Comparisons · PG-14 # Dated, sourced, corrections welcome We publish comparisons the way we would want to be compared: verifiable facts, a verification date on every table, links to the other side's own documentation, and a mandatory “where they win” section. ## The rules we publish by ### Verifiable facts only Every claim comes from the competitor's own docs, pricing pages or public materials — captured on a date and linked in the sources list. Vendor marketing is labelled as vendor-claimed. ### Where they win is mandatory Each comparison names the strengths of the other side. A table without concessions is an advertisement; we do not publish advertisements. ### Corrections welcome If a fact about your product is wrong or stale, tell us and we will re-verify and re-date it. The verification date is part of the page, not a footnote. ## Comparison programme | Comparison | Axis | Verification | Status | Page | |---|---|---|---|---| | Ledgerbook vs Formance | Hash-chained bi-temporal log, MIT-licensed open-source core, self-host proportions, read-only MCP | 2026-09-19 | ✓ Published | [Compare →](/compare/formance) | | Ledgerbook vs Blnk | OSS fintech ledger: chain enabled by default, immutable balances, self-hosted agent surfaces | 2026-09-19 | ✓ Published | [Compare →](/compare/blnk) | | Ledgerbook vs Crane | Restore-never-needed corrections, no-delete semantics, enterprise deployment | in preparation | ◇ Roadmap | — | | Ledgerbook vs Paprel | Cryptographic proofs, self-host and air-gap, enterprise depth | in preparation | ◇ Roadmap | — | | Ledgerbook vs Prosaic | Multi-jurisdiction, accruals depth, cryptographic evidence | in preparation | ◇ Roadmap | — | | Build vs buy | What the free option costs when audit evidence is a requirement | in preparation | ◇ Roadmap | — | | Ledgerbook vs QuickBooks | Under your books, not another app: ingestion from QBO-class estates, with no “replaces QuickBooks” claim | in preparation | ◇ Roadmap | — | | Ledgerbook vs ERP incumbents | Ingest and consolidate across SAP / Oracle / Microsoft / NetSuite-class estates; never framed as replacing them | in preparation | ◇ Roadmap | — | Sequencing note: the programme publishes in waves — Formance and Blnk carry pages today; Crane (wave 1) and the later rows are in preparation, with the order following where buyers ask first rather than a fixed queue. Rows publish only with dated, sourced cells; no row is ever filled with placeholder copy. ## The axes we compare on Seven dimensions, defined once and used on every comparison page. Cells hold only facts verified on 2026-09-19; a dash means *not verified at our bar* and is never filled from memory. Our own row carries phase tags, and the last two rows are adjacency families — close and reconciliation platforms, and planning/EPM tools — relationships rather than head-to-head rivals. | Vendor / class | Deployment | Immutability & corrections | Agent write path | Accounting depth | Provenance / audit export | Commercial shape | Where they win | |---|---|---|---|---|---|---|---| | Formance [F1] | MIT OSS core self-deploy; enterprise self-host (K8s) or private cloud; about 60% of paying customers self-host. | Append-only, hash-linked log; bi-temporal since v2 (Jun 2024); corrections as reversal entries, not mutation. | MCP is read-only and enterprise-cloud-gated (launched 2026-08-05); agent writes go through API + RBAC. | Money-movement / asset ledger; namespaced accounts; no multi-dimensional chart of accounts, no statutory posting validation, no ERP ingestion. | Hash-chained audit logs and SIEM export; no OCEL 2.1. | MIT core free; one annual enterprise subscription, ~$120K/year effective base per the AWS listing; no per-transaction metering on the base. | Regulated fintech/PSP stacks: 120+ payment connectors; SOC 2 Type II, ISO 27001 and DORA in place; a delivered enterprise self-host track record. | | Blnk [F2] | Apache-2.0 OSS deployable; Cloud tiers; enterprise on-prem / BYOC. | SQL trigger blocks updates; a SHA-256 chain exists and ships disabled until an operator enables it; balances mutate in place. | MCP routes through Blnk Cloud only — self-hosted deployments get no agent access. | Fintech ledger primitives (inflight transactions, refunds, a reconciliation engine); FX conversion was removed; no accounting dimensions. | Journal + balance snapshots; an opt-in chain-verification CLI; per-status webhooks. | $0 OSS / $100 sandbox / $1,000 production / Enterprise; ISO 27001 and GDPR posture. | Fintech dev teams wanting a complete OSS core at transparent prices; 1,100+ tests including invariant fuzzing. | | Crane Ledger [F3] | Cloud-only SaaS (Railway); no self-host. | Posted entries reverse-only; soft delete with a 7-day purge and no restore API; no hash chain. | MCP-native, about 90 tools; draft → validate → approve, with require_draft_approval and books_closed_through gates. | Hierarchical COA and categories; no first-class dimensions; multi-entity orgs; no scenarios. | Append-only event log (actor, API key, IP); per-entity history. | ≤$0.01/write metered + $1/org/month; 500 free credits; SOC 2 “ready” (not certified). | Agent-first builders wanting a fast MCP bootstrap, deep docs and machine-readable pricing. | | Paprel [F4] | SaaS-first (US/EU/APAC regions); a custom tier lists private deploy / BYOC — no self-host docs published. | Append-only policy: corrections are new entries and unbalanced writes are rejected; no cryptographic proof. | MCP-native; pending journals; tenant-scoped tokens with per-route and per-tool grants. | Embedded accounting engine (journals, multi-entity books, invoicing, payables); no disclosed multi-dimensional COA. | Append-only trail; HMAC-signed webhooks; no OCEL 2.1. | $149 / $499 tiers plus ops-metered custom; 14-day trial; ops = successful writes only. | The closest positioning match: governance-by-design pre-certification, a published OpenAPI 3.1 and a named case study. | | Prosaic [F5] | NZ SaaS only (open beta); firm-gated channel. | Journal-based writes via change-sets; no cryptographic immutability disclosed. | MCP server on their own host (OAuth 2.1 + PKCE); PENDING change-sets with admin approval and an autoApproveWrites opt-in. | Cash-basis GL; journals as the only write primitive; firm-first (GST, year-end). | Change-set audit trail; per-entity tokens; no OCEL 2.1. | $100 NZD/mo + $10/entity; APIs and MCP bundled, never metered per call; ~US$700K seed (2024); 50+ firms / 1,000+ clients claimed (Aug 2026). | NZ firms wanting BankLink-style standardisation; the best public agent change-set pattern; candid about scope limits. | | QuickBooks (Intuit) [F6] | Cloud SaaS only. | — | Official QBO MCP server (intuit/quickbooks-online-mcp-server) with documented plans for full API coverage. | SMB books of record with the largest accountant ecosystem; not a data layer. | — | Per-seat subscription plus ecosystem. | The SMB incumbent: familiarity, accountants, ecosystem, zero-infrastructure buyers. | | ERP incumbents [F7] | — (the row publishes cell-by-cell only when re-verified). | — | Agentic ERP features (Joule, Copilot, Oracle Fusion Agentic Applications) are real but bottlenecked by estate fragmentation. | Systems of entry and workflow; no neutral cross-estate ledger. | — | Enterprise platform licensing. | They are the system of record today — Ledgerbook ingests *from* them; it does not replace them. | | Close & reconciliation platforms [F8] (adjacency) | Enterprise SaaS; no self-host or air-gap published. | Match and evidence state lives above the books; corrections land in the ERP — no ledger-native typed corrections. | Governed matching automation — “AI suggests, humans confirm”; suggestions, not ledger writes. | Close workflow, matching cascades, reason codes, audit binders; not a ledger of record. | SOX evidence packs (vendor-shaped); no OCEL 2.1, no independently runnable chain verification. | Quote-led enterprise; public price points are not reliably published. | Mature close workbenches, auditor familiarity and process cadence — choose them for the close cockpit; we are the substrate their integrations read from. | | Planning / EPM tools [F9] (partners) | SaaS. | Version management and snapshots, not append-only ledgers; actuals arrive via integrations and extracts (CloudWorks-class projects reported at 6–12 weeks). | MCP surfaces emerging on the consumer side (Vena MCP; Pigment integrations) — reading ledger data, not governed writes into the books. | Modelling engines: multi-dimensional plans, drivers, allocations; no double-entry core, no per-scenario balancing invariant. | Snapshots and application audit logs; entry status is not exposed — no final-vs-provisional distinction. | Platform licensing (modules/seats); partner channel model. | Planning UX, model flexibility, adoption and ecosystem — the destination for the numbers we serve; plan quality is gated by ledger maturity, which these tools cannot see. | | Ledgerbook AI | Self-host + air-gap first-class: single node + basic HA + offline bundle PH-1; embedded class and multi-region residency ◇ Roadmap. | Chain hash on every Final entry, enabled by default, verify online/offline; corrections only as typed reversing entries PH-1; ZKP assertion contract = stub + one test vector PH-1, backend ◇ Roadmap. | MCP read/propose, same semantics and error taxonomy as HTTP, works air-gapped PH-1; full tool surface ◇ Roadmap; tiers T0/T1 PH-1, T2/T3 ◇ Roadmap; row-level access across tenant → ledger → entity → book → scenario → period → account, identical on exports PH-1. | Multi-dimensional COA; per-scenario double-entry; Provisional → Final lifecycle with typed corrections PH-1; compliance packs / statutory validation ◇ Roadmap; delta inheritance, promote and cross-scenario ops ◇ Roadmap. | OCEL 2.1 JSON export plus an independently runnable verifier; a provenance event on every mutation PH-1; evidence packages, XBRL GL / SAF-T ◇ Roadmap. | Quote-led, module SKUs, no per-transaction metering on the base; open-core licensing text TBD pre-launch. | Our own concessions, stated first: no certifications yet (targeted inside PH-2), no compliance packs or T2/T3 until PH-2, single-node scale class at PH-1 — if you need SOC 2 today, choose a certified vendor. | Sources, all fetched 2026-09-19: [F1] [formance.com/platform/deployment](https://www.formance.com/platform/deployment) · [/pricing](https://www.formance.com/pricing) · [MCP changelog](https://www.formance.com/blog/changelog/formance-mcp-server) · [AWS Marketplace listing](https://aws.amazon.com/marketplace/pp/prodview-6rjoxytk3a6lc) · [github.com/formancehq/ledger](https://github.com/formancehq/ledger) · [F2] [blnkfinance.com/pricing](https://blnkfinance.com/pricing) · [docs.blnkfinance.com (MCP, changelog, config)](https://docs.blnkfinance.com/cloud/reference/mcp) · [github.com/blnkfinance/blnk](https://github.com/blnkfinance/blnk) · [F3] craneledger.ai docs and pricing · [F4] paprel.com (product, architecture, security, pricing) · [F5] prosaic.works and developer.prosaic.works · [F6] [github.com/intuit/quickbooks-online-mcp-server](https://github.com/intuit/quickbooks-online-mcp-server) · [F7] estate analysis, same date · [F8] public close-platform materials, same date · [F9] public EPM materials, same date. Vendor marketing is labelled as vendor-claimed; the two published one-pagers carry per-cell dates and source lists. Deep dives on the adjacency rows: [N-sided reconciliation](/reconciliation) — the substrate close tooling reads from — and [FP&A consumption](/fp-and-a) — how planning tools get actuals without extracts. ## Comparison FAQ No. We are saying we are a different layer. Payment ledgers move money; we account for it — their exports are our inputs. Application vendors own the workflow; we are the ledger of record beneath it. Close platforms run the close cockpit above the books, and planning tools consume what the ledger serves. Where a competitor is genuinely ahead on an axis, the comparison says so, and we say why we are not competing there. Every row is dated and links its source; competitor marketing is quoted as vendor-claimed; each page carries a “where they win” section; and anything we have not shipped is labelled Roadmap in the same sentence as the claim. Corrections from vendors are applied and re-dated, and anything we cannot verify from public sources is marked as not verified at our bar rather than filled in. Write to us via [contact](/contact). We re-verify against your documentation, update the row and the verification date, and note the correction where it matters. A stale comparison is worse than no comparison — and the same channel takes corrections to our own column, including the facts we would rather not print. Sometimes you should — and where an OSS ledger fits, the comparison page tells you so. The honest answer is the delta: multi-entity consolidation semantics, statutory-grade immutability, OCEL 2.1 provenance and a runnable verifier, governance for agent writes, and support with an enterprise deployment story. That delta is what people pay for; the primitive itself is not the moat. ## Put your facts next to ours Design partners get a side-by-side evaluation pack — including the axes where we lose.