Ample Agent Pro: Frontier LLMs for tasks, planning and more

For a couple years, we've been watching from the sidelines to observe the best of what the Notions of the world could deliver with AI. At first, it was mostly gimmicks. But if there's one stable truth of the AI Era, it's that version numbers will keep on ascending, and every new version from the feuding four unlocks new possibilities.



Why these four? Third-party LLM Leaderboards (like arena.ai's) make clear there's no single winner in the adolescence of LLMs. For better and worse, this is going to be a back-and-forth competition. As we've collected more-and-more examples of "high value LLM interactions," our AI wishlist finally got long enough that we had no choice but to implement the best ideas and share them with the community.


And so today, we're thrilled to announce the general availability of Ample Agent Pro!


link⚔️ The models will figure it out

How is Amplenote's approach to AI different from other note and task apps?


The struggle is intense, it's no time to lose your noodle. Image generated by Ample Agent Pro


Our approach is all about the LLM models. Specifically, "which providers?" and "how outdated?" Every note app with AI has many features in common: summarization, proofreading, etc. Our users will come out ahead if the implementation of these features permits seamless access to every newly released LLM. Users can swap between providers, depending on what capabilities the task calls for.


Our approach ensures that, whatever the niftiest tricks LLMs are doing in their latest benchmarks, you can access the same powers from within Amplenote.


link🧐 What it does

Philosophy aside, whether an $8/month AI subscription is worth your time boils down to rather prosaic concerns: does it solve real-world problems?


While the plugin still has some rough edges, we believe this v1.0 launch is teeming with value. It starts with the breadth of features from AmpleAI, then adds a handful of special agents to help research, act, and organize.


Ample Agent's v1 feature lineup includes (✅ = core feature, free within monthly usage quota):

Task Agent. Progress on your delegate-able tasks (included in "core" during its beta release).

Note Agent. Transform note content, or automatically apply relevant tags to a note.

Dream Task Agent. Receive task suggestions informed by your quarterly goals (via Dashboard).

Deep Research Agent. Orchestrate multiple AI models to collaboratively deep-dive any topic (beta).

Image Generator. Generate contextually-relevant pictures, graphs or diagrams.

Thesaurus. Context-aware alternate word suggestions.

Insert Related Notes. Insert notes with content similar to the current note (beta).

Grocery Timesaver. Sort your grocery list by store sections/aisles.

Summarizer. Summarize a note, or notes linked from a note.

Complete Thought. Select any text and let AI finish your sentence.

Language Refiner. Select a phrase, describe what you'd like improved, get suggestions

Chat Agent. Engage in a long-form conversation grounded in your note content.

Rhyming Suggestions. Select a word or phrase, get contextually-aware rhyming suggestions.

Search Agent. Deep dive through notes in search of specified criteria.

Suggested Tasks. Analyze a (meeting?) note, extract the implied tasks.


In semi-related news, we fixed colors not toggling between light & dark mode in published notes this month, and this author is feeling gluttonous with his formatting toolbar. 🥳


link🎁 What a subscription gets you

If you sign up for an Ample Agent Pro subscription, you'll receive:

Access to the 15 features listed above, plus new ones added regularly

Free access to core AI features (the features from the list above), up to a monthly quota

Free updates (monthly, if not weekly)

Access to consistently updated frontier LLMs (as of April 2026, ChatGPT 5.4, Opus 4.7, Gemini 3.1 Pro, Grok 4.20 beta)

Amplenote support for any issues experienced using the plugin

Access to voting board to guide what Ample Agent Pro feature/improvement gets built next (coming week of April 27)


link🪄 Conclusion

We're excited to finally submit our entry into the marketplace of brain-multiplying AI tools. For a more detailed rundown of Ample Agent Pro, or to purchase a subscription, click here to learn more.


We'll write a more comprehensive Amplenote update in the weeks to come, but you can always check out live product changelog if you're curious what else we've been working on. 😅




link🗄️ Footnotes

A little bonus content for niche audiences that want to understand how "Amplenote is thinking about AI strategy" in general, or developers who might want to profit from building/updating an Amplenote plugin.


link🌍 AI Strategy and Competitive Comparison

According to our deep research queries, there are a handful of note & task apps that currently allow access to the major LLMs:

Amplenote. I've heard good things about these guys?

Airtable. "Enterprise-grade multi-provider setup," which is to say, supports multiple esoteric enterprise LLMs

Asana. Limited multi-provider access. Its AI Studio workflow builder has some choices.

Capacities. Built-in AI assistant supports OpenAI, Anthropic, Google, Mistral, and xAI (Grok) via BYOK API keys.

ClickUp. Exposes a model selector in its "Ask AI" chat interface with some models.

Notion. Built-in AI support of OpenAI, Anthropic and Gemini (no Grok at press time)

These apps share our belief that 1) AI is essential to expediting work 2) different AI providers excel in different domains.


Variance between implementations comes mostly from two areas:

Update cadence. New point releases on these LLMs are happening at least quarterly. The latest Opus is rumored to score 20-30 percentage points higher than the Opus before it. Thus, being even a few months outdated comes at a large cost to "potential productivity."

Arbitrage decisions. It's tempting to sell a subscription that transparently calls different AI providers, while charging the user a fixed cost. This approach is very convenient, and equally unscalable. The better the AI gets, the more the user depends on it, the greater the incentive of the "transparent LLM provider" to direct the user to a nano model, even if their query would have been bettered answered with a more powerful choice.

According to deep research queries, it seems that only Amplenote, and perhaps Capacities, ascribe to the principles:


1. The "right" update schedule is "basically immediate."

2. The most user-friendly price to connect with frontier models is $0.00.


Our solution does have its tradeoffs. Namely, that power users almost certainly will want to provide at least one AI provider API key of their own, to permit unlimited access to the highest-leverage features.


We think the "pros" far outweigh that con, because any power user already has accounts with multiple AI providers. The payoff is that you don't have to worry that Amplenote is going to send your high-impact questions off to some underpowered LLM. And, if the next release of any model places it firmly ahead of competition, you'll be able to tap into that power within days of its release.


link💰 Paid plugins

One reason that we opted to implement Ample Agent Pro as a plugin was to create a system that allows other Amplenote plugin authors to monetize their work. We already see a ton of value that Amplenote users glean from our directory of plugins; we can imagine that the quality of plugins will shoot much higher still if developers have financial incentive to build the best plugins imaginable.


Our guidelines for authors who want to author a paid plugin:

Offer a free version of your plugin; earn some upvotes

Email hello@amplenote.com your proposal for what plugin you would like to sell: what functionality will it offer, at what cost?

Given the breadth of possibility that exists at the crossroads of "Amplenote's Plugin API" (now featuring access to external calendar events, mood ratings, and everything seen on Mission Control Dashboard) and the possibilities that modern LLMs can enable.


Plot twist

Thanks for stopping by the Amplenote blog. Did you know that the content of this "blog post" is just a plain old note, lifted from the author's Amplenote notebook? Rich footnotes, industry-leading to-do lists, and a security-first mindset make us a solid option for modern writers. Try it out yourself.

Comments

Max Erofeev Unlimitedstar said 10 days ago

hoooray!

Bill Harding starFounderstar replied 7 days ago

Thanks Max! Lmk which of the functionality you think you'll use most?

Lars Kindem Unlimitedstar said 9 days ago

Ordering page fails

kasimasi2 starFounderstar said 8 days ago

Waiting on the Search Agent. Pretty much all I've ever cared about re:AI integration to my second brain...

Bill Harding starFounderstar replied 7 days ago

Search Agent is now available. We're hoping to include it in the "core" features soon, just need to finish validating its cost.

rajeshb89 Unlimitedstar said 7 days ago

Will this have regional pricing as well or just standard USD pricing?

Frank Sloan starFounderstar said 2 days ago

I'd like to be able to use local models (in LMStudio or the like) with it as well. I also think having an much simpler task agent look at my highest scoring tasks and recommend revisions on a regular basis with something like this for context (https://amazingmarvin.com/blog/three-simple-rules-writing-great-tasks-to-do-list/) would be awesome.