TreasuryFlow
Product
Cash positionThe morning number, every bank 13-week forecastUpload yours, we keep it live AI General LedgerBooks built from the bank IntegrationsBanks, books, ops, delivery
Solutions
For firmsFractional CFOs & bookkeepers IndustriesConstruction to public agencies
Pricing
Resources
Blog Free tools Monday Cash Close template Help center Customer story
Sign in See the demo Start free trial
Sign in See the demo Start free trial
Legal

Privacy Policy

Last updated: November 12, 2026 · what changed

TreasuryFlow ("we", "our", "us") is operated by Pantoll Ventures LLC. This Privacy Policy explains how we collect, use, disclose, and safeguard your information when you use our service.

1. Information We Collect

Account Information: When you sign up, we collect your email address and optional full name. We generate a hashed API key for authentication; we never store the raw key.

Financial Data (via Plaid): When you connect a bank account, Plaid transmits transaction data to our servers. We process this data using our privacy-first architecture:

  • We store exact transaction amounts so we can show them to you and the people and tools you authorize, in places such as your dashboard and exports, your morning email, the Slack answers you ask for, and an AI assistant you connect
  • Our AI categorization sees each amount only as a privacy-preserving magnitude bucket (Micro, Small, Medium, Large, XLarge), never the exact figure
  • We store merchant names, transaction dates, direction (inflow/outflow), amounts, and magnitude categories
  • We never store bank account numbers or routing numbers
  • Our application logs, error reports and product analytics are designed to exclude the exact amounts and balances from your connected accounts. Where a log line or error report has to describe a figure, it records only a size range or whether it is positive or negative

Billing Information: Payment processing is handled entirely by Stripe. We store your Stripe customer ID and subscription status. We never see or store your credit card details.

2. How We Use Your Information

  • To provide the TreasuryFlow service: transaction categorization, cash flow forecasting, variance analysis, bank fee analysis, working capital metrics, runway projections, and vendor intelligence
  • To generate AI-powered insights (forecasts, scenario analysis, the treasury assistant, and AR narratives) using Google Gemini. The assistant and these narratives are sent your actual figures (amounts, balances, and merchant or customer names) so they can answer questions about specific transactions. Google does not use what we send it to train its models. The categorization engine, by contrast, only ever sees magnitude ranges, never exact amounts.
  • To authenticate your API requests via hashed API key lookup
  • To manage your subscription and billing through Stripe
  • To send transactional communications (account creation, billing events)

3. Privacy-First Architecture

TreasuryFlow is built on a privacy-first principle. Your exact figures appear only where you and the people and tools you authorize use them, such as your dashboard and exports, your morning email, the Slack answers you ask for, and an AI assistant you connect. Section 4 lists the services that receive them. Our internal systems are designed to minimize exposure:

  • Our application logs, error reports and product analytics are designed to exclude the exact amounts and balances from your connected accounts. Where a log line or error report has to describe a figure, it records only a size range or whether it is positive or negative
  • Our AI categorization works on magnitude buckets, not exact amounts. The treasury assistant and our written summaries and insights, by contrast, are sent your actual figures (see Section 4)

Where data has been deidentified (for example, amounts reduced to magnitude buckets), we maintain and use it only in deidentified form and never attempt to re-identify it.

De-identified data and our own models. TreasuryFlow may use de-identified, aggregated data to train and improve its own models, for example, to improve how transactions are categorized. This is separate from the AI features in Sections 2 and 4, which use your data only to answer you. Before any data is used this way, we apply this standard:

  • We remove names, email addresses, and bank account and routing numbers
  • We remove any merchant or counterparty name that could tie a record to your business
  • We remove free-text memos and notes
  • We replace each amount with its magnitude bucket (Micro under $10, Small $10 to under $100, Medium $100 to under $1,000, Large $1,000 to under $10,000, XLarge $10,000 and over), never the exact figure
  • We combine records from many businesses, so no single business or person can be identified

We never sell this data, never use it to train another company's foundation models, and never share it with another customer. Data we obtain from a provider whose terms forbid it, including Xero, is never used for training.

4. Third-Party Services

These service providers process your data on our behalf. Here is what each one receives:

  • Plaid: Bank account linking and read-only transaction data. See Plaid's Privacy Policy
  • Stripe: Payment processing. We never see or store your card details. See Stripe's Privacy Policy
  • Google Cloud Platform: Hosting infrastructure: our application servers and encrypted databases run on Google Cloud. See Google Cloud Privacy
  • Google Gemini: AI features, run through Google Cloud (Vertex AI). For transaction categorization, Gemini receives transaction descriptions and merchant names with amounts bucketed into magnitude ranges (never exact amounts) and runs deterministically (temperature 0.1). The treasury assistant and narrative features are sent the actual figures described in Section 2 so they can answer questions about specific transactions. Features that read a photo, file, or email for you (for example, a receipt photo posted in Slack) send it to Gemini to read. We never send the bank account numbers or credentials we hold. A photo, file, or email you ask us to read is sent as it is. Google does not use your data to train its models. See Google Cloud's Privacy Notice
  • Anthropic (Claude): Internal operations. We use Claude in our own operations tooling, at the founder's direction, to investigate and maintain the service. It may read production data; by our policy it makes no change to customer data without the founder's approval. While it does this work, it can read the account data involved, such as your transactions and balances. It does not power any feature in the product. We use it under Anthropic's consumer terms with the “Help improve Claude” setting turned off on our account. Anthropic has no data processing agreement with us, and processing may occur in the United States and other regions. See Anthropic's Privacy Policy
  • Resend: Email delivery: receives your email address and the content of the emails we send you (for example, your morning cash brief). When an owner turns on invoice reminders, Resend also processes the billing contact's email address and the reminder text we send on the owner's behalf. See Resend's Privacy Policy
  • Sentry: Error monitoring: receives technical error reports so we can fix failures. Our error reports are designed to exclude the exact amounts and balances from your connected accounts (see Section 3). See Sentry's Privacy Policy

You choose whether to connect the services below. Each one is governed by that provider's own terms and privacy policy. Here is what we read from each one, or send to it:

  • Google sign-in: Optional. If you sign in with Google, we keep your name, email address, and Google account ID. We don't keep the other profile details Google shares, such as your photo.
  • Intuit QuickBooks: Optional. If you connect QuickBooks Online, we sync your invoices, bills, and accounting data into TreasuryFlow, and we write to your QuickBooks company only when you or someone you have authorized on your account explicitly posts an entry. See Intuit's Privacy Statement
  • Xero: Optional. If you connect Xero, we sync your accounting data read-only. See Xero's Privacy Notice
  • FreshBooks: Optional. If you connect FreshBooks, we read the email address on your FreshBooks login and the businesses it can see, your invoices, your clients, and your payments, to show who owes you money. We never change anything in FreshBooks.
  • Square: Optional. If you connect Square, we read your sales and settlement data to reconcile it against your bank deposits. See Square's Privacy Notice
  • Aspire: Optional. If you connect Aspire, we read your jobs, estimates, invoices, payments, and bank deposits; your properties and their collection notes; crew hours with their labor rates and costs; your project managers' names and email addresses; and the contact details of the customers on those jobs. We never change anything in Aspire.
  • CallRail: Optional. If you connect CallRail, we read your calls and form submissions, including each caller's name, phone number, email address, and any address they enter, and how they found you (source, campaign, search keywords, and landing page). We never change anything in CallRail.
  • Monday.com: Optional. If you connect Monday.com, we read the status and value columns of the one board you choose. We never change anything in Monday.com.
  • Toast: Optional. If you connect Toast or upload a Toast export, we read your orders, fees, refunds, and any Toast Capital repayments.
  • Eventbrite: Optional. If you connect Eventbrite, we read your organization name, events, ticket sales, and expected payouts.
  • Tripleseat: Optional. If you connect Tripleseat, we read your bookings, deposits, and balances due. The booking records we keep can include your clients' contact details.
  • Google Workspace (Gmail): Optional, available to some accounts. If your Google administrator gives us read-only access, we read the mailboxes you choose and keep bill details (such as vendor, amount, and due date) and the subject and participants of email threads about your money and jobs. For some of those threads, Google Gemini writes a short summary of the latest message. We do not keep the emails themselves. We can also list the names of the spreadsheets in your Google Drive so you can choose one, never their contents.
  • Slack: Optional. If you add the TreasuryFlow app to your Slack workspace, or give us a Slack webhook address for alerts, The TreasuryFlow Slack app below says what we read from Slack, what we keep, and what we send. See Slack's Privacy Policy.
  • AI assistants you connect (such as Claude): Optional. If you connect an AI assistant such as Claude to TreasuryFlow, it can read your TreasuryFlow figures through our read-only connector: cash and recent bank transactions (with account names and their last four digits), your forecast and upcoming debt payments, receivables by customer, vendor spend, jobs, project manager check-ins, marketing results by channel, your morning brief, your Today list, and bank connection status. For firm accounts, it can also read a summary across the client businesses you have access to. The assistant's provider's terms govern what it does with them.
  • Services you connect with an access token: Optional. Our integrations page lists the other services you can connect by giving us an access token, such as HubSpot, Shopify, and Meta Ads. For each one you connect, we read a fixed set of records, read-only, and keep them in your account.

Google Ads (purchase measurement): We advertise on Google. When you pay for a subscription, the confirmation page tells Google Ads that a purchase happened, so we can measure our advertising. It sends the amount you paid us, a reference for the payment that does not name you, and, if you came to us from one of our ads, Google's identifier for that click, which we keep in a cookie in your browser. Google's code may also set its own cookie on that page. As with any web request, Google also receives your IP address and browser details. This happens once, on that page only; no other TreasuryFlow page loads Google's advertising code. We never send Google Ads your email address, your bank or accounting data, or anything from your connected accounts, and we turn off ad personalization for it. See Google's Privacy Policy.

The TreasuryFlow Slack app

If you add the TreasuryFlow app to a Slack workspace, this is what we read from Slack, what we keep, and what we send to Slack.

  • What Slack sends us: Slack sends us the messages in the direct messages and channels the app is in, and notices such as the app being removed. We use only the messages sent to the app (direct messages, mentions of the app, and the /treasuryflow command), the asker's own earlier messages in the same thread, what people enter in TreasuryFlow's buttons and forms (such as a project manager's check-in, which we save to the job), and posts in a receipts channel you choose. We ignore all other messages and do not store them. For the events we act on, we keep Slack's event ID, without the message, so we never act on the same event twice.
  • Workspace members: When the app is added, and each time it checks who is linked, we read the workspace's member list from Slack. For each member whose Slack email matches someone with access to your TreasuryFlow account (you, your teammates, your bookkeeper, or your project managers), we keep their Slack ID, workspace ID, email address, and role, so that TreasuryFlow answers and messages only the people you have given access, and you can see who is linked. For every other member we keep only their Slack ID, never their email address, so you can see how many members are not linked. When a member who is not linked asks the app a question, we read their Slack name and email address and show them to the account owner in a direct message, so the owner can choose whether to invite them; we keep only a keyed fingerprint of that address, so an invite can go only to the address the owner saw. With the app's install we keep the workspace's name and ID, the receipts channels you choose, and, when the app was added from Slack, the Slack ID of the person who added it. A firm's install also keeps which Slack channel belongs to which client.
  • Questions and answers: To write an answer, we send the question, earlier questions and answers in the same thread, and the figures from your account it needs to Google Gemini (see Google Gemini above). Quick cash answers and reports are built from your saved figures without Gemini. So that a follow-up in the same thread has context, we keep each question and our answer, encrypted. A daily clean-up deletes them once the thread's last answer is a day old, so they are usually gone within two days. They are also deleted when the app is removed from the workspace and when you delete your account.
  • Receipts: In a receipts channel you choose, we read each photo, any text posted with it, and any reply in its thread that names a job. Google Gemini reads the photo. We keep only the details we read (such as the vendor, date, total, card's last four digits, and job) and the Slack reference to the photo, not the image or the message. Nothing is changed in QuickBooks or Aspire.
  • What we post in Slack: Answers, setup messages, and the other messages you set up or ask for, such as replies to receipt photos, bills sent to your bookkeeper, a direct message to the account owner when someone without access asks a question, project manager check-ins, and reminders to your bookkeeper and project managers about work waiting on them. An answer to a mention or to the /treasuryflow command is shown only to the person who asked. Reports you ask for go only to your own direct message. We do not answer in Slack Connect channels shared with another organization.
  • Sign in with Slack: If you add the app from Slack, Slack tells us your name, your email address and whether Slack has verified it, and your time zone. We use them to create your TreasuryFlow account or to link your existing one (an existing account, or an email Slack has not verified, is confirmed first with a code we email to it), and we revoke the sign-in token Slack gives us as soon as we have read them. An account created this way gets no scheduled email until you turn the morning email on, from the button in your Slack welcome message or in Settings. An existing account that links Slack keeps its email settings.
  • The app's Slack token: We store it encrypted. When the app is removed from the workspace or Slack revokes its token, we discard the token. When you delete your TreasuryFlow account, we revoke the token with Slack and delete your workspace links, saved questions and answers, and receipt records.
  • Slack alerts by webhook: Separately from the app, if you give us a Slack webhook address, we post alerts to it, such as low cash, possible duplicate charges, jobs heading for a loss, and broken connections.

5. Data Retention & Deletion

We keep your data for as long as your account exists. Canceling your subscription does not delete your account or your data. To delete them, use Close my account in Settings: it cancels billing, revokes bank access, and deletes your data. If a connection such as QuickBooks or Xero has to be closed by hand first, we tell you and finish the deletion within 30 days. You can also email privacy@treasuryflow.ai: we delete your data from our live database within 30 days of confirming that the request comes from you. Deleted data ages out of our backups within about 14 further days. Billing records held by Stripe are kept by Stripe as the law requires.

6. Security

We implement industry-standard security measures including:

  • API key authentication with one-way hashing of stored tokens
  • Rate limiting on public endpoints
  • CORS origin restrictions
  • HTTPS-only communication
  • Application logs and error reports designed to exclude the exact amounts and balances from your connected accounts

7. Your Rights

You have the right to:

  • Access your stored data through your TreasuryFlow account dashboard
  • Request deletion of your account and all associated data
  • Disconnect your bank account at any time

8. Contact

For privacy-related inquiries, contact us at privacy@treasuryflow.ai.

9. Changelog

  • November 12, 2026: Section 4 now has its own part on the TreasuryFlow Slack app: what Slack sends us and what we use, what we keep about workspace members and installs, what an access request shows the account owner, what we send to Google Gemini to write an answer, how long we keep a thread's questions and answers, what we keep from receipt photos, what we post in Slack, what Sign in with Slack gives us, and what happens to the app's token. It corrects two statements in the earlier Slack entry: we keep the email address only of the members linked to your account, not of every member, and an account created through Sign in with Slack gets no scheduled email until you turn the morning email on.
  • November 10, 2026: Section 4 adds Google Ads purchase measurement. When you pay for a subscription, the confirmation page reports the purchase to Google Ads: the amount paid, a payment reference that does not name you, and Google's click identifier if you came from one of our ads. No other page loads Google's advertising code, and no email address or financial data is sent. This replaces the September 2, 2026 statement that we run no Google Ads conversion tracking on any surface.
  • October 26, 2026: Section 3 now says TreasuryFlow may use de-identified, aggregated data to train and improve its own models, and sets out the standard we apply first: names, email addresses, account and routing numbers, identifying merchant and counterparty names, and free-text memos are removed, amounts become magnitude buckets, and records are combined across many businesses. This data is never sold, never used to train another company's foundation models, and never shared with another customer, and data from a provider whose terms forbid training, including Xero, is never used for it. The earlier text said we do not train any TreasuryFlow models on your data. Sections 2 and 4 now say plainly that it is Google that does not train on what we send to Gemini. Section 4 adds Anthropic (Claude), which we use in our internal operations tooling at the founder's direction, under Anthropic's consumer terms with model training turned off and without a data processing agreement.
  • October 26, 2026: Section 4 now says that when an owner turns on invoice reminders, Resend also processes the billing contact's email address and the reminder text we send on the owner's behalf.
  • October 20, 2026: Sections 1 and 3 now say where your exact figures appear (such as your dashboard and exports, your morning email, the Slack answers you ask for, and an AI assistant you connect) instead of saying they reach only an Excel ledger, that it is our AI categorization that sees magnitude buckets, and that our application logs, error reports and product analytics are designed to exclude the exact amounts and balances from your connected accounts; Sections 4 and 6 say the same. Section 4 now separates the providers that process your data for us from the services you choose to connect, and adds the services you can connect today that it left out: Google sign-in, FreshBooks, Aspire, CallRail, Monday.com, Toast, Eventbrite, Tripleseat, Google Workspace, the Slack app, AI assistants such as Claude, and services connected with an access token. It also says Gemini runs through Google Cloud and that a photo, file, or email you ask us to read is sent to it as it is. Section 5 now describes deletion as it works today: canceling a subscription does not delete your data, and you delete your account in Settings or by emailing us. The earlier text said data was deleted within 30 days of cancellation; nothing did that, so we corrected it.
  • September 2, 2026: Removed the Google Ads entry from Section 4. We no longer run Google Ads conversion tracking or Enhanced Conversions on any surface, and no hashed email addresses are sent to Google. Website measurement is first-party only.
  • July 5, 2026: Listed every service that processes your data (added Google Cloud Platform, Intuit QuickBooks, Xero, Square, Resend, and Sentry to Section 4) and clarified exactly what data Google Gemini receives for categorization versus the assistant.
  • May 22, 2026: Added the Google Ads conversion tracking and Enhanced Conversions disclosure (one-way hashed email addresses).
  • March 23, 2026: First published version.
The CFO morning brief, weekly.

Cash, runway, and benchmark insights from finance teams running TreasuryFlow.

Start free trial See the product tour
TreasuryFlow

Finance work done for owner-run businesses of any kind. It connects to your bank and your books.

$299/mo per company · 90-day free trial · Live in 5 minutes Read-only bank access via Plaid
Product
Features AI General Ledger Pricing Product tour Free tools For firms
Industries
Construction Landscaping Restaurants Property managers Multi-entity Public agencies
Compare
vs QuickBooks vs Float vs Mercury vs Trovata All comparisons
Company
About Blog Help center Status
© 2026 Pantoll Ventures LLC · 35 Miller Ave #1112, Mill Valley, CA 94941 · TreasuryFlow™
QuickBooks is a registered trademark of Intuit Inc. TreasuryFlow is not affiliated with or endorsed by Intuit.
Help center Status Contact Security Privacy Terms