Double Your Salary Plan: Let Hermes Help You Find the Money!

I built an AI that "finds money" for me every morning at 10 AM.

1. What Is Hermes Agent? And Why Can It Help You Find Money?

1.1 In One Sentence

Hermes Agent is an open-source autonomous AI agent framework developed by Nous Research, with v0.1.0 released in February 2026. It's not a "chatbot" — it's a personal AI worker that grows with you — the longer you use it, the better it understands you.

1.2 Hermes vs OpenClaw: What's the Real Difference?

Many people compare Hermes and OpenClaw, but they are fundamentally different things. OpenClaw is more like a "toolbox" where you manually configure each tool; Hermes is an "apprentice" that learns how to use tools on its own and even invents new ones.

Comparison	OpenClaw	Hermes Agent
Skill Source	User-written scripts	Auto-generated and optimized by the Agent
Memory Mechanism	Session-level memory	Layered persistent memory system (hot/cold storage separation)
Security Design	Hardened after the fact	Built-in from the ground up (prompt injection scanning, container hardening)
Iteration Speed	Regular updates	Iterated from v0.1.0 to v0.8.0 in 42 days [1]

Key Takeaway: OpenClaw is a temp worker you hired — Hermes is an apprentice you're training. The former leaves after the job is done; the latter gets better every time.

2. What Can It Do? Four Real-World Scenarios

2.1 Scenario 1: Daily Intelligence Butler

Imagine this: you set a single instruction — "Every morning at 8 AM, search Reddit and X for the latest open-source AI updates, compile a report, and send it to my Telegram."

A regular AI would execute mechanically, but Hermes is different. It observes your behavior: which items you clicked, which you replied to, which you ignored. Then, it quietly adjusts its filtering priorities. After three days, you'll notice the content pushed to you is increasingly on point. This isn't magic — this is behavioral learning.

2.2 Scenario 2: A Coding Partner That Never Forgets

What's the most painful thing for developers? Every time you switch to a new AI tool, you have to explain your project structure, naming conventions, and deployment process from scratch.

Not with Hermes. It remembers your codebase structure and where you got stuck last time. You close your laptop and go to sleep — it keeps working on a cloud VM. When you wake up the next morning, the progress report is already sitting in your WhatsApp.

2.3 Scenario 3: A Writing Assistant That Gets You

"I want a concise style, don't use the word 'boundaries'" — you said it once, and Hermes remembered.

It can remember your writing preferences, frequently used vocabulary, and off-limits topics across sessions. After a few uses, you don't even need to remind it anymore — it just knows what to write and what to avoid.

2.4 Scenario 4: A Researcher That Archives Methodologies

You ask Hermes to do a competitive analysis. It not only completes it but also automatically saves "how to do a competitive analysis" as a Skill. Next time you say "analyze Company XX for me," it directly calls the previous methodology, doubling the speed.

3. Hands-On Demo: Building a Side-Hustle Information Scraping System from Scratch

Prerequisites:

For WhatsApp deployment, refer to Tutorial 1: Deploying Hermes Agent to WhatsApp

For model configuration, refer to Tutorial 2: Integrating Xiaomi MiMo V2 Pro Model

Step 1: First Conversation — Tell It What You Want

Open WhatsApp and tell Hermes your core request directly:

"I want you to be my money-finding assistant, helping me find side hustles, remote work, or earning opportunities, filtering for high-quality, realistic opportunities."

Hermes will immediately start working. You'll see its "thinking process":

Step 2: Security Approval — It's Protecting You

When Hermes attempts to execute commands involving web scraping (such as curl | python3), the system will display a security warning:

Security Approval Interface

This is not an error — Hermes is protecting your server from unauthorized execution. Simply enter:

/approve session

This approves all similar operations for the current session. Using /approve always is not recommended, as you're still in the debugging phase.

Step 3: First Round of Results — Not Precise Enough

After approval, Hermes will deliver the first batch of results. But honestly, the first round is usually too broad:

"LinkedIn 13,000 jobs", "Upwork / Fiverr"…

Everyone knows these. The problem is:

No action path
No filtering logic
No prioritization

Step 4: The Key Move — Training Your Agent

This step determines success or failure. Start "training" it:

"This is not what I want. I do NOT want business setups like TikTok Shop, Shopee, or dropshipping. I want simple, individual tasks I can do online — translation jobs, video editing gigs, content writing. Must be something I can start TODAY."

Step 5: Output Evolution — It's Starting to Understand You

After several rounds of feedback, Hermes' output undergoes a qualitative transformation:

Now it gives you specific tasks, not platform lists. For example:

Task Type	Specific Content	Platform	Pay
Translation	English-Norwegian translation	Freelancer.com	$8-15/hr
Video Editing	30-60 second social media videos	PeoplePerHour	€30-250/batch
Content Writing	1500-2000 word SEO blog posts	PeoplePerHour	£55/article

Step 6: Set Up Scheduled Tasks — Let It Work Automatically

Now, have it scan for you automatically every day:

"Every day at 10am: Find 3-5 real-time gig tasks based on my preferences — posted within last 24 hours, low competition, actionable immediately. Send the results to my WhatsApp."

Now it pushes information to me daily and continuously optimizes based on my click behavior and feedback.

Step 7: Skill Consolidation — Making It Stronger Over Time

The final step: have it turn this workflow into a reusable skill:

"Turn this process into a reusable skill and improve it over time."

From now on, you have a self-evolving money-finding radar.

3. Five Core Capabilities of Hermes Demonstrated

Through this hands-on exercise, we validated Hermes' five core capabilities:

3.1 Remembers Preferences (Memory)

I only said once that "I prefer translation and video editing tasks," and it remembered. The next day, it started "choosing for me."

You can tell it your preferences anytime in WhatsApp:

I prefer translation and video editing tasks.
Avoid writing-heavy jobs.

3.2 Trainable (Feedback Loop)

Every time you receive a push notification, simply reply:

This one is good → It will push more similar content
This is useless → It will avoid this type of task

3.3 Takes Action for You (Actionable Agent)

Select a task and simply say:

Help me apply for this

It will write a proposal for you, tell you how to submit it, and even generate the application content.

Key Takeaway: It doesn't just find money for me — it helps me go get it.

3.4 Self-Evolving (Self-Evolving Skill)

You don't need to change any code. Just tell it:

Improve this system over time and make it more accurate.

It will adjust its filtering logic and improve result quality on its own.

3.5 Scheduled Pushes

Every morning at 10 AM, WhatsApp buzzes right on time:

"Today's actionable tasks: Translate a 300-word product description, $15, only 3 competitors, posted 10 minutes ago."

4. Training Tips: Lessons I Learned the Hard Way

After several days of training, here are my key takeaways:

4.1 The First Round of Output Will Always Be Off — Don't Give Up

The results Hermes gives you the first time are usually too broad. It's not that it can't do better — it just doesn't know you yet. You need to explicitly tell it what you don't want.

4.2 Be Specific with Feedback

Don't say "this is bad." Say "I don't want store-setup tasks. I only want per-task pay, something I can do today." The more specific you are, the faster it learns.

4.3 Use Security Approval Wisely

/approve session is your best friend. Don't use always during the debugging phase — save that for when things are stable.

4.4 Let It Consolidate Skills

Every time a complex task is completed, tell it to "Turn this into a reusable skill." This way, it can call it directly next time without re-reasoning from scratch.

I thought it would just give me information, but then it started telling me: "Here's what you can do today to make money."

I didn't build a tool — I trained an Agent.

Join Our Discord Community!

Scan the QR code to join the user discussion group and unlock a variety of practical and fun use cases!

🚀 Tencent Cloud Developer Community & Support

1️⃣ Tencent Cloud Developer Community

Unlock more advanced use cases on Discord

Click to join the community

2️⃣ Dedicated Service Support

Join via WhatsApp / WeCom for exclusive technical support

Channel	Scan / Click to Join
WhatsApp Channel
WeCom