Dear Reader,

Yesterday. I shared with you how my PhD defence ended up being a stress test for my note-taking system and the AI agents that run on top of it. (Click here if you missed it)

But what are AI agents, and how exactly can they help in your academic work?

To explain it in one analogy:

They are technically very similar, but while a calculator performs a single step, an AI agent executes many steps to achieve a specific goal. Just like a piece of software on a computer written for a specific purpose. This makes agents, naturally, far more versatile, autonomous and powerful.

The key difference between an agent and a computer in this analogy is that computers don’t program themselves, but AI agents do.

But what can agents do for YOU, in your day-to-day work?

• They reveal the big picture: Import a PDF, run it through an agent, and it will find the bits relevant to your research based on your other papers and notes.
• They connect the dots for you: Give an agent access to your notes and PDFs, and it surfaces links between ideas you wrote months apart and relates them to the meeting you had back in January.
• They find the tiniest of details: Agents can compare 100s of papers and notes to, for instance, identify that one paper that worked with that exact method you are trying to implement.
• Organise your notes/papers: Agents understand the content of your notes and connect to your Zotero collection. If you give them clear outlines for how you want your digital information structured, they can do it for you.
• Help you write: Using AI for academic writing is not new, but agents take it to another level, as they can help you outline papers, based on your notes and papers, for the first time crafting a narrative.

This list can go on and on. But let me show you the five use cases that helped me prepare my PhD defence in just a few days and explain what an AI agent is in practice in this video:

This video is part of the Effortless AI Agents Course, which I am releasing today. All of these examples are taken directly from my PhD defence. Once you get used to using agents, you can run them in parallel and move at a speed in your research, unmatched by anything in the past.

That's what saved me.

PS: So how did my PhD defence go? It was fun; some said I was a bit combative, which, however, landed well. My observation about people is that mirroring their communication style usually works best. So examiner #3 was quite harsh — I met him there, and it was a very fruitful discussion! He honestly helped me develop my own arguments, and I ended up submitting an abstract on our discussion to the NZ Ecological Society the same day.

New Course: Effortless AI Agents for Academics

This course will teach you how to set up a system of AI agents running alongside your academic notes and papers. The agents know your research, can find connections to notes and papers you touched years ago, dig through your Zotero library, find literature online, and do any organisational chore you can imagine.

Part 1: Setting up an AI system for academic work

The technical foundation to run AI agents. You only need to set it up once, and I am here to help (read about the founding cohort below). By the end of Part 1, you have a working AI that can already search, read, and reason across your notes, your Zotero library, and the web.

• Connect AI with your notes & papers: Using Obsidian and an AI plugin, your vault becomes a workspace where a research assistant reads and writes right alongside you.
• Connect to Zotero & Consensus: Using so-called MCP servers, you can give your agent access to your local Zotero library and the Consensus search engine (over 200 million+ papers online), making the assistant capable of helping with any literature review.
• Change things without fear: When you give an agent access to your notes, you want security that it won’t change or delete anything. I will show you how to set up visual software that tracks every change to the character and lets you undo anything your agent does if needed.

Part 2: Teaching AI how you work

This part turns your generic AI into one that understands your conventions, your rules, and red lines. These policies are essential to avoid losing cognitive sovereignty to AI. I provide you with a step-by-step workflow script so you don’t miss anything important.

• Don’t explain yourself: Define how you structure your notes, where important settings are found, and how your agents should talk to you. Record everything in a file once, and the prompting shrinks.
• Define a clear PDF policy: Make sure your AI always works with full-text PDFs when asked a question and does not guess from the title or abstract.
• Define a clear AI policy: AI-generated content (like summaries) should never be mixed with your thoughts and should be clearly marked. An efficient policy file makes sure your AI always honours this decision without being reminded.
• Turn repetitive chores into one click: Capture any workflow once and turn it into a skill you reuse forever.

Part 3: Using AI Agents for Academic Work

A showcase of real academic workflows I use in my work that you can copy, adapt, and improve for your own domain. By the end of Part 3, you have a library of workflows and the confidence to invent your own.

1. Everyday agent moves: Deep search across your vault, detailed look-ups, synthesis across many sources, and turning messy PDFs into clean, linked notes.
2. Connect & import sources: A connector skill that cross-references new material against your existing notes, and a Zotero importer that pulls papers, reads their full PDFs, and files them into your network automatically (in bulk, if you like).
3. Systematic literature review: A scoring skill that ranks papers against your own criteria table, so you can screen a whole collection in minutes.
4. Visual thinking: Turn a long, tangled text into a timeline or mind map, or let AI build a lit review matrix using Obsidian Bases without touching code.
5. Automations using AI: Now you don’t need technical know-how to configure powerful plugins (like QuickAdd) that make working with Obsidian frictionless.

Release Bonus: Free Founding Cohort Membership (only for the next 4 days)

Setting up AI agents is still quite technical, even in 2026. So, as a bonus, I decided to invite everyone into a founding cohort. This will be an exclusive space to get support from me and to share what you accomplish with others. In the space, I will also upload a few more videos on topics you request. So if you want a more transformative experience and peers to share the journey with, get the course in the next few days, and you’ll automatically be added to the founding cohort space.

Tomorrow, I want to show you how you can set up this system and give you some free tutorials to get started, even if you don't want to join the course yet.

Until then, wishing you my very best, Ilya

Ilya, The Effortless Academic

I will show you exactly what I mean by “AI system”. If you have been following EA for a while, you are probably familiar with my tendency to think in systems and workflows.

A system is multiple components working together to create more than the sum of their parts.

For instance, an AI chat is useful on its own. But if you give it access to your Zotero library, you can suddenly do much more (organise it, scan it, export it, etc).

The AI system I have developed over the last few months is exactly that. Make sure to read the email tomorrow if you want to set it up.

Previous parts of this series​
Part 1: My PhD Defence: 6 days to rebut 30 pages of critique​

This email is part of a 4-day series on AI agents. If you don't want to know about it, unsubscribe from this series.​
Want to leave forever: Unsubscribe from everything (never hear from me again)