AI Twins, Coding Dominance.
A BNY CIO built an AI twin of herself. Claude Code is dominating startup dev. Here's what both mean for accounting firm owners right now.
A BNY CIO Built an AI Twin of Herself — Finance Leaders Are Next
A BNY executive built an AI digital twin trained on her communication style, using it to prep speeches and — according to her — hold nightly strategy conversations with herself. This isn't a novelty act. When C-suite executives start offloading judgment and communication to a model that sounds and thinks like them, the question for accounting firm owners becomes: who is actually advising the client? The billable relationship has always been built on trust in a specific person. That assumption is getting stress-tested.
If your value is your judgment, you need to know what happens when clients can rent a version of that judgment for $20/month.Claude Code Is Eating Startup Development — Your Firm's Automation Backlog Has No Excuse
A survey of 20+ startup founders and VCs found that Claude Code has become the dominant AI coding tool in that world — not because it's perfect, but because it ships working code fast enough to matter. For accounting firms, this is directly relevant: the internal tools you've been putting off — custom reconciliation scripts, client onboarding automations, month-end checklists that run themselves — are now buildable by a senior accountant willing to spend a weekend with the right tool. You don't need to hire a developer. You need one person on your team who's willing to learn prompt-driven development.
Identify one manual process your team repeats every month and assign someone to prototype an automation with Claude Code before your next all-hands.Google BigQuery's Cross-Engine Iceberg Support Solves the Multi-Source Data Problem
If you're pulling client data from QuickBooks, a payroll system, and a banking API into any kind of reporting pipeline, data format mismatches are probably costing your team hours every close cycle. Google Cloud's new cross-engine Apache Iceberg support in BigQuery lets teams read and write the same data tables across different engines without conversion — meaning cleaner, faster pipelines when your data lives in more than one place. This is infrastructure-level news, but the practical impact is fewer broken exports and less manual reconciliation of data that should already agree.
If your close cycle involves manually stitching together exports from three or more systems, this is the category of tooling that eliminates that problem permanently.The Billable Hour Isn't Dying — It's Being Hollowed Out From the Middle
Everyone keeps saying AI will kill the billable hour. That's wrong — and it's distracting firm owners from the real threat. What's actually happening is that mid-tier work — the $150-$250/hour stuff that junior and mid-level staff do — is being compressed, not eliminated. A BNY executive can now rehearse a board presentation with an AI twin. A startup founder can ship internal tooling without an engineer. The work isn't disappearing; the hours required to do it are shrinking, and clients are starting to notice the math. Firms that haven't restructured their pricing around outcomes instead of time are going to feel this by Q1 2027.
Your clients aren't going to tell you they think they're overpaying for hours that AI is quietly cutting in half — they'll just switch firms.