🚀 Introducing beemine.ai - AI-powered Beeminder companion

[troylar]

Fair point (and this is why we’re in free beta mode).

This language was based on my original intent to require user encryption key for maximum security (which we still could do).

I’m changing the language on the security section today:

What it currently says:

End-to-End Encryption - All sensitive data, including your API keys and credentials, is encrypted using industry-standard AES-256 encryption before being stored. Even we can’t see your encrypted data.

What it will say:

Secure Server-Side Encryption - All sensitive data, including your API keys and credentials, is encrypted at rest using industry-standard AES-256 encryption. We decrypt and use your API keys only to make authorized AI calls on your behalf, following standard practices for SaaS automation platforms.

Does this provide better clarification?

Thanks for correcting me!

[troylar]

One of the features I’m adding is an “beginner mode” user toggle.

When we show Beeminder jargon, the words will have an indicator that let you hover over the word to see an explanation.

This will also apply to the AI where it can break down your goals in Beeminder terminology, but explaining how the terms apply to your specific goal.

[troylar]

I’m moving the AI analysis page to a sidebar chat.

You can can @mention your goals to ask about specific things.

I’m also working on letting it take actions on your goals as well.

I’d love feedback here . . . what kinds of things would you like to ask/do?

[redacted]

That works!

This language was based on my original intent to require user encryption key for maximum security (which we still could do).

Seems likely to be more trouble than it would be worth (IMO).

[troylar]

BeeMine AI Vision v0.2.0

I’m still working on the foundational features before I start adding the really cool stuff.

This release provides a couple new features that I’ve personally wanted for a while.

Intelligent Data Backfilling
When I add new goals, I don’t always have the time to setup my automation and I tend to get too busy to update every day.

I’ve added a new feature that let’s me edit and add historical data quickly across several days.

You can add one day or an entire data range.

And if you pick a date range, you can choose to skip days for existing data and/or skip weekends.

Not only that, but you can edit and delete existing datapoints as well.

And if you’re making lots of changes, don’t worry!

Your changes are staged, so you can review all of your changes before you commit!

Goal Assistant
The other big feautre is the Goal Assistant sidebar.

This is JUST in its infancy .

Play around with this and let me know what you think.

I’ve added the ability to @mention your goals for context.

Right now, NOTHING is saved in the database–all of the history is stored in your browser.

I’ll clearly notify you when we start saving conversations so you can choose to opt in.

Thanks for all the feedback so far—keep it coming! Let me know how the new chat workflow and history batching fit into your Beeminder routine and what you’d like to see in v0.3.0.

Written with AI

Release date: 2025‑09‑26
Live at: https://beemine.ai
Changelog: beemine.ai - AI-Powered Goal Tracking for Beeminder

TL;DR

• Persistent sidebar chat keeps the AI assistant on screen while you work.
• Goal history upgrades let you add, edit, delete, and batch datapoints with unit-aware validation.
• Batch staging workflow previews multi-day changes before you commit them.

What’s New

Added

• Chat — Sidebar companion that stays open next to the dashboard, plus @goal mentions for goal-specific conversations.
• Dashboard · History — Rich editing in the goal history dialog with unit-aware inputs, multi-day range selection, and quick filters.
• Dashboard · History — Staged batching so you can review additions, updates, and deletions together before saving.

Changed

• Marketing — Security copy now clarifies that API keys are encrypted server-side with AES-256 and only decrypted to run authorized AI calls (no end-to-end claims).

Why It Helps

• Stay in flow: The chat stays visible, so coaching and context follow you across goals.
• Confident edits: Unit-aware validation and previews reduce mistakes when backfilling data.
• Batch efficiency: Queue a week’s worth of updates, sanity-check them, and push once.
• Transparent security: Messaging now matches the actual encryption model, avoiding overpromising.

Try It in 60 Seconds

1. Visit https://beemine.ai and open the Goals Dashboard.
2. Start a conversation in the sidebar, mentioning a goal with @slug.
3. Open the goal’s History dialog, pick a date range, and stage a few datapoint changes.
4. Review the staged list and hit Save changes.
5. Check the marketing page for the updated security explanation.

[troylar]

The next foundational feature I’m building into Beemine is Intelligent Goal Health Score.

When I’m look at my goal dashboard, I see a static chart that shows me:

• What I’ve done
• Where I am
• When I might derail

What’s missing?

• WHY does it look like that?

and

• HOW should I interpret my data and translate it into forward-moving actions?

Intelligent Goal Health Score will let you create custom rules around your goals to help you analyze your goals the way YOU want to by your own custom scoring mechanism.

You can create global rules that apply to all of your goals as well as custom rules that apply to named filters that you create in the dashboard (by tag, by name matching, etc.).

Then on your goal cards, you’ll see your current score based on your active rules.

Here are some examples of questions you’re scoring can help you surface:

• Is this goal prone to derailing more than others?
• Are there gaps in my data?
• Was my last datapoint longer than X days ago?
• Are my daily datapoint values for this goal averaging below X?
• Are my datapoints being entered on a regular cadence or am I always playing catchup?
• Is my safety buffer exceeding X?

To start, I’m going to keep it simple with a simple scoring mechanism focused on Beeminder-specific metric interpretation.

Beyond that, I want to start thinking about how we can create a layer above the Beeminder goals that gives us a more holistic view using a similar scoring mechanism.

• How am I doing with my health?
• How am I doing with my personal growth?

I’d love to get feedback and suggestions on this!

[troylar]

If you were wondering if Beemine.ai was just another fly-by-night-flash-in-the-pan . . . you were very wrong.

I’ve spent about 30 hours in the past two weeks toiling over some really amazing new features that I’m excited to use myself.

I’m in this for the long .

My goal is to release this week, but I’m also working a full-time job and a side hustle, so that’s tentative.

Here’s a quick run-down of the new features . . .

Feature #1: Smart Coaches

The “AI Insight” tab has been replaced with a real sidebar chat feature.

You can now have conversations about your goals and even ask for advice.

But I need a crusty old war-torn veteran to kick me in the pants.

Use the Tough Love coach!

How about a no-nonsense ‘say it like it is’ kinda person?

Try the Minimalist or Data Analyst!

Here’s an example of the Tough Love coach:

Feature #2: Take Action in Chat

I’m REALLY excited about this one.

You can now ask the chatbot to modify goals, add/remove/update datapoints, step down . . . pretty much anything you can do in the API.

Feature #3: Beeminder AI Wisdom

Behind the scenes, I’ve LOADED up the AI knowledge base with GADS of wisdom from The Guru @dreev himself.

To start, I pulled in all of the help and every blog post into single markdown file.

Obviously, Beeminder has changed over the years and we can’t blindly use everything.

I’ve been crafting and tweaking and honing the knowledge to distill the best practices, suggestions and general all-around Beeminder street smarts.

• Help me interpret my goals.

• What other ways am I NOT using Beeminder?

• Am I using it the right way for my goals?

Feature #4: Context-Aware Goals

There are seven goal types.

By default, AI sees datapoints and a goal, so it assumes the datapoint should lead to the goal.

Obviously, that’s NOT how we want to think about our goals.

I’ve been working really hard at teaching the AI to understand the nuances of the different goal types and respond based on those nuances.

I’m also updating the dashboard to show a context-aware view of each goal so that the action deadline is appropriate for the goal type.

(UI is not done yet)

Welp, those are the new features coming this week!

This is JUST the beginning . . . I have a massive backlog that is going to extend beyond Beeminder, but more on that later . . .

[dreev]

@troylar

Behind the scenes, I’ve LOADED up the AI knowledge base with GADS of wisdom from The Guru @dreev himself.

Ha! I’m getting way too much credit here. The help docs are mostly @shanaqui (painstakingly curated, refreshed, and refined for years now, based on what minimizes real-world user-confusion) and the blog is often as much @bee as me. (If the blog ever makes you laugh out loud, it was probably Bee’s contribution.) Not to mention all the Nicky posts and other workerbee posts and guest posts. Big team effort there.

(Now I’m marveling at the 400+ blog posts spanning almost 15 years. Very proud of that thing! Also I think it’s wonderful that you’re putting it all to use getting these coachbots to understand Beeminder so deeply.)

Anyway, amazing work on all these Beemine.ai features. You’re putting us to shame! Especially doing this as a side project.

PS: I was curious about the blog authorship breakdown and asked the robominions to click through the archives and pie-chart it. I can’t vouch for these numbers but I guess they’re plausible:

So I see how you accidentally gave me too much credit. To reemphasize, many posts where I’m the official author are heavily assisted by Bethany (and I suppose vice versa). No AI [writing] assistance yet though. Unless you count this bit from the 2022 post on self-blackmail:

But I’m very biased so… let’s see what GPT-3 thinks?

No, Beeminder is not self-blackmail. The whole point is that you’re making a commitment to yourself, and then holding yourself accountable. If anything, Beeminder could be seen as a way of rebuilding trust in your future self.

Mic drop. (I’m still marveling at how this AI, having ingested the whole internet, knows enough about Beeminder to describe it coherently and often correctly in its own words.) But in case you find that unsatisfying, let’s hear from some real live human users…

PPS: When I say no AI assistance on the blog I mean writing-wise. Our daughter used to illustrate the blog until AI destroyed her job. Anyway, back to Beemine.ai and whether AI is about to steal our jobs…