Claude for excel is already amazing. Fully capable of doing junior work. Formatting is great. Can refactor large multi-tab spreadsheets. It just burns tokens. If OpenAI is going to subsidize this on the monthly enterprise plans for a while then it's a game changer.
Claude for Excel (I work in finance) was one of the absolutely critical reasons we added Anthropic enterprise licenses. But they've turned out to be quite expensive ($100/day for heavy users). We'll see what OpenAI's quotas are.
From my experience, LLM performance in these areas is being massively oversold. I have repeatedly tried using Claude to modify a range of models typical of investment banking / private equity / sellside research contexts, and the results have been generally disastrous. On multiple occasions, the xlsx would no longer open.
Just my experience, it’s not a solution but rather a productivity tool. I mostly use it for tasks I can do myself but it would probably take 20-30min to dial in - now Claude can do it in 2-3min. (E.g. in a data table - add a new column that checks column a if the data is a, do x, if the data is b, do y, if the data is c, do z - then combine that with the word after the hyphen in column b —- or another example —- create a new sheet that is the same format as sheet one but show calculates the difference between column a and b bot for sheets 1-12 in a summary)
I don’t get good results when I just have Claude build things on its own - but for these types of specific productivity tasks I can save a couple of hours here and there.
I work with large files a lot, running claude code on it is not token intense at all. Probably because it does a lot with scripts. But its a bit more raw, but i think in the end more powerful. Have to pick a good excel library and language. I do node, maybe python can work as well
Work in a firm similar to yours and we have been going to though the motions of figuring ways for the bullpen to make use of these tools and would love to hear your thoughts if you would be willing to share!
We already have all-electric trainers like the Bye eFlyer https://byeaerospace.com/ so I can see this "working", but I'm not certain how effective it would be compared to something as well-tested as the "stealth" version of the MH-6 helicopter that's been in production for about a decade.
Additionally, the basic non-stealth MH-6 airframe and power-plant configuration has been around since the 1960s so its base flight characteristics are well-known.
I agree with a lot of the "how to" suggestions here. One addition is that you need to have a small number of key non-development employees really guide the development with strong opinions on functionality. This will likely mean your strongest folks in the departments that the software will serve. I would avoid building this by committee. 1-2 of your best folks who will be the ultimate end users should have massive decision-making power in what gets developed, in what order, and for what purpose.
Also, don't underestimate maintenance in your cost assumptions. Even after you've developed the product, you will need several people full-time just to maintain, update, and bug fix, let alone add new features that got sidelined during the initial development in order to meet the MVP.
Matt Levine who writes for Bloomberg is a great daily read. He posted about this about a week ago and it provides a good overview of how this type of thing is priced/insured.
And his analysis is almost embarrassingly bad. I cannot fathom how someone who worked in investment banking at Goldman could be so far off in how to price this.
The Epicurean Dealmaker wrote a fantastic post a while back, and I will endorse but not repeat his entirely correct explanation of why IPOs are supposed to price at a discount; just go read it.
The most interesting bit is: "Wednesday's proposal doesn't require companies to verify that individuals meet income thresholds set by the law, officials said. Instead, the SEC asks for comment on whether verification steps are needed."
Potentially receiving funding from a non-accredited investor has been one of the remaining sources of risk to a startup taking crowdfunding.
I've scraped pretty much all of the funding and company data from crunchbase and reorganized it so that it shows a number of different insights, including what you see here: for any VC Firm, who do they co-invest with most often, and on what specific deals.
Total funding for different rounds and various time periods is also available on the homepage of the site. This data really interests me and I hope it can maybe help someone evaluate a VC Firm or to be used as a tool when fundraising.
We started a company called ChubbyBrain (CB is our tribute to that). 0% of our data is from crunch.
In terms of where it comes from:
80% of our data comes via software we've built to parse news, SEC, investor, corporate websites (we crawl about 12k of them daily).
About 20% of our data comes directly from investors. The biggest contingent is angel data which we get via a partnership we have with Silicon Valley Bank and the Angel Capital Association.
Claude for Excel (I work in finance) was one of the absolutely critical reasons we added Anthropic enterprise licenses. But they've turned out to be quite expensive ($100/day for heavy users). We'll see what OpenAI's quotas are.