What's Up with AI?

Home   Events   Jobs   About Us

I'll say upfront what I'll probably repeat on this blog more than once: I think AI is currently overrated in accounting relative to the hype around it, and I think Excel fluency remains underrated. That opinion is exactly why I think it's worth tracking AI honestly here — not as a hype machine, and not as a dismissal, but as a working professional trying to figure out what's real, what's marketing, and what's actually useful. This post is where that tracking starts. How We Got Here, Briefly AI didn't arrive suddenly in 2026 — it's been building for years, but the pace has genuinely compressed. The general-purpose chatbot era (roughly 2022–2024) proved large language models could write, summarize, and answer questions convincingly. What's changed since is the shift from models that talk to models that act. That shift has a name: agentic AI. Instead of asking a model a question and reading its answer, 2026's frontier systems are increasingly built to complete multi-step tasks with real autonomy — researching, drafting, checking their own work, and in some cases running for extended stretches without a human in the loop. Anthropic, OpenAI, and Google DeepMind have all pushed in that direction, and Chinese labs (DeepSeek, Qwen, GLM) have closed the performance gap enough that some benchmarks now show leading global models separated by only a few percentage points. The competitive frontier, in other words, is no longer just "which company has the smartest model" — it's "which company's agents can actually finish a real task correctly." Pricing has also been weaponized in a way that matters for small businesses and independent professionals specifically. Cheaper models have gotten dramatically cheaper — orders of magnitude cheaper than flagship models for many tasks — which means the AI tools showing up in accounting, legal, and small-business software are no longer gated behind enterprise-only budgets the way they were even two years ago. Where the Real Adoption Is (and Isn't) Here's the part that deserves more attention than it gets: broad AI adoption and AI actually working are two different numbers, and the gap between them is the most important fact in this space right now. Roughly 88% of organizations report using AI in some form — but only about 39% report a measurable bottom-line impact from it. Separately, a widely cited pattern in enterprise AI shows the large majority of AI projects failing to deliver their intended value. That's not an argument against AI. It's an argument against assuming a tool works just because everyone's talking about it — which happens to be the exact standard I try to hold my own work to. Where AI is showing real, measurable value: in accounting specifically, firms using AI report cutting real time off the monthly close, handling meaningfully more clients per week, and in some categories cutting document-processing costs by roughly 80%. In software development, AI-assisted engineers report merging substantially more code. In customer support, a majority of service organizations have deployed AI agents and many report a strong return per dollar invested. These are not hypothetical use cases — they're operational ones, in categories close to what accountants and finance professionals do every day. Where it's still mostly hype, or at least unproven: general "AI will run your business" claims, and consumer-facing "AI advisor" tools that give confident-sounding advice without the accountability a licensed professional carries. That distinction — assistive tool versus unsupervised advisor — is going to come up again and again on this blog, because it's the exact line I think matters most for accounting, law, and medicine specifically. The Uncomfortable Part There's a labor story underneath the technology story that's worth naming honestly rather than glossing over. In 2026, entry-level hiring in some AI-adjacent fields has taken a real hit — software developer employment for workers in their early twenties has reportedly fallen sharply, even as senior engineers using AI tools report a real productivity boost. That's not a hypothetical future risk. It's happening now, and it's a preview of a conversation the accounting and legal professions need to have before it happens to them, not after. I don't think the honest answer to that is panic, and I don't think it's denial either. It's specificity: which tasks are genuinely automatable versus which require judgment, accountability, and a license — and building your own skills toward the second category deliberately, rather than hoping the first category doesn't come for you. What This Blog Will Track Going forward, this space will follow three things, plainly and without the sales pitch: what AI tools are actually being adopted in accounting, finance, and adjacent professional fields; what's separating the tools that deliver real value from the ones riding the hype cycle; and what that means for professionals trying to stay relevant rather than replaced. I'll be direct when something is overhyped, and I'll be direct when something genuinely works — including the tools I end up recommending through the1accountant4u down the line. That's the whole point of writing this under my own name instead of a faceless brand account.