Enhancing MLO with an AI MCP Server for Smarter Integration (try 2)

[Daniel Clark]

I’m a dedicated user of My Life Organized (MLO) and love how it keeps my tasks in sync across Android, iOS, Windows, and soon the web. I also rely on Shortwave Email and the Bee AI Wearable, where I’ve been using an MCP Server (via [beemcp on GitHub](https://github.com/OkGoDoIt/beemcp)) to seamlessly integrate my email and Bee. This setup transforms email into an intelligent extension of my workflow, allowing AI to create todos in Bee from emails, reference past decisions, and understand my priorities—all without switching platforms. I’d love to extend this capability to MLO by adding an AI MCP Server, and I believe it could take MLO to the next level.

### What is an AI MCP Server?

The Model Context Protocol (MCP) is an open standard from Anthropic that lets AI models connect to apps and data sources through a universal interface. An MCP Server for MLO would expose its todo lists, priorities, and actions to AI assistants (e.g., in email or wearables), enabling them to read and interact with MLO data securely and efficiently. Learn more at Anthropic’s [MCP introduction](https://www.anthropic.com/news/introducing-the-model-context-protocol).

### Advantages for MLO

Adding an MCP Server would bring these benefits:

- **Context-Aware AI**: AI assistants could see my MLO todos and priorities, incorporating them into responses in email or other platforms.

- **Seamless Todo Creation**: I could create MLO todos directly from emails or conversations without leaving the app I’m in.

- **Unified Workflow**: The AI would understand my full thought process across MLO, email, and Bee, eliminating context loss.

- **Smarter Interactions**: By observing patterns in MLO, the AI could offer better suggestions and automate tasks based on my preferences.

This mirrors what I enjoy with Bee and email via beemcp: todos from emails, full conversation context, and a system that gets smarter by seeing my workflow holistically.

### How to Implement It

Implementing an MCP Server is straightforward with available resources:

1. **Understand MCP**: Check out the [MCP specification](https://modelcontextprotocol.io/spec) for details.

2. **Build the Server**: Use [MCP SDKs](https://docs.modelcontextprotocol.io/sdks) (Python, TypeScript, etc.) to expose MLO’s data and actions (e.g., read todos, create tasks).

3. **Secure It**: MCP includes authentication to ensure only authorized AI platforms access the data.

For inspiration, the [beemcp repository](https://github.com/OkGoDoIt/beemcp) shows how this works for Bee and could guide an MLO implementation.

### Why It Matters

An MCP Server would make MLO a central hub for AI-driven productivity, enhancing its value across my devices and platforms. I’d be thrilled to see this integration and am happy to discuss further.

Thanks for considering this!

-Danny

[imajeff]

1. you forgot to write (try 1) on your first post, Lol
2. It's not clear how this AI stuff will work considering the unwanted eternal hype where it has never been nearly as helpful as everyone selling something seems to claim... as well as my experience being that AI only works for typical, but it fails for me because I'm not typical. May I point out that MLO is not typical either.

On Saturday, July 5, 2025 at 8:18:14 AM UTC-6 Daniel Clark wrote:

I’m a dedicated user of My Life Organized (MLO) and love how it keeps my tasks in sync across Android, iOS, Windows, and soon the web. I also rely on Shortwave Email and the Bee AI Wearable, where I’ve been using an MCP Server (via [beemcp on GitHub](https://github.com/OkGoDoIt/beemcp)) to seamlessly integrate my email and Bee. This setup transforms email into an intelligent extension of my workflow, allowing AI to create todos in Bee from emails, reference past decisions, and understand my priorities—all without switching platforms. I’d love to extend this capability to MLO by adding an AI MCP Server, and I believe it could take MLO to the next level.

### What is an AI MCP Server?

The Model Context Protocol (MCP) is an open standard from Anthropic that lets AI models connect to apps and data sources through a universal interface. An MCP Server for MLO would expose its todo lists, priorities, and actions to AI assistants (e.g., in email or wearables), enabling them to read and interact with MLO data securely and efficiently. Learn more at Anthropic’s [MCP introduction](https://www.anthropic.com/news/introducing-the-model-context-protocol).

 [...]

[Daniel Clark]

All I can say is that so far the AI from my Bee Wearable (http://bee.computer) combined (via MCP) with Shortwave email has been amazing. It knows about most things I say and write, plus records family history etc. I recently realized how valuable this will be after people I love pass away - I'll have access to facts about family history I'd be unlikely to write down or remember otherwise. And it is good at answering questions based on this knowledge base.

The only thing I find annoying is that I now have multiple places with TODOs. Hence the suggestion.

--
You received this message because you are subscribed to the Google Groups "MyLifeOrganized" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mylifeorganiz...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/mylifeorganized/debb6373-9d2c-4cf7-af19-6641a353c25fn%40googlegroups.com.

[imajeff]

Now, I'm in no way speaking for the MLO team. I have nothing against anyone trying to make good use of modern AI, I've been studying AI since the 80's and it's only that I understand the problems apparently better than those who said basically we developed it but we don't know how it works.

MLO team is doing it of course. You should look for the recent conversation on https://groups.google.com/g/mylifeorganized/

where they have already responding to the question and said they are working on the Web version of MyLifeOrganized, it was already announced that will have a good API, but they said that will be the API to support AI integration. 

Fingers crossed, we'll see how it goes

[Hammy Havoc]

For anyone reading in future, whilst I'm very much anti-unethical AI in terms of training and ripping off hard-working folks with gen AI, if an MCP or API integration actually worked, I would be interested in seeing how someone runs a local LLM to organise their tasks and make sense of it all from an existing workspace. For this to not be counterproductive, it would need to be Clippy-like suggestions about moving some tasks to certain projects or making them sequential sub-tasks of another. But this would sincerely need to work well (and I don't think a next-word prediction model is going to cut it) or it would be as annoying and mostly useless as Clippy. Would love to be pleasantly surprised, but I don't think it'll materialize as I personally feel LLMs are a dead-end technology for most of the problems they're being thrown at.

You could probably do it without an LLM just by using RegEx for keywords and looking at dates and times when tasks were created, or looking at the archive of tasks that are already complete, looking for patterns in when they're created, what order they're completed in, context etc.

The API excites me quite a bit.

[ad-mx]

I'm pretty sure that mcp and AI integration is necessary nowaydays because large volume of generated code also generates need for tasks the dev has to perform around that code => having mcp in task app would be life-saver, agent just adds them to the inbox or proper project with right annotations at the end of its session. MCP is basically API agent understands and it works exceptionally well.

Best regards

AD

[Jeff Smith]

If you understand how AI works, you will know it is more hype than anything else. Good search engine, if you're willing to make sure you learn everything so that you spot what is not correct that it spit out. In the end, they focus on hype because it sells and they have spent WAY WAY more money than they should have (which by the way decreases value of the dollar)....

I will admit it is quite impressive, the amount of programming and expensive retraining going into it so there are big results, just not as useful as the entire public is being told. Modern business is now all about cash grab before the general public finds out it wasn't going to be worth it. A decade ago and 10 decades ago they were always pushing excitement saying "just around the corner". Now that they are saying "we finally made it around the corner" here we are being told that any moment now we will have the amazing things in our own hands but I still only hear rumors that it finally exists... somewhere I have not yet seen for myself. Convince me by showing me the actual value, not merely showing some specific example that doesn't relate to what I need AI to do. Not just say "if you give us more resources it will finally be in your hands."

This thread is a very example of what I'm talking about. One day GPT 3 started spitting out words that it heard intelligent people say and you said wow this AI is now intelligent. The worst thing you could have done is fall all over that Sam guy just because he said, "We don't know how it works"... This aint the movies so it never was magically intelligent now from lightning or something that would not cause that. 

On Wed, Jan 14, 2026 at 5:06 AM ad-mx <chie...@gmail.com> wrote:

I'm pretty sure that mcp and AI integration is necessary nowaydays because large volume of generated code also generates need for tasks the dev has to perform around that code => having mcp in task app would be life-saver, agent just adds them to the inbox or proper project with right annotations at the end of its session. MCP is basically API agent understands and it works exceptionally well.

Best regards

AD

środa, 22 października 2025 o 23:06:31 UTC+2 Hammy Havoc napisał(a):

For anyone reading in future, whilst I'm very much anti-unethical AI in terms of training and ripping off hard-working folks with gen AI, if an MCP or API integration actually worked, I would be interested in seeing how someone runs a local LLM to organise their tasks and make sense of it all from an existing workspace. For this to not be counterproductive, it would need to be Clippy-like suggestions about moving some tasks to certain projects or making them sequential sub-tasks of another. But this would sincerely need to work well (and I don't think a next-word prediction model is going to cut it) or it would be as annoying and mostly useless as Clippy. Would love to be pleasantly surprised, but I don't think it'll materialize as I personally feel LLMs are a dead-end technology for most of the problems they're being thrown at.

You could probably do it without an LLM just by using RegEx for keywords and looking at dates and times when tasks were created, or looking at the archive of tasks that are already complete, looking for patterns in when they're created, what order they're completed in, context etc.

The API excites me quite a bit.

On Wednesday, 16 July 2025 at 17:30:45 UTC imajeff wrote:

Now, I'm in no way speaking for the MLO team. I have nothing against anyone trying to make good use of modern AI, I've been studying AI since the 80's and it's only that I understand the problems apparently better than those who said basically we developed it but we don't know how it works.

MLO team is doing it of course. You should look for the recent conversation on https://groups.google.com/g/mylifeorganized/

where they have already responding to the question and said they are working on the Web version of MyLifeOrganized, it was already announced that will have a good API, but they said that will be the API to support AI integration. 

Fingers crossed, we'll see how it goes

--

 "It is not what I believe that will make-or-break; It is what I'm doing about it." --Jefferson Smith

[Alexander W.]

IMHO... adding MCP to the mix would be more harmful than useful.
---

MLO works perfectly without any cloud if you use the core stuff only.

Moving this to a browser is - IMHO - a very bad idea.

The UI for Windows PC is great and enables most sophisticated use cases and scenarios.

My take would be as a customer who loves and lives with MLO - do not move away from the core value propositions of MLO for daily handling and usecases.

---

what is possible and works fine is exporting MLO-XML and then work with AI on it. That makes sense, but only for insights creation and adding the widsom of AI - which is the collective knowledge of humanity - in a AI-fuzzy way to the mix and use the haluzinations for creativity injection ;).

---

currently AI is not able to really "think" or execute precisely stuff - yet.

Thats way using it for the core features of MLO would hurt a lot ;).

just my2cents - A.

[Daniel Clark]

The Shortwave email client is AI and is super useful. I've never been able to do Inbox Zero before, but now I can. It is very good at creating todo items and calendar events from emails, easily referencing numerous different email threads. Being able to interface it's TODO list with MLO's would be awesome. Increasingly the value of software is in its ecosystem, not it as a standalone thing. MCP will open MLO to be part of an entire ecosystem of AI tools. The AI part isn't even that important, for my use case at least it would just be a way to maintain sync. This can of course be done with non-AI code, the difference is that with AI the "glue code" can be a set of prompts created by a non-programmer user instead of a developer using an API.

--

You received this message because you are subscribed to the Google Groups "MyLifeOrganized" group.
To unsubscribe from this group and stop receiving emails from it, send an email to mylifeorganiz...@googlegroups.com.

To view this discussion visit https://groups.google.com/d/msgid/mylifeorganized/4f9224a4-9dce-488d-9a2c-8a4786925915n%40googlegroups.com.

[Joel]

Completely support this idea.  AI integration is long overdue. The human mind is not designed to be burdened with the mundane.  My Life Organized is uniquely the best platform for it's implementation.  

AI is a massive opportunity for GTD implementation improvements.  

[Daniel Clark]

Re: AI, another data point is that the https://goblin.tools/ app is like #4 in paid productivity Android apps. It uses AI to break down tasks into small steps automatically.

To view this discussion visit https://groups.google.com/d/msgid/mylifeorganized/2cec225d-31b5-4c94-b065-58abb18ac986n%40googlegroups.com.

[Daniel S]

Is MCP happening?  I'd very much like to have AI assistant manage my MLO tasks.

Open Claw and/or Claude assistant with MCP connections.

I'm just getting into Shortwave too.

Just having Claude be able to review emails, notes in notion, then send a message to team in Slack is huge.  Being able CRUD to MLO would be a very big win for that ecosystem 

[Johan Aubry]

Hi,

I fully support this idea. An MCP integration would make a lot of sense for MLO, especially given how structured its data already is.

One thought: Cloud Sync is already an always-on service that manages user data across devices. If MCP could be exposed (optionally and securely) at that level, it might be a natural integration point for automation and external tools.

This would enable use cases like creating tasks from external systems, syncing workflows, or building custom automations — without changing MLO’s core experience.

Of course, keeping this optional and under user control would be key.

Very interesting direction!