AI Agent Vocab 101: The Terms That Actually Matter

Ilya Gordey — Sun, 22 Mar 2026 21:30:22 +0000

Smart people keep asking me the wrong questions about AI agents. Not because they're not smart — because they're missing the vocabulary.

───

I've been building AI agents that trade real money on prediction markets. And every time I explain what I'm doing to someone outside the AI bubble, I hit the same wall: they don't have the words to ask the right questions.

"So it's like a bot?" Kind of. "Is it safe?" Depends what you mean by safe. "Who's in control?" That's actually the most important question — and most people can't even frame it yet.

Here are the terms that changed how I think about all of this.

───

🧠 What an AI Agent Actually Is

LLM (Large Language Model)
The brain. Trained on enormous amounts of text, it predicts what comes next — whether that's a word, a sentence, or a decision. GPT, Claude, Gemini: all LLMs. An LLM alone just talks. It can't do.

Agent
An LLM plus tools plus a loop. The defining difference: an agent can affect the world outside the conversation. It can read your email, place a trade, send a message, run code. If an LLM is a consultant giving advice, an agent is the assistant who actually books the flight.

Skill / Tool
What the agent can reach for. Search the web. Check a wallet balance. Place a Polymarket order. Each capability is a "tool" or "skill" — discrete actions the agent can choose to invoke. The agent decides when to use them. You decide which to give it.

───

⚙️ How It Thinks

Token
LLMs don't read words — they read tokens. Chunks of text: sometimes a full word, sometimes a syllable, sometimes punctuation. Everything about capacity and cost is measured in tokens. When people say "this model has a 200k context window" — that's 200,000 tokens, roughly 150,000 words.

Context Window
The agent's working memory. Everything it can "see" at once: the conversation history, the documents you fed it, the tools it has, the task you gave it. When the window fills, old content gets dropped. This is why long-running agents sometimes "forget" earlier instructions — they literally ran out of room.

Hallucination
When the model generates something confident, fluent, and wrong. Not lying — it has no concept of truth. It's pattern-matching on what a plausible response looks like. For a writing assistant, hallucinations are annoying. For an agent managing money, they're dangerous. This is why agent design matters: you constrain the agent so a hallucination can't cause real-world damage.

🏗️ How Agents Are Built

System Prompt
The instructions baked in before the conversation starts. This is where you define the agent's personality, constraints, and goals. "Never delete messages." "Only trade markets with >$50k liquidity." "Always ask before spending more than $100." The system prompt is how you control what the agent does with the power you've given it.

Harness
The scaffolding around the LLM. System prompt, tool definitions, memory, error handling, retry logic — everything that turns a raw model into a useful agent. The model is the engine. The harness is the car. A powerful engine in the wrong chassis will still crash.

MCP (Model Context Protocol)
A standard for connecting AI agents to external data — files, calendars, databases, APIs. Think of it as USB-C for AI

I Built an AI Agent That Trades Polymarket 24/7 — Here's How It Works

Ilya Gordey — Sun, 22 Mar 2026 21:18:31 +0000

Prediction markets. Whale wallets. And the question that kept me up at night.

It Started With Watching Wallets

I was manually trading on Polymarket. Losing, mostly — not because my analysis was wrong, but because by the time I'd read the news, checked the odds, and placed the trade, someone had already moved the market.

Then I started looking at on-chain data. Some wallets were consistently early. 58%+ win rate across hundreds of trades. Not lucky — systematic.

They weren't smarter. They were faster. And they clearly had automation.

So I started building mine.

What I Learned Building an Autonomous Trading Agent

Getting the data is easy. Acting on it is hard.

Polygon is public. You can read every wallet, every trade, every timestamp. The signal is all there — whale opens a position at 2am, market moves 20% by morning.

The problem is execution speed and geo-blocks. Polymarket blocks most of the world (including where I live — Thailand). The CLOB API is finicky. And placing an order requires: USDC approval, wallet signing, order construction, submission — four steps that need to happen in seconds.

I ended up routing everything through Vercel's Tokyo region (hnd1) — the one datacenter Polymarket doesn't block. It's janky but it works.

Non-custodial was non-negotiable.

I didn't want to hold user funds. So every user gets their own Polygon wallet, private key encrypted with AES-256-GCM, stored only on their side. The agent signs transactions locally. We never touch the money.

The leaderboard changed everything.

When I added a public leaderboard showing which agent strategies were performing — people started competing. Optimizing not just for profit but for rank. It turned a trading tool into a game.

The Bigger Realization

Halfway through building this I had a thought that I can't un-have:

People already deposit money into AI agents. They just don't call it that.

Your Meta Ads budget? That's a deposit into an AI agent. It decides targeting, bids, timing — autonomously. You fund it and hope it earns more than it spends.

Polymarket is the same model, but transparent. Every trade is on-chain. You can audit exactly what the agent did and why.

We're at the beginning of a shift where "giving money to an AI and letting it run" becomes normal. Not just for ads. For trading, for content, for operations.

I don't know how fast this happens. But it's already happening.

Where It Is Now

The tool is live at polyclawster.com. Early, rough around the edges, but real — real wallets, real trades, real on-chain history.

Building this in public. Sharing the wins and the stupid mistakes along the way.

If you're doing anything similar — AI agents with real-world stakes, prediction markets, on-chain automation — I'd genuinely love to talk: https://twitter.com/ilyagordey

DEV Community: Ilya Gordey

AI Agent Vocab 101: The Terms That Actually Matter

I Built an AI Agent That Trades Polymarket 24/7 — Here's How It Works