One in four new accounts comes back not in good order.

Every account a firm opens can be rejected by the custodian for a missing field, a bad document, an unresolved signer. It's the #1 operational friction in wealth management — and it lands on the client.

SmartIQ & multiple vendors (manual NIGO 25–31%); Plaid 2025 (ACATS rejection 30–50%); industry onboarding surveys 2025.

The pain is acute right now.

• Advisors are moving firms in record numbers — the Commonwealth→LPL diaspora alone is re-papering tens of thousands of accounts.
• Custodian consolidation concentrated the rules; learn one registry (NFS) and you cover the bulk of the BD market.
• AI finally makes upfront validation feasible — propose the answers, then verify every one against deterministic rules.
• Reg S-P (2024) raised the bar on data handling and breach scoping — firms need provable, gated workflows, not PDFs.

Everyone cleans up NIGOs after rejection.
We gate the flow before submission.

Garbage in, garbage out — so we don't let garbage in.

AI proposes. The rules engine disposes.

One authoritative, fail-closed checkpoint on every path — running identically in the browser as you work and on the server as the source of truth.

• Capability matrix — 39 registration types, transcribed from the real NFS custodian registry.
• Fail-closed validation — deterministic rules; an unknown type defaults to deny, never to a permissive fallback.
• Upfront routing — holds, reviews, and screening fire off the data, before submission.
• Drift — every later change re-runs the same gate before it posts back. In good order stays in good order.

We tore down the incumbent. It only claims to be the gate.

We read a production account-opening system's code. Its 61 validation rules are advisory — wired into no submit path, hard-blocking in exactly two places. We built the gate it claimed to be.

A working gate on real registry data.

• Entity-of-entity resolved to who signs — beneficial owners computed by multiplying ownership through sub-entities; CDD ≥ 25%, enforced.
• One record, reused — members entered once, applied to accounts by role. No duplicate entry.
• Change re-runs the gate — flip a client to non-US and every one of their accounts is held for AML review before it can post.
• App-level encrypted PII + an append-only, field-level audit for 17a-4.

Not just new accounts — the whole back office.

Account opening, cashiering, address changes, transfers — every request lands in the right queue, gets checked, and only goes out in good order. Need a new kind of work tracked? A firm adds a queue itself, in minutes.

Do more with less.
Scale the book, not the back office.

The AI era rewards firms that grow without growing headcount. Account opening and maintenance is the back office — manual, error-prone, and the bulk of operations time. Sluice automates the fail-closed gate, so the same team clears far more volume.

Grow AUM and accounts without growing the back office. That's the margin story — and why this lands now.

Docupace (manual ≈ 60% of staff time; ~65% effort reduction); industry NIGO-labor estimates.

A concentrated, underserved buyer.

NFS-first reaches the bulk of the BD market through one registry. Incumbents are enterprise-priced and backend-only — Docupace (~$50–100M rev), Fenergo ($1.16B val), FNZ/Appway. None gate upstream.

FINRA YE2025; SEC/IAA 2025; Cerulli; onboarding-software market syntheses 2023–2025.

We built the real thing first.

Sluice is built by operators who ran a production account-opening system — we've seen the 35-type matrix, the entity bubble-up, and exactly where the incumbent fails. That domain truth is the moat: a generic engineering team can't reverse-engineer the rules; they have to live them. AI proposes; we already know what “in good order” means.

SaaS per firm. Never a transaction fee.

Per-firm + per-seat subscription, white-labelable to the firm's brand. We don't tax the flow — we clean it. Expansion is natural: open → maintain → onboard → reconcile to the book of record, each a module on the same gate.

• Land on the onboarding desk; expand to maintenance + the CRM/book-of-record reconciliation.
• Custodian-pluggable: NFS first, then Pershing / Schwab / Apex on the same engine.

Working product. Now: design partners.

• Shipped — a live, gated product on real NFS registry data: the gate, entity resolution, the relationship/record model, re-gate-on-change, encrypted PII + audit.
• Now — opening 3–5 design-partner pilots in the hottest vein (Commonwealth→LPL diaspora; Kestra / Cambridge / &Partners). [confirm signed pilots]
• Next 12–18 mo — convert pilots to paid; add Pershing; ship the maintenance + CRM-reconciliation modules.