How can I built an AI Agent, that messages me the top 3 economy news I need to know as an international management graduate
That is an excellent project to take on! Having a tailored automated assistant that delivers high-value economy news is a fantastic way to keep your edge as an international management graduate. Before we dive into designing and building the agent, I would love to learn a little more about how you would prefer to construct and receive it.
Do you prefer a no-code workflow approach (like n8n or Make) or a coding approach using Python (e.g., LangChain/LangGraph)? Also, what messaging app or channel (e.g., Telegram, Email, WhatsApp) would you like the agent to use?
I wondered whether I could use an AI Agent
You can absolutely use an AI agent for this. To give you the exact blueprint, we just need to choose how we want to build it and where you want your daily top 3 summaries sent.
I prefer a visual, no-code setup like n8n or Make.
A visual, no-code approach using platforms like n8n or Make is a brilliant choice! It lets you focus entirely on the logic of your agent — how it filters macroeconomics and international trade news — without getting bogged down in syntax errors. Before we map out the node-by-node architecture, let's nail down which platform you'd like to use and where you want your daily top 3 summaries sent.
What are the pros and cons of n8n vs. Make for this?
Let's break down how n8n and Make stack up against each other so you can choose the best engine for your news agent. Both are incredible visual workflow builders, but they approach execution, pricing, and AI integrations with very different philosophies.
n8n vs. Make: The Core Differences
To make this comparison concrete, we can look at three critical pillars: pricing structure, self-hosting flexibility, and how they handle AI nodes.
Here is a quick snapshot of how they compare across these areas:
| Feature | n8n | Make |
|---|---|---|
| Core Philosophy | Node-based, highly developer-friendly | Visual mapping, highly intuitive for beginners |
| AI Native Nodes | Advanced native LangChain & Agent nodes | Basic HTTP & direct LLM API modules |
| Hosting | Cloud or Self-hosted (completely free) | Cloud-only |
| Pricing Model | Active workflow & execution-based | Per-operation billing (rapidly consumes quota) |
1. Pricing & The Self-Hosting Superpower
Make operates on a strict cloud-only model where you pay per "operation." Every single step in your workflow—fetching a feed, filtering a headline, passing text to an LLM, and sending a Telegram message—costs one operation. If you fetch fifty news articles a day, you will burn through Make's free tier of 1,000 monthly operations in just a few days. n8n is famous for its fair-code self-hosted option. You can run n8n completely free on your own local machine, a Raspberry Pi, or a cheap cloud server. If you do choose their hosted cloud tier, they bill based on executions (runs of the entire workflow) rather than individual operations, which is significantly more generous for data-heavy tasks like news aggregation.2. Node Ecosystem: Native AI vs. Direct APIs
When building an actual , n8n is currently the stronger platform because it has a dedicated, built-in advanced . It features specialized nodes for LLMs, persistent memory, vector databases, and agentic tools natively. You can literally drag and drop a "Tool Agent" node, wire a chat model to it, and hand it a Google Search tool without writing a single line of orchestration code. In Make, there are no native "agent" or memory nodes. Instead, you have to manually configure direct HTTP requests or use the standard OpenAI/Anthropic modules to handle the completion loops yourself. It is absolutely doable, but it requires you to build the agentic "thinking loop" manually.Given that you want to construct an AI agent that filters and processes a lot of incoming news, n8n's native AI toolset and generous execution model make it a highly compelling choice. However, Make is incredibly fast to set up if you do not want to deal with hosting.
Which of these platforms feels like the right fit for your project?
How much will it cost me to set this up?
Let's pull back the curtain on the actual numbers. Because we are looking at a visual setup, your ongoing costs are going to be split between two main buckets: the workflow engine (running the nodes) and the **LLM
The Real Cost of Daily News Automation
Your overall budget depends heavily on whether you choose to self-host n8n or use a cloud tier, plus how many tokens you consume when analyzing the daily macro news. Here is exactly how the math breaks down.
| Cost Component | Path A: Self-Hosted n8n (Recommended) | Path B: n8n Cloud | Path C: Make.com (Core Plan) |
|---|---|---|---|
| Platform Subscription | $0 / mo (Free software) | $20 / mo (Starter tier) | $9 / mo |
| Hosting Infrastructure | $0 to $5 / mo (Free local machine or cheap VPS) | Included | Included |
| LLM Token Usage | $1 to $3 / mo | $1 to $3 / mo | $1 to $3 / mo |
| Total Estimated Cost | $1 to $8 / month | $21 to $23 / month | $10 to $12 / month |
Let's Do the Token Math
To figure out your actual API costs, we have to look at the volume of text your agent will process. Let's assume you fetch 30 global economics headlines each morning. You feed those 30 headlines into a lightweight, highly efficient model like GPT-4o-mini or Gemini 1.5 Flash to filter and select the top 3 most critical events. This process is highly optimized, costing roughly 0.015 cents per execution. Once the top 3 articles are chosen, your agent fetches their full content (averaging 1,500 words or ~2,000 per article). It passes these 6,000 input tokens to the LLM alongside a system prompt engineered for executive analysis, generating a detailed 500-token summary output. Using standard API rates, this high-quality analysis costs about $0.03 to $0.05 per day. Over a month, your total AI "brain" cost is only about $1.00 to $1.50.If you go the self-hosted n8n route, you can run the agent completely free on your laptop to test it out. If you eventually want it running 24/7 in the cloud without keeping your laptop open, you can host it on a virtual private server like Hetzner or DigitalOcean for about $4 to $5 a month.
Now that you see the cost breakdown, does the self-hosted n8n path sound like a fun option to explore, or would you prefer the absolute zero-maintenance setup of n8n Cloud or Make?