Best Idea(s) for 2025

[winjitsu]

7 hours ago, benchmark said:

It's not the data, but the usability of the excel sheet that makes it stick. I agree that always connect to the 'right data source' is better, until dashboard becomes easy to create/use/share, excel has a long life line. 

I don't want to hijack this conversation away from ideas, but we are already at the point where Claude Code / ChatGPT can make a dashboard in one sitting. 

Once upon a time, I was an analyst, and similarly, about 10 years ago, we were asked to see if we could move things off of Excel into a business intelligence platform like Tableau (some workbooks are taking five minutes to open). We reached out and got some quotes that probably totaled in the six figures, including implementation over several months and annual seat costs. How far we've come. 

I'm not predicting apocalypse, but I do think financial analysts are gonna be in a similar boat with software developers. Before an executive might ask, "Hey, can you rerun those numbers and tell me what the breakdown was by geography and cost?" And the FA would go back to their workbook and play with the data and change around the pivot tables and get back. But that actually can be done by AI in thirty seconds. 

 

11 hours ago, brobro777 said:

 

See bro that sounds way too complicated. These lawyer types don't wanna deal with that - they'll just pay MSFT some more money and move on to being awful people that they are. Besides, they don't want their precious legalese to be tampered with AI, let alone put into some weirdo box!

 

No!

 

 

We must work with different lawyers  All my friends (mid-level) use AI, especially in legal drafting, proof-reading and help with emails (though of course, they then paste that into a word doc/outlook). I imagine the 1Ls have been using AI extensively over their studies.

Proprietary formats like Word .doc actually do not play well with LLMs and .txt and .md files are much better, ditto .csv work better than .xlsx. I don't think it's much of a stretch to think in the future we're going to be working more with AI-centric formats than human-centric formats. 

Partner level, though, yeah, they're never gonna change. 

Side note: China has made the lift from Office to Kingsoft. It actually isn't that bad. 

All to say, I don't think the Office moat is unassailable and we're in for very interesting times this next decade. 

Edited January 19 by winjitsu

[bargainman]

12 hours ago, winjitsu said:

I don't want to hijack this conversation away from ideas, but we are already at the point where Claude Code / ChatGPT can make a dashboard in one sitting. 

Once upon a time, I was an analyst, and similarly, about 10 years ago, we were asked to see if we could move things off of Excel into a business intelligence platform like Tableau (some workbooks are taking five minutes to open). We reached out and got some quotes that probably totaled in the six figures, including implementation over several months and annual seat costs. How far we've come. 

I'm not predicting apocalypse, but I do think financial analysts are gonna be in a similar boat with software developers. Before an executive might ask, "Hey, can you rerun those numbers and tell me what the breakdown was by geography and cost?" And the FA would go back to their workbook and play with the data and change around the pivot tables and get back. But that actually can be done by AI in thirty seconds. 

 

 

We must work with different lawyers  All my friends (mid-level) use AI, especially in legal drafting, proof-reading and help with emails (though of course, they then paste that into a word doc/outlook). I imagine the 1Ls have been using AI extensively over their studies.

Proprietary formats like Word .doc actually do not play well with LLMs and .txt and .md files are much better, ditto .csv work better than .xlsx. I don't think it's much of a stretch to think in the future we're going to be working more with AI-centric formats than human-centric formats. 

Partner level, though, yeah, they're never gonna change. 

Side note: China has made the lift from Office to Kingsoft. It actually isn't that bad. 

All to say, I don't think the Office moat is unassailable and we're in for very interesting times this next decade. 

 

Well MSFT is trying to build copilot into excel and python too  When I last tested it (6+ months ago) copilot sucked and didn't look at the spreadsheet content, more just help file stuff).  But MSFT just needs to add enough AI/MCP whatever integration to prevent mass migration away.

 

I suspect they're working to get xls and llms to work better together since they can access all chat gpt's IP.

[Spekulatius]

4 minutes ago, bargainman said:

 

Well MSFT is trying to build copilot into excel and python too  When I last tested it (6+ months ago) copilot sucked and didn't look at the spreadsheet content, more just help file stuff).  But MSFT just needs to add enough AI/MCP whatever integration to prevent mass migration away.

 

I suspect they're working to get xls and llms to work better together since they can access all chat gpt's IP.

It’s still the same way. Copilot is still not able to look into your spreadsheet or formulas and it works just like a typical help suggestion. It basically really just the old paperclip with a few upgrades but the paperclip looked better, imo.

It‘s surprising how bad it is. I don’t get how people are concerned about all sorts of sticky software getting replaced by AI but not MSFT. I think Google might be able to build something with their own AI agent that helps with  conversions and remove a big pain point and that could make MS Office vulnerable. Thats just one example how this could go. 

[UK]

3 hours ago, bargainman said:

 

Well MSFT is trying to build copilot into excel and python too  When I last tested it (6+ months ago) copilot sucked and didn't look at the spreadsheet content, more just help file stuff).  But MSFT just needs to add enough AI/MCP whatever integration to prevent mass migration away.

 

I suspect they're working to get xls and llms to work better together since they can access all chat gpt's IP.

So they where first with all this AI (remember, they were about to make Google dance) and pardon my french, but this copilot is such a shit it is not even funny. The only thing you do with it is just try to find how to disable forever.

 

And how about this: https://www.techradar.com/televisions/your-lg-tv-may-get-an-unremovable-microsoft-copilot-app-in-its-next-update-and-yes-users-are-annoyed

[UK]

3 hours ago, Spekulatius said:

It basically really just the old paperclip with a few upgrades but the paperclip looked better, imo.

[bizaro86]

4 hours ago, bargainman said:

 

Well MSFT is trying to build copilot into excel and python too  When I last tested it (6+ months ago) copilot sucked and didn't look at the spreadsheet content, more just help file stuff).  But MSFT just needs to add enough AI/MCP whatever integration to prevent mass migration away.

 

I suspect they're working to get xls and llms to work better together since they can access all chat gpt's IP.

 

I have a couple things I use google sheets for (mainly for Google Finance API) and I was blown away how much better their AI suggestions are. I made a change to a drop down, and it popped up a box asking if I wanted to make a related change to a certain set of cells - and that was exactly what I was planning to do next and it knew and did it automatically. Only saved me 60 seconds, but definitely made me think about my strong preference to do everything in Excel...

[Hektor]

12 hours ago, Spekulatius said:

It’s still the same way. Copilot is still not able to look into your spreadsheet or formulas and it works just like a typical help suggestion. It basically really just the old paperclip with a few upgrades but the paperclip looked better, imo.

It‘s surprising how bad it is. I don’t get how people are concerned about all sorts of sticky software getting replaced by AI but not MSFT. I think Google might be able to build something with their own AI agent that helps with  conversions and remove a big pain point and that could make MS Office vulnerable. Thats just one example how this could go. 

I guess MSFT will enhance their AI offering when their customers start asking for it, not when their competitors are rolling them out. MSFT apparently has a moat, which helps.

[Hektor]

9 hours ago, UK said:

The only thing you do with it is just try to find how to disable forever.

[brobro777]

On 1/19/2026 at 4:01 AM, winjitsu said:

I don't want to hijack this conversation away from ideas, but we are already at the point where Claude Code / ChatGPT can make a dashboard in one sitting. 

Once upon a time, I was an analyst, and similarly, about 10 years ago, we were asked to see if we could move things off of Excel into a business intelligence platform like Tableau (some workbooks are taking five minutes to open). We reached out and got some quotes that probably totaled in the six figures, including implementation over several months and annual seat costs. How far we've come. 

I'm not predicting apocalypse, but I do think financial analysts are gonna be in a similar boat with software developers. Before an executive might ask, "Hey, can you rerun those numbers and tell me what the breakdown was by geography and cost?" And the FA would go back to their workbook and play with the data and change around the pivot tables and get back. But that actually can be done by AI in thirty seconds. 

 

 

We must work with different lawyers  All my friends (mid-level) use AI, especially in legal drafting, proof-reading and help with emails (though of course, they then paste that into a word doc/outlook). I imagine the 1Ls have been using AI extensively over their studies.

Proprietary formats like Word .doc actually do not play well with LLMs and .txt and .md files are much better, ditto .csv work better than .xlsx. I don't think it's much of a stretch to think in the future we're going to be working more with AI-centric formats than human-centric formats. 

Partner level, though, yeah, they're never gonna change. 

Side note: China has made the lift from Office to Kingsoft. It actually isn't that bad. 

All to say, I don't think the Office moat is unassailable and we're in for very interesting times this next decade. 

 

 

The prosecution rests!

 

[UK]

On 1/8/2026 at 8:36 AM, brobro777 said:

 

Boy you value investor guys are harsh as hell - newspapers from 2000? What you gonna say next, Adobe Creative cloud with 41mil paying subscribers is Yellow Pages?

 

Damn!

 

https://uk.finance.yahoo.com/news/goldman-issues-a-blunt-warning-to-beat-up-software-stock-investors-141358103.html

[brobro777]

4 hours ago, UK said:

https://uk.finance.yahoo.com/news/goldman-issues-a-blunt-warning-to-beat-up-software-stock-investors-141358103.html

 

 

You and your Goldman Funboys, get a room!

 

 

 

[Longnose]

20 hours ago, UK said:

https://uk.finance.yahoo.com/news/goldman-issues-a-blunt-warning-to-beat-up-software-stock-investors-141358103.html

 

I dunno man i still struggle to get behind this narrative that software is dead. having vibe coded several little projects myself thus far its an amazing enabler but to say you're gonna build something that replaces these businesses and all that they address? Man i don't know about that.  The bottom end stuff that supports the lower tier market sure. competition will get fierce in that market. But anything that supports enterprise IMO will continue to persist and likely dominate its space. getting an enterprise creative team off adobe? doesn't seem likely.  getting your business to switch CRM's from salesforce to some vibe coded solution? unlikely.  But sure your an entrepreneur with a sub 50M$ business and you don't wanna pay for all the bells and whistles of those companies sure you can get away with some vibe coded solution. but if/when you mature. you'll join the safety/security of the big player packages when you can afford it. Or you can risk getting sued for using images nanobanana created without verifying copyright or getting sued cuz your vibe coded CRM accidently leaked PII for all the customers you've been collecting personal information on. 

[UK]

6 minutes ago, Longnose said:

 

I dunno man i still struggle to get behind this narrative that software is dead. having vibe coded several little projects myself thus far its an amazing enabler but to say you're gonna build something that replaces these businesses and all that they address? Man i don't know about that.  The bottom end stuff that supports the lower tier market sure. competition will get fierce in that market. But anything that supports enterprise IMO will continue to persist and likely dominate its space. getting an enterprise creative team off adobe? doesn't seem likely.  getting your business to switch CRM's from salesforce to some vibe coded solution? unlikely.  But sure your an entrepreneur with a sub 50M$ business and you don't wanna pay for all the bells and whistles of those companies sure you can get away with some vibe coded solution. but if/when you mature. you'll join the safety/security of the big player packages when you can afford it. Or you can risk getting sued for using images nanobanana created without verifying copyright or getting sued cuz your vibe coded CRM accidently leaked PII for all the customers you've been collecting personal information on. 

Just was messing around with bro:), Goldman will do, what they allways do (rock the boat whichever way), personally, I still have no idea or strong conviction about all this.

[Longnose]

3 minutes ago, UK said:

Just was messing around with bro:), Goldman will do, what they allways do (rock the boat whichever way), personally, I still have no idea or strong conviction about all this.

Haha fair enough. I have a large position in ADBE right now.  I know sentiment will likely persist negative for some time. But I cant bring myself to believe that AI will enable their demise. If anything It will grow their TAM IMO.