Your AI stack is useless
until your accounting
is usable.

The average enterprise lease accounting platform costs $25,000/year (source: Vendr/Finantrix 2026 data) — and revenue recognition typically requires a separate vendor at similar cost. Big 4 managed services run $50K–$200K/year. Ledger Layer Starter is $49/month ($588/year) and includes both lease and revenue modules. Professional is $499/month ($5,988/year), bundles both modules, and automatically includes every new module added to the roadmap (IFRS 9, ASC 326 / CECL, hedge accounting, IFRS 17, share-based payment, taxes, business combinations, and more) at no extra cost. Same deterministic engine. Same audit trail. Same hash-verified output. No implementation project. No per-seat pricing. No per-agent pricing.

No. AI can interpret and suggest, but it cannot guarantee deterministic, audit-grade outputs. Language models are probabilistic — they produce plausible answers, not verified ones. Ledger Layer separates interpretation from computation: AI extracts and normalises data, but all accounting results are produced by a deterministic engine with human approval. This separation is enforced architecturally, not just by convention.

No — and this is the most important thing about Ledger Layer. The AI layer extracts and normalises data from your inputs only. All accounting calculations — present value, amortisation, revenue recognition, journal entry amounts — are performed exclusively by the deterministic engine. The AI has zero access to any compute function. Your auditors can rely on every number.

Yes. Ledger Layer accepts Excel uploads directly. The AI reads the structure of your workbook, extracts lease and contract terms, and routes them to the engine. You review and confirm before anything is calculated. Bulk import of 200 leases completes in under 60 seconds.

Once approved, a journal entry is immutable — no delete, no edit. The correct treatment is an explicit reversal JE, which itself goes through the full approval workflow. This mirrors standard accounting practice and produces a clean audit trail.

The engine supports the full range of modification events: extensions, early and partial terminations, remeasurements for variable-in-substance payments, and rate changes — all per IFRS 16 and ASC 842 prescribed treatment. Each modification triggers a remeasurement run from the modification date, producing updated schedules and modification-event journal entries.

None. FX and IBR rates live in dedicated tables. Every calculation traces back to a specific persisted rate row — effective date, source, and who uploaded it. The IBR matrix supports currency × term bucket × effective date across unlimited currencies.

Yes. Ledger Layer exposes a full MCP server with 75+ schema-validated tools. Your agent connects via Bearer token with the same role-gated access as a human user. The key distinction: the agent can propose actions, but it cannot approve journal entries — that authority always stays with a human.

Yes — Mode 3 (Self-Hosted). Docker image, your infrastructure, your local AI provider. The deterministic engine runs identically in all three deployment modes. No data leaves your network. Available on the Enterprise plan.

Four roles, each with distinct authority:

OWNER — Full workspace control. Manages billing, users, and all approval authority.
ADMIN — Can approve journal entries, manage entities, and configure workspace settings. Cannot manage billing.
PREPARER — Can upload data, run the engine, and draft journal entries. Cannot approve their own work.
REVIEWER — Read-only access to portfolio, journals, disclosures, and audit trail. Cannot modify data.

Same roles apply to human users and AI agents. An agent with a PREPARER token can extract and compute but cannot approve. Enforced at the API layer. MFA is enforced for ADMIN and OWNER.