mailing, status machine, admin queue, Telegram.
+- ERPNext order/sync, audit log, two-tier no-login model.
+
+## Open risks / dependencies
+
+- **Original-signature legal risk** for machine wet-ink (see legal doc). Keep the
+  true client-wet-sign fallback and disclose method where required.
+- **RON / notary** needs a licensed partner per state.
+- **Apostille / embassy** legalization is jurisdiction-specific and manual at the
+  edges; price for the concierge labor.
+- **UPL (unauthorized practice of law)** boundaries for estate/court docs - offer
+  document preparation + filing logistics, not legal advice; partner with
+  attorneys where the line is close.