The Nexairi Dispatch · Monday, May 11, 2026 · Issue #13
PCAOB puts audit AI on the standards clock
The regulator flagged audit tech as a research priority. Standards are coming — document now.
Good morning, friends. Two regulators moved last week: the PCAOB put audit AI on its formal research agenda and the SEC proposed cutting quarterly filing requirements in half, which sounds like less pressure until you realize internal close discipline is much harder to maintain without an external deadline. Digits moved fixed asset schedules inside the general ledger, eliminating one of the most common sources of month-end reconciliation errors. And Google quietly made Gemini available to every Workspace accounting firm for free — most just haven't turned it on yet.
📊 ACCOUNTING — Digits Moves Fixed Asset Schedules Into the Ledger
What happened: Digits launched Schedules, a feature that moves fixed asset depreciation and prepaid expense tracking directly into the general ledger. The system detects which transactions need a schedule, drafts the amortization math and holds the entry for accountant approval before anything posts.
Why it matters: Most month-end close errors in fixed asset schedules start because the schedule and the ledger don't talk to each other. Moving this workflow inside the accounting system closes that gap and gives reviewers a clear source trail. Firms doing high-volume close work stand to recover real time on reconciliation.
What to watch: Whether Xero and QuickBooks build equivalent embedded schedule management. This kind of feature tends to spread across mid-market platforms within 12 months of a smaller vendor proving the concept.
⚖️ REGULATION — PCAOB Puts Audit AI on Its Research Agenda
What happened: The PCAOB listed audit technology as a formal research priority on May 5. The regulator is examining how firms use AI tools during audits, with plans to develop standards covering tool selection, supervision and testing.
Why it matters: This is the first clear signal that audit technology will get its own standards — not just fall under existing quality control rules. Any firm using AI in audit work without documentation of how those tools were chosen and supervised is now behind where standards are heading.
What to watch: The PCAOB's research-to-standard timeline typically runs 18 to 24 months. Firms that build documentation practices now will be ahead of the formal requirement rather than scrambling to meet it.
💼 FINANCE — SEC Wants to Let Companies Report Twice a Year
What happened: The SEC proposed on May 5 to let public companies switch from quarterly to semiannual external reporting. The change would be optional — companies that prefer quarterly filings could keep them.
Why it matters: Fewer external deadlines don't mean less internal close work. Boards, lenders and operations still need quarterly data. When the pressure of a regulatory deadline disappears, internal reporting becomes the only thing maintaining close discipline — and internal deadlines are easier to slip.
What to watch: Whether early adopters who switch to semiannual reporting see finance team productivity gains or experience drift in internal reporting quality within the first 12 months.
🤖 AI TOOLS — Google Gemini Is Already in Your Accounting Firm
What happened: Google added Gemini AI features to every Business and Enterprise Workspace plan at no added charge. For accounting firms already on Google Workspace, that means AI built into Docs, Sheets, Gmail and Drive — no new software purchase needed.
Why it matters: Gemini can draft engagement letters from your existing templates and prior client emails, categorize expenses in Sheets and route invoice attachments into a tracking spreadsheet automatically. Workspace Studio lets staff build no-code workflows, and the AI Control Center governs which tools each person can access. Most firms haven't activated any of them.
What to watch: The window to get ahead of peers closes as adoption normalizes. Firms that test and deploy one Gemini workflow this quarter will be ahead of clients who will expect AI-assisted service as a baseline.
Outside Nexairi
Maryland Residents Billed $2B for Out-of-State AI Data Centers — Tom's Hardware
Maryland is fighting a $2 billion power grid upgrade cost that state regulators say should fall on AI hyperscalers, not residents. The case is before federal energy regulators and raises a direct question about who pays when AI infrastructure scales.
OpenAI Fires Back in Week 2 of the Musk Trial — MIT Technology Review
OpenAI's president testified that Musk initially pushed the company toward a for-profit structure and sought near-total control. Week two shifted the narrative from Musk's grievances to OpenAI's account of what he wanted and didn't get.
DOL Launches a Free Text-Message AI Literacy Course for Workers — Princeton CITP
The Department of Labor partnered with Arist to offer AI basics through a text message course — no app or subscription required. A Princeton review finds the program a useful first step but limited in depth for workers who need practical AI skills.
Most Teams See 10 to 20% Gains From AI, Not the Promised 10x — sshh.io
Poor planning, unchanged handoffs and no skill-building investment cap most teams at 10 to 20% AI productivity gains. Getting to 2x or more requires restructuring how people work, not just adding tools to existing processes.
Tool Worth Knowing: AgentPeek
AgentPeek puts a live view of Claude Code and Codex sessions in your Mac menu bar. For teams running AI automation agents in the background, it cuts the context-switch cost of checking whether they're done.
Deeper Read
You Need AI That Reduces Maintenance Costs — jamesshore.com
James Shore's case for evaluating AI tools not on initial output quality but on whether they reduce long-term maintenance burdens. For any firm adopting AI where the hidden cost is ongoing correction and oversight, this is the right frame.
Local AI Needs to Be the Norm — unix.foo
The case for running AI models locally rather than sending data to cloud providers. For accounting and legal firms with sensitive client data, this is a practical data governance argument worth sitting with.
Quick Hits
- PS3 emulator team asks developers to stop submitting AI-generated pull requests
- Task Paralysis and AI — why more AI access sometimes means less gets done
- AI malaise: MIT Technology Review on why uncertainty is growing as AI goes mainstream
- Local AI as a privacy default — why cloud AI has a data exposure problem
- Touchless Payroll: how to evaluate vendor AI claims before you switch
- Karbon's The Loft: why AI training needs to be a firm-wide benchmark
- One AI lab could dominate enterprise stacks by 2027 — the concentration risk for CFOs