For the past year and a half, I've watched AI change what's possible in our profession, and in my own life.
It's fundamentally changed how I see the future.
So I wrote down what I actually believe: where I think we're headed, and what I think we should do about it.
Whether you're a CPA or someone who's wondered if you've already missed the boat on this stuff, I hope you'll read it.
https://t.co/jiwXDPeDZX
@ChrisHervochon yea It's solid but I'm spoiled from claude. But it sounds like we are in the same boat and working to solve the same problem. Ill DM you.
@ChrisHervochon Sick. Would love to connect on this. I’ve been running Qwen2.5‑14B locally on my 32gb MacBook, and while it's impressive, the limited context window and the speed are an issue. I'm considering upgrading my hardware to run larger local models as well.
Seven weeks ago, this was just a tweet on #TaxTwitter asking if any accountants wanted to talk AI weekly... now we have so many cool resources for you!
Come say hi Wednesday at 2pm ET.
https://t.co/h1k6pTYouh
“If you do instantly what others do eventually, you’ll be successful.”
As a solo CPA, I think about this all the time while everyone around me screams, “AI will replace CPAs.”
Yea a lot of them won’t make it.
The bigger risk is clearly the CPA who waits too long to understand how to use it.
“But AI makes mistakes!”
So do I. So do you. It’s ok. Learn from those mistakes.
“These models aren’t secure!”
Neither is blindly trusting every piece of software, portal, spreadsheet, or email workflow you already use.
The answer isn’t to ignore AI or dismiss it.
The answer is to learn where it helps, where it fails, and how to build better systems around it.
CPAs who do that now will have a massive advantage over the ones who wait until they’re forced to catch up or retire.
Introducing GLM-5.2: Frontier Intelligence, Open Weights
- Significant improvements in coding and agentic tasks
- Strong long-horizon capabilities with a 1M context window
- Two levels of reasoning effort: GLM-5.2 (max) pushes the limits, while GLM-5.2 (high) strikes a strong balance between performance and token efficiency
- MIT-licensed open weights
- Same API pricing as GLM-5.1
Tech Blog: https://t.co/LAsxUdN0JZ
Weights: https://t.co/g0A1C4UWx4
API: https://t.co/Kc3E22cbN7
Coding Plan: https://t.co/Nk8Y98HNhU
Chat: https://t.co/WCqWT0qCQb
@ITmarko@ITmarko it really is and is why I'm advocating hard for clearer authoritative guidance, because right now it feels like we're all trying to figure it out alone.
@CPATaxTeam In this case I actually wasn't working with any specific client information, I was building out workflows and internal tools. Agents are great at building, but following what they built and the why is the challenging part IMO.
The more work I offload to AI, the further I position myself from the source knowledge.
When I delegate execution to agents, I lose the implicit context that naturally comes from being the one doing the work. Because of this distance, I am finding that my review process takes longer than it did in the past.
My focus now is on building clear audit trails and making the review process as easy and reviewable as possible.
Easier said than done, but switching outputs from md files to html has helped a lot.
Everyone wants tax work to be a vending machine: put in $10k, get $10k back.
But, that’s not how it always works.
The best money you spend here is the kind you can’t see on an invoice.
It’s the disaster that quietly never happened. It’s the landmine you never stepped on. It’s the stress you never carried because someone else was watching the field.
You don’t pay for what shows up on the return. You pay for what doesn’t.
My Hermes agent sent me this for my morning quote:
"Alice had it right. Some mornings you have to run as fast as you can just to stay in the same spot. Winning the morning in an accelerating world isn't about the perfect routine. It's the boring, repeatable systems that keep you from falling behind before the day even starts."
I definitely feel like Alice in today’s fast moving world of Ai. I feel like I have to move as fast as I can just to keep up.
Just got to keep on keeping on.
Anyways, back to work.
I've been speaking to some lovely accountants who are in love with the mission at Open Accountants and they've requested to verify skills related to GAAP and IFRS, which is something I am on board for. So from next week we'll not only have tax skills on the site but also GAAP and IFRS related skills verified by experts in the field.
Thank you to the accountants that are so mission driven to make this happen.
If your an accountant and would like to contribute hit me up, the opportunity for you (besides ensuring that LLMs get the info it spits out right) is that you get free AI based SEO. We're seeing instances where an LLM refers the accountant who verified the skills in question for follow ups and further advice.
I hope we can all collectively take the industry into the age of AI safely, it requires a group effort.