ARNew toolAdded Oct 17, 2026
QuickBooks now has an agent that chases unpaid invoices
Intuit's August 2026 QuickBooks update added a Payments Agent that follows up on unpaid invoices for you, alongside Intuit Intelligence, which handles routine bookkeeping and flags shifts in cash flow.
Why it matters: If you're on QuickBooks, you may already pay for automated collections follow-up and not be using it.
Try thisOpen QuickBooks and check whether the Payments Agent and automatic invoice reminders are available on your plan. Turn reminders on for 3 days before and 1 day after the due date.
Tool: QuickBooks · Source: Self Employed
SpendNew toolAdded Oct 17, 2026
Ask Claude or ChatGPT about company spend, then act on it
Payhawk's Fall '26 release connects its spend platform to Claude and ChatGPT. You can ask questions about company spend in plain English and then code, approve or chase items from the chat, with each change previewed before it happens.
Why it matters: This is where finance software is heading: you ask the AI you already use, instead of logging into one more system to build a report.
Try thisWhatever spend tool you use, ask its vendor whether it offers an MCP connector for Claude or ChatGPT. Then try: "Which vendors' spend rose more than 20% versus last quarter, and who approved it?"
Tool: Payhawk · Source: Payhawk
CloseNew toolAdded Oct 17, 2026
A general ledger whose agents do the bank recs and accruals
Digits, a 2026 CPA Practice Advisor Innovation Award winner, is an AI-native ledger whose agents handle bank reconciliations and accrual schedules and escalate only the items that need professional judgment.
Why it matters: The close work that eats the most hours is exactly what these agents target. The 'escalate only judgment items' design is the control to look for.
Try thisList your five most time-consuming close tasks and the hours each takes. That's the scorecard to bring to any close-automation demo.
Tool: Digits · Source: CPA Practice Advisor
AuditNew toolAdded Oct 17, 2026
AI agents that run your SOX tests on the whole population
Bead AI, another 2026 Innovation Award winner, uses AI agents to execute a team's existing SOX testing plans, evaluate many kinds of evidence, and test full populations instead of samples.
Why it matters: Sampling exists because people are expensive. When testing every transaction costs about the same as testing 25, the audit conversation changes.
Try thisPick one control you sample today (for example, JE approvals). Ask: what would we learn if we tested 100% of them every month?
Tool: Bead AI · Source: CPA Practice Advisor
TaxNew toolAdded Oct 17, 2026
An agent that schedules and confirms tax payments in 60+ jurisdictions
Remitian uses agentic AI to schedule, format, route and confirm tax payments across more than 60 jurisdictions, with human approval and high-risk payments routed to review.
Why it matters: Multi-state payment calendars are pure tracking work with real penalties for a miss. A human approves; the agent does the remembering.
Try thisExport your tax payment calendar for the next 90 days. Count the jurisdictions and due dates. That number is your business case.
Tool: Remitian · Source: CPA Practice Advisor
SpendNew toolAdded Oct 17, 2026
A controller agent that chases receipts inside Microsoft Teams
Payhawk's Financial Controller Agent works inside Teams: it chases missing receipts, handles submissions and approvals, and matches several receipts at once.
Why it matters: Receipt chasing is the most hated job in finance. Putting it where employees already chat gets receipts back faster than another email.
Try thisCount last month's card transactions still missing receipts on day 5 of the close. That's the metric to beat.
Tool: Payhawk · Source: Payhawk
CloseHow they do itAdded Oct 17, 2026
Before you let AI close the books, write down your policies
On a QuickBooks accountants' podcast, Pilot's CEO described a human-reviewed autonomous close: define your policies and checklist first, feed complete source data, let the platform run the close, then review a summary of what it did and what it couldn't. The discussion also covered how small errors survive for years without review, from a $23 payroll liability carried for five years to a revenue schedule that overstated revenue for two.
Why it matters: The lesson isn't the tool. It's that AI only closes well when your policies are written down and someone still reviews with skepticism.
Try thisWrite your close policies on one page: capitalization threshold, accrual method, bad debt approach, allocation rules. Any AI close tool will need it, and so will your next new hire.
Tool: Pilot · Source: Earmark CPE
CloseNew toolAdded Oct 17, 2026
Continuous anomaly checks instead of a month-end scramble
Canopy's Close Automation continuously flags uncategorized expenses, duplicate transactions and other ledger issues during the month, rather than waiting for the close.
Why it matters: Finding a duplicate on the 8th is a two-minute fix. Finding it during the close is a delay.
Try thisAsk Claude to review this week's transaction export for duplicates (same vendor, amount and date) and uncategorized items. Run it every Friday.
Tool: Canopy · Source: CPA Practice Advisor
CloseIdeaAdded Oct 17, 2026
Turn a spreadsheet you rebuild every month into an agent
One of the more practical agent patterns: take an existing working paper (a prepaid schedule, an accrual rollforward, a payroll tie-out) and have AI turn it into a repeatable process that reruns each period from fresh data.
Why it matters: You don't need a new system. The process is already documented in your spreadsheet; AI can make it run itself.
Try thisPrompt: "Here is my prepaid amortization workbook and last month's GL detail. Document the steps I follow each month, then build a process that rebuilds the schedule from new GL data and flags anything that doesn't tie."
Tool: Any AI assistant · Source: Ledge
Summaries are ours; follow the source link for the original. Listing a tool isn't an endorsement, and no vendor pays to be included. Educational content, not professional advice.
