For Web3 Finance Teams
Stop spending hours every week reconciling on-chain activity against your books. GhostWork builds the first version of a 3-way recon workflow across wallets, accounting system, and bank data, with exceptions surfaced for human review.
Your team should be doing finance, not data assembly. The pilot gives you a working recon surface on real data, then a runbook your team can keep improving.
Manual treasury recon is a tax on your team
Every hour spent exporting, matching, and chasing discrepancies is an hour not spent on decisions that matter.
Wallets multiply. Reconciliation doesn't scale.
One Ethereum mainnet wallet was manageable. Now you have wallets across 6 chains, a Safe multisig, a CEX account, and a custodian. Each one needs to map to a line in your books. Nobody on your team has time to do this right.
Month-end closes take days instead of hours.
By the time you've exported CSVs, converted token prices at the right historical rate, and matched transactions against your accounting system, you've lost a week. And there's always something that doesn't reconcile.
Mismatches surface at the worst time.
A mismatch shows up in an audit or a board report, not your monthly recon run. Because the run was manual, incomplete, and nobody had a clean exception queue. You fix it under pressure.
How it works
A scoped Phase 1 build on your real treasury data. We map the recon workflow, build the pipeline, and hand over the runbook and source materials.
Step 01
Connect your sources
We map the sources that matter first: wallets via RPC or indexer, accounting data from QuickBooks, Xero, NetSuite, or exports, and bank or custodian feeds where available.
Step 02
Recon logic gets built
A focused automation ingests on-chain events, converts them to functional currency using agreed price sources, and matches them against your ledger entries.
Step 03
Mismatches surface immediately
When a transaction doesn't match, the agent flags it with context: the raw event, the expected entry, and what's missing. You get a clean exception list, not a spreadsheet to diff yourself.
Step 04
Close package gets easier to review
Reports are generated from the same source data. Your team gets transaction-level detail, price sources, and reconciled balances in one place instead of assembling every tab manually.
What the pilot delivers
$5,000 flat for the 30-day scoped pilot. Hardening and handoff happen inside the same pilot once the first surface is defined.
3-way recon running on your real data
On-chain, off-chain, and bank: all three sources matched against each other on your real data, not sample data.
Historical price reconciliation
Token transactions converted through agreed price sources at the relevant timestamp or block. Fewer spreadsheet assumptions, more traceable inputs.
Review-ready reporting
Monthly close packages in a format your finance team can review and adapt for auditors. Transaction-level detail, not only summary exports.
Exception triage workflow
When something doesn't match, you get context: the gap, the explanation, and what to do next. Not a diff to figure out yourself.
The pilot is deliberately narrow
Treasury data gets messy fast. The first engagement is not an enterprise transformation project. It is a fixed-scope build that proves one recon surface on real data.
If the pilot exposes a broader finance automation roadmap, we scope that separately after the first working system is in front of your team.
Common questions
Book a 20-minute recon audit
Tell us what you have: chains, wallets, accounting system. We'll tell you what the recon pipeline looks like for your setup and what it would take to automate it.
Book a Recon AuditTreasury recon is one of our most common pilot surfaces. See the Ghost Ops Pilot for a 30-day automation pilot with a flat fee.