Ledger Lens - every transaction categorized, every categorization on record
Month end starts with a stack of client bank statements and an accountant reading them line by line: interpret, categorize, key it in, next line. Ledger Lens, built on Blue Mesh, does the reading and the categorizing, pauses on the lines it should not decide alone, and writes down why every line went where it went. Your people start where the judgment is.
From statement stack to reviewed ledger
The workflow is a sequence of agents on a canvas, and the sequence matters: extraction before categorization, categorization before review, and a log entry at every step. Here is what each stage actually does.
Intake and extraction
Statements arrive as banks send them, PDF and Excel, one layout per institution. Ledger Lens extracts the lines into structured records, per client and account.
- 01.01PDF and Excel statement upload
- 01.02Line-level extraction into structured records
- 01.03Multi-client, multi-account handling
- 01.04Every extraction logged
Categorization with reasons
Models categorize each transaction against the categories your firm configures. A line that cannot be placed cleanly is flagged as an exception, not forced.
- 02.01Categories your firm configures
- 02.02Exception flagging instead of forced matches
- 02.03Reason recorded on every categorization
- 02.04Consistent treatment across the whole batch
Review, export, and the record
Flagged lines queue for an accountant, and nothing posts past that gate unreviewed. Approved output exports to the books with its record attached.
- 03.01Exceptions reviewed by a person before posting
- 03.02Export to books and workpapers
- 03.03Approvals attributed and timestamped
- 03.04The full trail retained under your rules
Modern bookkeeping, with the working shown
Accounting firms categorize client statements by hand. It delays bookkeeping, invites errors, and a query months later finds no record of why.
Blue Mesh powers Ledger Lens. Teams upload PDF or Excel, models categorize and flag exceptions for an accountant to rule on, and every step is logged.
The stack still arrives at month end, but the firm's people start at the exceptions, not line one. How a figure was produced is on record.
Built for work that gets audited
An audit trail built for the audit
Every decision and result is logged for compliance audit: the line that was read, the category it was assigned, the reason, the flag, the approval. Financial records answer to auditors and tax authorities, and what they ask for is exactly this record.
Exceptions stop and wait
The workflow pauses at a human checkpoint before anything ambiguous posts. That is a limit built into how it runs, not a feature to admire: an accountant rules on every flagged line, and the ruling is logged with their name on it.
Client data stays in your environment
A client's statements are their financial data, not material to lend a vendor. The platform deploys private cloud, on-premises, air-gapped, so the statements are read where your controls already apply.
The boundary, stated
Ledger Lens categorizes and documents. It does not exercise accounting judgment: exceptions go to your people, the treatment of a hard line is theirs to decide, and the accountant signs the books. A firm that wants software to make those calls unsupervised is not what this is for.
- BM Studio Where the statement workflow is assembled: intake, extraction, and categorization as agents on a canvas, with an approval step before anything posts.
- Trust and security Where client statements sit while the platform reads them: private cloud, on-premises, air-gapped, with the controls named rather than asserted.
- How we compare Fifty platforms on what each one publishes, plus six questions worth putting to every vendor on your shortlist.
Bring one client's statements
One client, one month, your categories. We will run the workflow live and show you the record it leaves behind: every line, every exception, every approval.