Intuit Just Let AI Send Client Invoices Without a Human Click

Adam spent 20 minutes looking for a $36 receipt. His finance team sent three Slack messages. Someone made a sticky note.

Ramp would have matched it automatically the moment he swiped. Auto-coded, in-policy, synced. Nobody had to ask Adam for anything.

This is what finance looks like when it runs itself.

Your team can be Adam. Or they can not be Adam.

Don’t be like Adam

Intuit Lets AI Send Invoices Unattended

Summary: QuickBooks now lets small business owners create, review, and send invoices or purchase orders through Claude or ChatGPT conversations, moving past the old read-only integration. Why it matters: A client can now generate and dispatch a real invoice without you — or anyone — reviewing it first. Catch: Intuit is describing its own feature rollout here, and there's no independent testing yet of how often AI-drafted invoices contain errors a human would have caught. Unknowns: Pricing for the expanded access and what error-handling or review process sits behind it aren't in Intuit's announcement — worth pressing your QuickBooks rep on before you tell a client to turn this on. My take: This is worth testing now if you're already deep in the QuickBooks ecosystem — better to find the failure mode on a dummy file than have a client find it on a real one. It's not a reason to panic, but it is the same loose-guardrails problem this issue's Heads Up lead digs into, and firms that wait for Intuit to add a review step will be the last to know it's missing. [Read more →]

Clients Value Trust Over AI Speed — Karbon survey data (350 U.S. business owners) show trust is the #1 reason clients pick an accountant, cited by 39% — more than any other factor, on the way to a +49 NPS score. Clients aren't AI-resistant either: 89% just want transparency on where it's used, and 81% say their accountant is more valuable when AI handles routine work and the saved time goes into a strategy call. Why it matters: If your firm markets itself as AI-powered, lead with judgment and relationship for the advisory side of your pitch — the data backs touting AI speed on the routine/compliance side, since that's exactly where clients say they want it freed up. [Read more →]

QuickBooks Rebuilds Its Banking Center — A practitioner review of the redesigned QuickBooks Online Banking Center found real improvements to transaction review, categorization, matching, and reconciliation, alongside a few rough edges still being worked out — most notably, the new AI categorization starts learning from scratch on existing files and takes about two months of daily use to reach ~90% accuracy. Why it matters: If you're doing bookkeeping cleanup this week, don't judge the new interface's AI suggestions on day one — re-test your categorization rules and give the AI a few weeks to relearn the client's patterns before assuming your old shortcuts still work. Also confirm your firm's QBO plan has actually received the update, since Intuit is rolling it out gradually. [Read more →]

The AI Filed the Return. So What Exactly Are We Charging For? — A well-known accounting AI evangelist let an agent file a client's return start-to-finish with almost no manual intervention, and it worked — while a separate industry survey the same month had AI leads at major firms saying autonomous agents "aren't there yet." The author's test for telling the two apart: run a tool with zero human review, and if the output would've differed materially from what you'd have filed, it's assisted, not autonomous, no matter what the sales page claims. Why it matters: Apply that same test to whatever AI tool you're already running before you decide what to bill for it — the tools doing genuine unattended work are rarer than the marketing suggests, and knowing which bucket yours falls in should drive whether you're pricing by the hour or by the outcome. [Read more →]

ProSeries' Built-In Planner Goes Underused — A 35-year CPA and longtime ProSeries user walks through practical uses of the built-in tax planner, including instantly modeling a client's updated capital gains scenario. Why it matters: If you're already paying for ProSeries, test the planner on one client this week before pitching tax planning as a separate, higher-fee service from prep. [Read more →]

Before you recommend the QuickBooks AI invoicing feature to anyone, send yourself a test invoice through it first. Watch exactly what data it pulls, what it gets wrong, and whether there's a review step you can force before send — you want to know the failure mode before a client finds it for you. [Read more →]

Speak your prompts into ChatGPT or Claude and get detailed, paste-ready input that actually gives you useful output. Wispr Flow captures what you'd cut when typing. Free on Mac, Windows, and iPhone.

Try Wispr Flow free

Governance: Firms Adopt AI Faster Than They Govern It: Thomson Reuters research finds tax and accounting firms have rolled out AI tools faster than they've built governance around them, with the firm framing the next phase as reining in unsanctioned staff use while keeping approved tools accountable. This isn't a distant compliance exercise — if staff at your firm are already running client data through AI tools nobody signed off on, that's an active, uncontrolled channel for confidential information right now. It connects directly to this issue's Featured Tool: the same absence of a review step that lets an AI send an invoice unattended is what lets ungoverned tool use spread inside a firm. Why it matters: Audit which AI tools your staff are actually using this week, not just the ones on your approved list, before a client or regulator asks the question for you. [Read more →]

Audit Readiness: Confidence Outpaces Reality: Schellman's "State of AI Governance Report 2026," surveying 525 US professionals, found most organizations believe they'd pass a compliance audit on their AI governance, though only about a quarter call their program fully mature. Why it matters: When a client tells you their AI governance is audit-ready, assume that confidence is overstated and ask to see the actual documentation, not just hear the assurance. [Read more →]

Skills: Directing AI Beats Delegating to It: A joint KPMG/UT Austin field study of 500+ early-career professionals found employees with nearly identical skills got very different results from AI depending on whether they directed and refined its output or simply handed off the task. Why it matters: Train junior staff to actively question and refine AI output as its own skill, assuming tool access alone won't produce good work on its own. [Read more →]

Every story this week points at the same gap: the tools are moving faster than anyone's rules for using them. That gap is where your value as the CPA in the room actually goes up, not down.

—Alex

You don't have to scroll every AI thread, track every new tool, or watch every demo. 

The Rundown AI breaks it all down for you — the latest AI news, tools, and tutorials in one free 5-minute email every morning. 

Trusted by 2M+ professionals at Apple, Google, and NASA.

Sign up to start learning.

📬 Know an accountant drowning in manual bookkeeping? Forward this.