Every close, you log into client bank and credit-card portals by hand to download the statements QuickBooks won’t auto-fetch — especially the small banks and credit unions Plaid can’t connect. Twenty clients across thirty institutions is thirty separate logins, every single month. It’s the real job nobody bills for: document logistics, not accounting.
We automate that one step for you, on your own computer. Not in our cloud. Not with your client passwords on our servers. The automation runs inside your own Chrome, riding the login session you already have — and there’s no AI anywhere in it.
| Pilot | Retainer | |
|---|---|---|
| Price | Free | $99/month, flat |
| Institutions | 2 | up to 10 (more: ask) |
| Duration | 2 weeks | monthly, cancel anytime |
| Setup | screen-share call | included |
| Breakage fixes | included | included, 48h response |
No contract, no per-statement fees, no surprise usage billing. The automation files are plain JSON on your machine — yours forever, even if you cancel.
Tools like Hubdoc, LedgerDocs Bank Fetching, or other cloud fetchers do something similar — by storing your clients’ bank passwords on their servers and logging in from their cloud. Banks are actively killing that: credentials in a datacenter, behind a clean datacenter IP, is exactly the pattern their bot-detection now device-blocks, and it’s often a terms-of-service problem.
Taprun is built the other way around. The automation runs on your machine, in your browser, with your own login — so to the bank it looks like you, the returning human, because it is. Credentials never leave your computer — that’s architecture, not a policy promise. And the replay is deterministic (a fixed, auditable program — not AI improvising clicks on a live bank account), fully logged, so you have a clean record for your E&O file.
You already pay $47–93/month for cyber liability cover because you hold client bank logins. Keeping those logins on your own machine — never in anyone’s cloud — is the point.
For statements you need the same result every month, traceable line by line — not a model that might quietly mis-key a number or “decide” something. Taprun doesn’t fetch by guessing transcripts or reading the screen with AI; it replays the exact recorded steps. Nothing learns, nothing trains on your client data, nothing improvises on a bank account.
Email hello@taprun.dev with two client banks you’d like off your plate at month-end. We’ll reply within 24 hours with a setup-call link.
Prefer to build it yourself? Taprun’s engine is free and the spec is MIT — see the homepage and the general done-for-you portal pulls page. The service is for people who’d rather buy the outcome than the tool.
How do I get client bank statements QuickBooks won’t fetch? QuickBooks’ bank feed only connects ~3,000 US institutions through Plaid/Finicity; small banks and credit unions on cores without an API simply aren’t covered, so QuickBooks tells you to upload them manually. The remaining options are: ask the client every month (slow), log into each portal yourself (tedious), or have the login-and-download step automated on your own machine — which is what this service does.
Is it safe to give a cloud tool my client’s bank login? Storing a client’s bank password on a third-party server is a security and terms-of-service risk, and it’s the exact pattern banks are now blocking. The safer pattern is to keep the login on your own machine and automate locally — the credential never crosses to anyone’s cloud.
Can I download bank statements into QuickBooks automatically without AI? Yes. This is a deterministic recorded automation — it replays the same login → download steps every month, fully logged, with no AI model reading the screen or making decisions.
What about small banks and credit unions Plaid can’t connect? Those are exactly the institutions this is built for. If a human can log in and download the statement, the automation can replay that on your machine — no API required.