What stands out to me is how MCP brings the same power that used to require tons of manual scripting, API wrangling, or jumping between tools and makes it accessible through one central, interactive conversation.
We’ve technically been able to achieve some of this before, but the level of effort (and technical overhead) was so high that most people never touched it.
Now, with MCP, it’s like the wall between thinking and doing is finally dissolving. One interface, infinite interactivity.
Yes, with MCP everything feels seamless and this is where the trend is going. AI should be able to interact with our most fav apps to make our lives better ofc!
I guess it also depends on what personality type you have. If I have one meeting with someone and the AI automatically creates a bunch of tasks, project items and draft emails all over the place, I'd go absolutely insane. Most of that will be like a seagull crapping all over my tidy office, I'm sure. But in general, I agree. Cool stuff is possible.
I think it depends whether the AI is accurately doing so. I think most people will find this helpful if it's accurate.
But I do understand project has a lot of nuance and context. And sometimes people who have tendency to control are less likely to leave out everything to AI. At least for me, most of task creation works are done mostly by AI so I can edit it later on. It saves me so much time.
As a busy mum, coming out of maternity leave and starting to make time for my personal endeavours, this is game changing and I can’t wait to try it. One question though…… is all the data Claude is getting from us going back being put back into it’s training data? I notice, this only works in desktop version of Claude.
As far as I know they don't. Anthropic is quite conservative in training their models and they emphasize strongly on privacy too.
Beside, getting MCP data and use it on training models could cost them a lot to process all informations. I don't think that's the best approach to feed their models. But, keep it in mind: dont do it if you are not comfortable :)
We will not use your Inputs or Outputs to train our models, unless: (1) your conversations are flagged for Trust & Safety review (in which case we may use or analyze them to improve our ability to detect and enforce our Usage Policy, including training models for use by our Trust and Safety team, consistent with Anthropic’s safety mission), or (2) you’ve explicitly reported the materials to us (for example via our feedback mechanisms), or (3) you've otherwise explicitly opted in to the use of your Inputs and Outputs for training purposes.
just ask Claude to tell u in plain english on how to implement it, feel free to DM me in case u find issue or wait on my later post to take u through the step by step implementation.
Nice post about MCP and its use cases. One thing I always wonder when hearing about AI agents and MCP tools is whether the problem would not be better solved by a workflow, which is more robust and stable for relatively simple tasks such as turning a client conversation into action items in Notion.
ah yes, simple task better be automated but some work requires more nuance in our thinking, so I don't particularly automate it yet but use AI in between as thinking partner to generate better outputs.
I love the how you're giving people the ability to be able to connect all of their systems. It really drives growth in a space where AI is finally able to take coherent and logical actions.
Whats the difference in the gemini ecosystem that is already integrated with gmail/gtasks/notes and takes notes in g meet? Would you say it is similar capabilities l?
But Gemini itself on their core platform can't make decision or actions toward their own tools. It's not interconnected. You need to go to gmail, gdrive, gdocs to actually use Gemini separately.
But with MCP, you can do all these things inside Claude.
Would you mind preparing an in-depth post how to integrate MCP into Claude Desktop? Thanks!
I will :)
Would definitely be a huge plus!
What stands out to me is how MCP brings the same power that used to require tons of manual scripting, API wrangling, or jumping between tools and makes it accessible through one central, interactive conversation.
We’ve technically been able to achieve some of this before, but the level of effort (and technical overhead) was so high that most people never touched it.
Now, with MCP, it’s like the wall between thinking and doing is finally dissolving. One interface, infinite interactivity.
Yes it works same as API but automated by AI.
Yes, with MCP everything feels seamless and this is where the trend is going. AI should be able to interact with our most fav apps to make our lives better ofc!
The Industrial Revolution of digital knowledge work.
I guess it also depends on what personality type you have. If I have one meeting with someone and the AI automatically creates a bunch of tasks, project items and draft emails all over the place, I'd go absolutely insane. Most of that will be like a seagull crapping all over my tidy office, I'm sure. But in general, I agree. Cool stuff is possible.
I think it depends whether the AI is accurately doing so. I think most people will find this helpful if it's accurate.
But I do understand project has a lot of nuance and context. And sometimes people who have tendency to control are less likely to leave out everything to AI. At least for me, most of task creation works are done mostly by AI so I can edit it later on. It saves me so much time.
As a busy mum, coming out of maternity leave and starting to make time for my personal endeavours, this is game changing and I can’t wait to try it. One question though…… is all the data Claude is getting from us going back being put back into it’s training data? I notice, this only works in desktop version of Claude.
I'm sure it's going to be exciting!
As far as I know they don't. Anthropic is quite conservative in training their models and they emphasize strongly on privacy too.
Beside, getting MCP data and use it on training models could cost them a lot to process all informations. I don't think that's the best approach to feed their models. But, keep it in mind: dont do it if you are not comfortable :)
Anthropic’s Privacy Policy (Privacy Policy \ Anthropic at https://www.anthropic.com/legal/archive/86d180d7-2d2c-4821-9258-b2224acc06ea) states the following at the end of Section 2:
We will not use your Inputs or Outputs to train our models, unless: (1) your conversations are flagged for Trust & Safety review (in which case we may use or analyze them to improve our ability to detect and enforce our Usage Policy, including training models for use by our Trust and Safety team, consistent with Anthropic’s safety mission), or (2) you’ve explicitly reported the materials to us (for example via our feedback mechanisms), or (3) you've otherwise explicitly opted in to the use of your Inputs and Outputs for training purposes.
Thank you!
Wyndo, this is such a solid step forward in using Claude MCP!
I’ve been noticing that Claude’s integrations with other apps are way ahead of what we’re seeing from ChatGPT or even Gemini right now.
Really appreciate how clearly you broke it all down, the phased approach makes it so much easier to get started without overwhelm.
Honestly, this is exactly where we are heading, and it’s exciting to see things coming together like it.
Claude to me is pushing more boundaries here :)
Love to see they are going toward this direction and differentiate themselves from ChatGPT.
Yeah, with this direction, Claude actually already become an AI agent that can make decision which tools to use to achieve the goals from their users.
It’s inevitable, but AI and its owners will know nearly everything about everybody, soon.
Thank you for this Wyndo! Can you share how you setup memory MCP? One my biggest pain points is getting projects or chats within Claude to communicate
you can use this one: https://github.com/doobidoo/mcp-memory-service
just ask Claude to tell u in plain english on how to implement it, feel free to DM me in case u find issue or wait on my later post to take u through the step by step implementation.
Nice post about MCP and its use cases. One thing I always wonder when hearing about AI agents and MCP tools is whether the problem would not be better solved by a workflow, which is more robust and stable for relatively simple tasks such as turning a client conversation into action items in Notion.
What are your thoughts on that?
ah yes, simple task better be automated but some work requires more nuance in our thinking, so I don't particularly automate it yet but use AI in between as thinking partner to generate better outputs.
What did you use for the google calendar MCP?
It’s available on Claude web and desktop
Do you know if it’s their own MCP or if they leveraged something? I assume it’s their own
I love the how you're giving people the ability to be able to connect all of their systems. It really drives growth in a space where AI is finally able to take coherent and logical actions.
i think this is the next shift of AI where it can actually connect and automate most of our workflows and escape the chat box prison :)
Wait until you can get those brain implants.
Great summary
Super helpful letter, thank you.
Whats the difference in the gemini ecosystem that is already integrated with gmail/gtasks/notes and takes notes in g meet? Would you say it is similar capabilities l?
But Gemini itself on their core platform can't make decision or actions toward their own tools. It's not interconnected. You need to go to gmail, gdrive, gdocs to actually use Gemini separately.
But with MCP, you can do all these things inside Claude.
No need to jumping around multiple tools.
One single platform to rule them all!
Fascinating ! Thank you so much for sharing
Would love to connect with you directly and chat about where you plan on taking this next
Glad to know u resonate with this :)
DM is open, happy to chat!
Thanks for the input