Lately my company has been doing a lot of complex accounting and reporting in spreadsheets. Overall was surprised by how well both GPT and Claude handled some of these extremely tedious tasks. Not uncommon to have an hours-long task compressed to minutes.
My anecdotal experience is GPT 5.2 Pro is decently ahead of Claude Opus 4.5 in this category when it gets to the tricky stuff, both in presentation and accuracy. The long reasoning seems to help a lot. But, apparently the benchmarks do not agree.
The deterministic part (calculations) is done by Excel.
The non-deterministic part is turning human instructions ("calculate the NPV over 10 years for X given Y") into Excel.
This is already a non-deterministic process (humans are non-deterministic!). The question is if an AI model can be more reliable than humans, and I can't see any reason why it wouldn't be.
The correct path is pretty clear, so the logits for following that path are going to be a long way from off-path.
For something like this the real problem is training the model to use Excel (which will show up by it being confused which sheet it is on or trying to use the wrong window or things like that), not the non-determinism.
so basically what you're saying is that: it doesn't do the math, it tells the math-doing-thing what math to do. Basically, instead of humans using Excel, imagine AI using Excel?
Yet I don't understand the aha moment here? It might save analyst time but aren't there already enough automation that you don't really need to tell the AI to tell the math-doing-thing to do the math because the math-doing-thing is already optimized for most general functions? What are we gaining from adding the non-deterministic process here when the real non-deterministic process is still the human being prompting what to do?
Seems like a solution to a non-problem from my pov.
Sure in the sense that you're setting up a program that has inputs and outputs etc. But then all math is programming. All language is too, even speaking can be considered programming if you're stretching the definition enough. But I will disagree that setting up spreadsheets = basically software engineering.
The LLM just needs to make sure it uses it appropriately. Doing that bit is the non-deterministic part, but the NPV calculation itself is completely deterministic.
When you setup a spreadsheet, you choose a column or cell to show the NPV. You use the npv formula on this cell and apply it to inputs from other columns or cells.
This process of deciding what data to put in what column and how to apply NPV to it is non-deterministic. You could choose to put it in column A, or column B or maybe even row 3 - it depends!
Do you really think you come off as a good person posting things like this? This place is to satisfy intellectual curiosity, not to lord your supposed intelligence over others, without even proving it exists in the first place.
Just a reminder that an accountant who might say "I use Currency format" is still working in binary floating point
as the format is just used as a display mask. And using VBA macros with the Currency type will hit problems at the boundary, when values move between the worksheet and the macro. The tool is broken in a way that proper accounting software is not...
Afaik Excel is wicked fast at math (can handle ginormous spreadsheets, some real magic under the hood), while correctly doing decimal based math (1/5 is 0.2 exactly, not some floating point monstrosity)
I just hope the divine powers of Luca Pacioli might help you, before is too late. A whole set of accounting bodies and real world disasters expressly advise against its use and spreadsheets in general.
Also... Excel does not use exact decimal arithmetic. It stores numbers in IEEE-754 binary floating point,
which cannot precisely represent many decimal values used in accounting. Microsoft documents this explicitly:
– results depend on formula structure and rounding order
That alone disqualifies it as Accounting ledger. Historically you have lots of
famous examples of bad outcomes of treating Excel as financial infrastructure...
Large audit firms also publish entire risk frameworks, ( and I am sure PWC does also...) explaining why uncontrolled spreadsheets are a major source of financial misstatement in the first place.
This entire site is full of idiots confidently spouting nonsense. Probably time to move on. "Don't use Excel for accounting" has to be the dumbest thing I've read in a year, even with all of the LLM bullshit going on.
And that's including the people this morning who are apparently excited for LLMs to file their taxes.
I've been having the same feelings lately, especially around the AI doomers coming in with some weird observations that make no sense. I usually don't comment, but seeing a lot of these types of overgeneralized responses.
You don't really deserve a response, but even the idea that you could tell someone to not use Excel for accounting and seriously project that you know anything about how accounting works in the real world is hilarious.
You could think 500 random internet articles - it doesn't matter. The assertion is still ridiculous. You literally will not be able to get a job as an accountant without being able to use Excel. Don't be an idiot.
You can also link to hundreds of articles showing that dynamic types cause errors. People don't care, they still use them when it makes sense to use them. There is no better mainstream alternative for accounting than Excel right now, period.
No, not a single reply, because your statement wasn't "Excel can cause problems sometimes". That's a different conversation. You said "Don't use Excel for accounting", which is something you could only say if you have no idea at all about how accounting works in the real world.
I'm not arguing about whether or not Excel is perfect, or if it causes problems, or if it makes demons fly out of your nose. I am educating you about the real world and how it's important to understand that no matter how "well akshually" technically correct you are, it doesn't matter at all if the advice you're giving is WRONG and not useful to people.
So, I repeat. Idiots confidently spouting nonsense.
There's a big difference between geeks and hippies. Too many of the former here and not enough of the latter here - people who can understand other people, culture etc and then think about how technology can fit into ones life.
It's shocking and honestly pathetic. "Hey guys the software you use is not accurate" being met with such derision is how we regress as a society, not move forward.
"This entire site is full of idiots confidently spouting nonsense."
Too right - this goes for the arrogant geeks at the within the orgs of LLM producers who still havent displaced call center workers with their technology.
Comical stuff. Really is.
The lack of disconnect with what people actually do in jobs is absolutely breathtaking.
> Too right - this goes for the arrogant geeks at the within the orgs of LLM producers who still havent displaced call center workers with their technology.
Some have shares in these AI companies and need to off-load them either on secondaries or IPO.
These are the clear incentives to show performative results to give the impression that you can replace accountants with AI without any of the risks involved.
The benchmarks look good. Slide decks and spreadsheets look better. The people must use Claude Cowork and have their Claude Code moment and figure out the consequences. It will be really interesting to see articles like this (https://mitchellh.com/writing/my-ai-adoption-journey) written by people who actually care about accuracy in places like KPMG to get their perspective on things.
I remember over hearing some normal people on the bus talking about essentially orchestrating some agent scraper to pull and summarise news from 40 different sites he identified as important which put him quite ahead of his peers. These were non-technical people orchestrating an agent workflow to make them better at work.
Though there’s not much that tickles my software brain here. But the agents are coming for us all.
Based on the article... is this basically just making Claude better at formatting and data presentation, or does it also get better at analysis? I get the impression it's the former.
> The side-by-side outputs below show how output quality has improved from Claude Opus 4.5 to Opus 4.6.
Disclaimer: I use AI to code (and I code for finance) and I love Anthropic.
But: for f-ck's sake, I cannot click on the picture and have it show up in full. It stays at its tiny size, impossible to read the numbers. I had to right-click and "open in a new tab".
AI is, somehow, definitely still not fully there yet.
And then you hand it to your boss who takes a 20 second look at it and asks why you made a projection that assume massive revenue growth and 3 years of perfectly flat utilities, insurance, G&A - no inflation etc.
It does look really promising as a skeleton starting point though. Like generate it, delete numbers and populate by hand.
Not unlike the boilerplate start we saw in AI coding a couple years back
You'll use a ton of AI but it won't wipe the humans out. In the end you'll have a compositional change, likely nothing catastrophic imo. In part because there is a buck to stop and Claude ain't got no hands...
Anthropic does anything to keep the Claude hype going; from fearmongering ("AI bad, need government regulations") to wishful thinking ("90% of code will be written by AI by the end of 2025" —Dario) to using Claude in applications it has no business being in (Cowork, accessing all your files, what could go wrong?) to releasing "research" papers every now and then to show how their AI "almost got out" and they stopped it (again, to show their models are "just that good") to prescribing what the society should do to adapt to the new reality to doing worthless surveys on "how AI is reshaping economy, but mostly our AI not others".
Now this is going to be interesting to watch to see if the finance bros financing this AI wave to get rid of SW engineers will keep financing getting rid of their own.
Lately my company has been doing a lot of complex accounting and reporting in spreadsheets. Overall was surprised by how well both GPT and Claude handled some of these extremely tedious tasks. Not uncommon to have an hours-long task compressed to minutes.
My anecdotal experience is GPT 5.2 Pro is decently ahead of Claude Opus 4.5 in this category when it gets to the tricky stuff, both in presentation and accuracy. The long reasoning seems to help a lot. But, apparently the benchmarks do not agree.
Edit - noticed OpenAI specifically focuses on finance use cases in their gpt-5.3-codex blog as well https://openai.com/index/introducing-gpt-5-3-codex/
I feel like I'd be really skeptical of results from a non-deterministic model for something as precise as accounting....
The deterministic part (calculations) is done by Excel.
The non-deterministic part is turning human instructions ("calculate the NPV over 10 years for X given Y") into Excel.
This is already a non-deterministic process (humans are non-deterministic!). The question is if an AI model can be more reliable than humans, and I can't see any reason why it wouldn't be.
The correct path is pretty clear, so the logits for following that path are going to be a long way from off-path.
For something like this the real problem is training the model to use Excel (which will show up by it being confused which sheet it is on or trying to use the wrong window or things like that), not the non-determinism.
so basically what you're saying is that: it doesn't do the math, it tells the math-doing-thing what math to do. Basically, instead of humans using Excel, imagine AI using Excel?
Yet I don't understand the aha moment here? It might save analyst time but aren't there already enough automation that you don't really need to tell the AI to tell the math-doing-thing to do the math because the math-doing-thing is already optimized for most general functions? What are we gaining from adding the non-deterministic process here when the real non-deterministic process is still the human being prompting what to do?
Seems like a solution to a non-problem from my pov.
Setting up spreadsheets is just programming in an unconventional environment.
And it has all the requirement translation requirements of programming
Sure in the sense that you're setting up a program that has inputs and outputs etc. But then all math is programming. All language is too, even speaking can be considered programming if you're stretching the definition enough. But I will disagree that setting up spreadsheets = basically software engineering.
"The non-deterministic part is turning human instructions ("calculate the NPV over 10 years for X given Y") into Excel."
erm theres an NPV function built in excel.
> erm theres an NPV function built in excel.
Exactly!
The LLM just needs to make sure it uses it appropriately. Doing that bit is the non-deterministic part, but the NPV calculation itself is completely deterministic.
Youre not quite getting it, are you.
When you setup a spreadsheet, you choose a column or cell to show the NPV. You use the npv formula on this cell and apply it to inputs from other columns or cells.
This process of deciding what data to put in what column and how to apply NPV to it is non-deterministic. You could choose to put it in column A, or column B or maybe even row 3 - it depends!
The LLM does that part non-deterministically.
Do you really think you come off as a good person posting things like this? This place is to satisfy intellectual curiosity, not to lord your supposed intelligence over others, without even proving it exists in the first place.
But humans non-deterministically use that or use a hand-rolled formula.
Humans dont enter numbers non-determinstically. WTF are you on about lmao.
Have you ever done anything related to corp finance/valuation in a professional setting? Highly doubt it.
[dead]
I know plenty of nondeterministic accountants
With tool use you do reduce the risks.
It's not like these models calculate.
Yeah, seriously, I use AI all day every day but that terrifies me.
[flagged]
Dont use Excel for accounting....
HN ignorance on display...
The Journal of Accountancy - "Bugged by Excel’s calculation errors" - https://www.journalofaccountancy.com/issues/2014/mar/excel-c...
ICAEW (Institute of Chartered Accountants) — "Rounding Errors Revisited" (Excel Tips #447) - https://www.icaew.com/technical/technology/excel-community/e...
Just a reminder that an accountant who might say "I use Currency format" is still working in binary floating point as the format is just used as a display mask. And using VBA macros with the Currency type will hit problems at the boundary, when values move between the worksheet and the macro. The tool is broken in a way that proper accounting software is not...
"Floating-point arithmetic may give inaccurate results in Excel" - https://learn.microsoft.com/en-us/troubleshoot/microsoft-365...
An Excel spreadsheet is a bad source of truth for data. It's fine for analytics though.
[flagged]
[flagged]
Do they use it as a source of truth or do they import that data in a spreadsheet to let the user make their own calculations?
Not the person you responded to but I'm genuinely curious.
Afaik Excel is wicked fast at math (can handle ginormous spreadsheets, some real magic under the hood), while correctly doing decimal based math (1/5 is 0.2 exactly, not some floating point monstrosity)
A yes PWC :-) That is what I would expect :-))
I just hope the divine powers of Luca Pacioli might help you, before is too late. A whole set of accounting bodies and real world disasters expressly advise against its use and spreadsheets in general.
You even have dedicated pages...
https://en.wikipedia.org/wiki/List_of_spreadsheet_mistakes
EuSPRIG Horror Stories - Spreadsheet mistakes
https://eusprig.org/research-info/horror-stories/
For example the (ICAEW) UK chartered accountants explicitly warn auditors. And research has shown that around 94% in these contexts contain errors:
"Errors in Operational Spreadsheets" - https://faculty.tuck.dartmouth.edu/images/uploads/faculty/se...
Also... Excel does not use exact decimal arithmetic. It stores numbers in IEEE-754 binary floating point, which cannot precisely represent many decimal values used in accounting. Microsoft documents this explicitly:
https://learn.microsoft.com/en-us/office/troubleshoot/excel/...
It also only preserves 15 significant digits and silently zeroes anything beyond that:
https://support.microsoft.com/en-us/office/excel-specificati...
So this means:
– decimal values are approximated
– rounding errors accumulate
– large financial values lose precision
– results depend on formula structure and rounding order
That alone disqualifies it as Accounting ledger. Historically you have lots of famous examples of bad outcomes of treating Excel as financial infrastructure...
The JPMorgan London Whale - $6B loss A flawed Excel risk model and copy paste errors understated risk and helped produce multi-billion losses. https://en.wikipedia.org/wiki/2012_JPMorgan_Chase_trading_lo...
TransAlta energy trading — $24M loss Simple spreadsheet row misalignment in bidding model cost aprox 10% of the company profit. https://www.lumeer.io/spreadsheet-for-project-management/
Fannie Mae - $1.3B reporting error wrong spreadsheet formula caused more than $1B accounting misstatement. https://www.theguardian.com/technology/2024/oct/28/microsoft... https://archive.is/w1cjj
Fidelity Magellan — $2.6B error Missing minus sign in a spreadsheet overstated capital gains by billions. https://en.wikipedia.org/wiki/List_of_spreadsheet_mistakes
Large audit firms also publish entire risk frameworks, ( and I am sure PWC does also...) explaining why uncontrolled spreadsheets are a major source of financial misstatement in the first place.
This entire site is full of idiots confidently spouting nonsense. Probably time to move on. "Don't use Excel for accounting" has to be the dumbest thing I've read in a year, even with all of the LLM bullshit going on.
And that's including the people this morning who are apparently excited for LLMs to file their taxes.
I've been having the same feelings lately, especially around the AI doomers coming in with some weird observations that make no sense. I usually don't comment, but seeing a lot of these types of overgeneralized responses.
'oh you do x? You don't do y? You're an idiot'
That's not productive.
I have the impression today you will be one of the 10,000: https://xkcd.com/1053/
But you are correct, this entire site is full of idiots confidently spouting nonsense... :-)
"Floating-point arithmetic may give inaccurate results in Excel" - https://learn.microsoft.com/en-us/troubleshoot/microsoft-365...
"Classification of Spreadsheet Errors" - https://arxiv.org/pdf/0805.4224
"A study conducted by Coopers & Lybrand found errors in 90% of the spreadsheets audited."
"Impact of Errors in Operational Spreadsheets" - https://arxiv.org/abs/0801.0715
"What We Don’t Know About Spreadsheet Errors Today" - https://arxiv.org/pdf/1602.02601
You don't really deserve a response, but even the idea that you could tell someone to not use Excel for accounting and seriously project that you know anything about how accounting works in the real world is hilarious.
You could think 500 random internet articles - it doesn't matter. The assertion is still ridiculous. You literally will not be able to get a job as an accountant without being able to use Excel. Don't be an idiot.
You can also link to hundreds of articles showing that dynamic types cause errors. People don't care, they still use them when it makes sense to use them. There is no better mainstream alternative for accounting than Excel right now, period.
Not a single reply on the technical and governance arguments?
I think we agree...this site is indeed full of idiots confidently spouting nonsense. :-)
No, not a single reply, because your statement wasn't "Excel can cause problems sometimes". That's a different conversation. You said "Don't use Excel for accounting", which is something you could only say if you have no idea at all about how accounting works in the real world.
I'm not arguing about whether or not Excel is perfect, or if it causes problems, or if it makes demons fly out of your nose. I am educating you about the real world and how it's important to understand that no matter how "well akshually" technically correct you are, it doesn't matter at all if the advice you're giving is WRONG and not useful to people.
So, I repeat. Idiots confidently spouting nonsense.
Hah. These are the same weirdos who spoke like this to Jobs: https://www.youtube.com/watch?v=oeqPrUmVz-o
There's a big difference between geeks and hippies. Too many of the former here and not enough of the latter here - people who can understand other people, culture etc and then think about how technology can fit into ones life.
It's shocking and honestly pathetic. "Hey guys the software you use is not accurate" being met with such derision is how we regress as a society, not move forward.
nothing is accurate in the hands of incompetent people
Did you actually read these or just Google for problems?
> "Floating-point arithmetic may give inaccurate results in Excel"
Well.. yes? That's what floating point numbers do.
https://stackoverflow.com/questions/40168077/floating-point-...
So what?
> "Classification of Spreadsheet Errors"
It's a taxonomy of the types of bugs spreadsheets have.
There are these for most other programming languages too.
etc, etc
Look, my first job was translating an Excel spreadsheet into an actual application and we found errors all the time.
Once Excel spreadsheets get too complicated they tend to fall apart. This is very well known by people who use it heavily.
It doesn't mean that it is useless.
"This entire site is full of idiots confidently spouting nonsense."
Too right - this goes for the arrogant geeks at the within the orgs of LLM producers who still havent displaced call center workers with their technology.
Comical stuff. Really is.
The lack of disconnect with what people actually do in jobs is absolutely breathtaking.
> Too right - this goes for the arrogant geeks at the within the orgs of LLM producers who still havent displaced call center workers with their technology.
Some have shares in these AI companies and need to off-load them either on secondaries or IPO.
These are the clear incentives to show performative results to give the impression that you can replace accountants with AI without any of the risks involved.
Unfortunately if those jokers intend to IPO later this year it wont go down too well. They should've done it 6 months ago.
[flagged]
The benchmarks look good. Slide decks and spreadsheets look better. The people must use Claude Cowork and have their Claude Code moment and figure out the consequences. It will be really interesting to see articles like this (https://mitchellh.com/writing/my-ai-adoption-journey) written by people who actually care about accuracy in places like KPMG to get their perspective on things.
I remember over hearing some normal people on the bus talking about essentially orchestrating some agent scraper to pull and summarise news from 40 different sites he identified as important which put him quite ahead of his peers. These were non-technical people orchestrating an agent workflow to make them better at work.
Though there’s not much that tickles my software brain here. But the agents are coming for us all.
Based on the article... is this basically just making Claude better at formatting and data presentation, or does it also get better at analysis? I get the impression it's the former.
> The side-by-side outputs below show how output quality has improved from Claude Opus 4.5 to Opus 4.6.
Disclaimer: I use AI to code (and I code for finance) and I love Anthropic.
But: for f-ck's sake, I cannot click on the picture and have it show up in full. It stays at its tiny size, impossible to read the numbers. I had to right-click and "open in a new tab".
AI is, somehow, definitely still not fully there yet.
And then you hand it to your boss who takes a 20 second look at it and asks why you made a projection that assume massive revenue growth and 3 years of perfectly flat utilities, insurance, G&A - no inflation etc.
It does look really promising as a skeleton starting point though. Like generate it, delete numbers and populate by hand.
Not unlike the boilerplate start we saw in AI coding a couple years back
>Like generate it, delete numbers and populate by hand.
This is called 'Month End Close' in accountant speak.
It's time to sell hedge fund stocks!! Jokes aside, I took the CFA exam last week and now I'm starting to worry about my career...
I wouldnt worry. Unless you are just memorising stuff and dont actually understand anything - then you should.
Just like developers, accounting won't go away. Just less people needed.
Since demand is so insane high. We will just get into an equilibrium.
Not like developers, where like 25 percent of people want to do something with IT...
You'll use a ton of AI but it won't wipe the humans out. In the end you'll have a compositional change, likely nothing catastrophic imo. In part because there is a buck to stop and Claude ain't got no hands...
Article did not load on my tablet :sweat_smile:
Anthropic does anything to keep the Claude hype going; from fearmongering ("AI bad, need government regulations") to wishful thinking ("90% of code will be written by AI by the end of 2025" —Dario) to using Claude in applications it has no business being in (Cowork, accessing all your files, what could go wrong?) to releasing "research" papers every now and then to show how their AI "almost got out" and they stopped it (again, to show their models are "just that good") to prescribing what the society should do to adapt to the new reality to doing worthless surveys on "how AI is reshaping economy, but mostly our AI not others".
It’s just marketing, no evil intended.
Their chart only goes up to 70.
At least it starts at zero.
Now this is going to be interesting to watch to see if the finance bros financing this AI wave to get rid of SW engineers will keep financing getting rid of their own.