I wouldnt be surprised if the next target is slack like communication tool.
Then they would own the whole stack for organizational context, which is the biggest problem in enterprise AI.
I wouldnt be surprised if the next target is slack like communication tool.
Then they would own the whole stack for organizational context, which is the biggest problem in enterprise AI.
Introducing Claude Meetings: add Claude to your teamโs video call and it builds docs, whiteboards, and other artifacts with you while youโre still talking.
Starting to roll out in beta on Team and Enterprise plans.
@ivanfioravanti Had a similar idea for gym, to see which machines are used most and how long, not sure what would be the use case for gym owners though
@volosatovde I agree it s too much)) but i wouldnt be surprised if companies started treating these conversations as valuable data and making changes around communication
Building a company brain is essentially figuring out the API calls between humans.
Once you know who ask whom for what - you get the hidden service layer.
@volosatovde That is so true! And such a paint point for org context
I wonder if we will see communication tools that helps capture context better, e.g. slack but no dms
@thomaspaulmann I usually ask it to make a simple video to summarize its work and use elevenlabs for voice.
My morning routine turned it into watching videos where agent reports its work, showing ui actions, db entries, explaning the logic etc.
Today we demoed AI company brain to entire team (50+ people). We did a long QA session at the end. Some observations based on questions:
- Most questions were about the accuracy of the answers.
- Teams wanted to define what a metric really is within the company context, e.g. our conversion rate is not the conversion rate the llm thinks at the moment
- But nobody wants to teach AI their process. They want their definitions respected without writing docs for it
- Who can see what is a big deal. Features around permission/access control are necessary at this point
- No one really cares about MCP/CLI or different models
- They really want systems that can tell when it doesn't know the answer
- People are tired of ai tools, seems like they have been promised a lot and burned before
Take away for us: adoption within enterprises will be very long and painful process. Key to this is going to be trust - systems that can consistently provide answers, cite sources and say when they can't answer.
There are so many challenges to be solved for extracting company knowledge and processes - especially around meetings & privacy, that I can see why there is no mainstream org-context product yet.
Some edge cases we've seen so far:
- AI notetakers join calls as "people". 1:1 plus a notetaker looks like a 3-person meeting... Now if you want to only index group meetings (which is what most clients wants) you gotta figure out the real people
- Same people can joins from two Google accounts, now you need to figure out deduplication
- Names of products, companies, metrics or people often doesnt survive AI transcripts. you need to handle entity resolution at the meeting level
It almost feels like there should be a separate product that transcribes meetings based on the org context
@KenTheRogers Why not just ask Claude to maintain the app? Proactive agents should handle it easily.
If you are okay with vibe-coded app, just ask it to fix issues if/when they break.