Pleadwright vs. the three alternatives creditor-side counsel actually evaluates.
Side-by-side, across the same six observable facts /trust and /methodology already attest to — statute-aware rules, fail-closed gates, clickable Why? rationale, versioned citations, no-migration pilots, and citation-ready audit packets.
Mentally walk each row top-to-bottom. Read the Pleadwright column against the three alternatives a partner would consider — full-time partner review of every consumer-facing message, a generic LLM drafting tool with no encoded FDCPA / TCPA / validation / Mini-Miranda rule table, and a legacy on-premise collection workflow suite. The cells name the statutes so a countering partner sees exactly where the audit trail starts.
Axis · Pleadwright vs. alternatives
Fingerprint pending · counsel-sign-off envelope incomplete
| Axis | Pleadwright | Partner review of every consumer-facing message | Generic LLM drafting tool | Legacy on-premise collection suite |
|---|---|---|---|---|
№ 01 whyPleadrwright.proofs.0.title | Encoded FDCPA (§1692c, §1692e, §1692g), TCPA (§227), Reg F (§1006.34), and §1692e(11) Mini-Miranda — every rule shipped under a versioned citation. | Partner reads the letter and signs off. Coverage depends on which partner picked up the file that morning; re-derivation is anecdotal. | The model drafts in plain prose. FDCPA, TCPA, validation, and Mini-Miranda are not in the rule table; they appear, when they appear, as prompt instructions. | Ships with a static rule pack the vendor last updated in 20XX. No per-rule version, no per-rule reviewer, no clickable cite back to the codified registry. |
№ 02 whyPleadrwright.proofs.1.title | Deterministic 5-state gate (Permitted · Human Review · Blocked · Missing Evidence · Waiting for Approval). Fail-closed on bankruptcy stay (§362), active representation, open dispute, or a closed SOL window. | Partner intuition catches the worst cases. Borderline matters — soft SOL, recent dispute, ambiguous FDCPA carve-out — sail through on partner bandwidth alone. | No gate taxonomy. A "stop" appears when the human catches a problem on review. State-space is the drafting tool's 0-shot typo rate, not codified law. | Workflow rules treat validation, dispute, and representation as workflow-fields, not as hard-stops. The suite routes "until counsel responds." Counsel does not always respond. |
№ 03 whyPleadrwright.proofs.2.title | Every emitted decision carries a clickable Why? payload that threads to (a) the authority line — rule id + triggered rule text + codified registry entry — and (b) the evidence row — the matter-file fields the rule read. | Partner rationale is defensible — but live in the partner's head. Re-derivation = ask the partner to re-read the matter file and remember why they approved it. | Model output exposes a confidence score and the prompt. Neither is the rule the letter cites; neither is the evidence row the rule read. | Workflow log shows the template name + revision id. The link from template to codified §1692g text is in the vendor's admin tool, not the audit packet. |
№ 04 whyPleadrwright.proofs.3.title | Every rule ships with a versioned citation: statute + section + codified registry entry id. A reviewer can trace a citation back to the worksheet that signed it. | The partner signs; the signature is the citation. There is no worksheet; there is no entry id; there is no diff between this letter and last month's letter signed by a different partner. | The model's training corpus included §1692g. The version of the model is opaque; the cite it offers back is what the corpus happened to contain at training time. | Versioning exists at the template level. The statutory citation embedded in the template rarely updates without a vendor release; release notes are not a citation trail. |
№ 05 whyPleadrwright.proofs.4.title | In a pilot or sandbox, reads synthetic or redacted matter files and keeps generated artifacts in a sandboxed audit vault. No production write-back, no shadow database to migrate, and no connector promotion is implied. | No migration; the matter file already lives on the firm of record's drive. The cost is partner-hours, not infrastructure — but at full partner rate. | OAuth into the matter store, or copy-paste into the prompt. Either way: the matter file is now also a prompt history the vendor retains for safety / tuning. | On-prem install + dual-write reconciliation + a parallel intake queue. Months of vendor services; the matter file lives in two places for the duration of the rollout. |
№ 06 whyPleadrwright.proofs.5.title | Every evaluation writes a timestamped row to the audit trail: rule_version, policy_version, triggered rule id, matter-file snapshot, and a 64-char hex decision hash. Replay the same input six months from now → same hash. The trail is citation-ready. | Audit packet is the partner's signed letter. Replay is whatever the partner can remember at deposition. Six-month replay not possible without the partner's testimony. | Audit trail is the prompt + the response + the token ledger. There is no rule version, no policy version, and no decision hash — the same input today and six months from now returns different prose. | Workflow-level logs cover user actions. Statutory version, decision hash, and the link back to the rule that fired are not rows in the export. |
Reviewed under the same counsel-sign-off envelope /methodology and /trust attest to. Open the methodology fingerprint for the literal reviewer name, signature date, and rule-table version.
Evidence incomplete
Hand this comparison to a counterparty.
After a live demo, the operator walks the same six axes against the alternatives the buyer's counsel already evaluated. The matrix below renders identically in the presentation and in the post-walk email; cite the row, click through to the proof, and the audit trail is one click further.