DEV Community: Amr Mohamed

7 Side Hustles That Actually Pay in 2026 - Ranked by Time-to-First-Dollar

Amr Mohamed — Fri, 16 Oct 2026 18:23:21 +0000

7 Side Hustles That Actually Pay in 2026 — Ranked by Time-to-First-Dollar

Most "make money online" lists are recycled. They tell you to start a blog, do surveys, or drop-ship, but they skip the real question: how fast does the money actually land?

I've tracked what actually produces income over the last 18 months so you don't waste another weekend on dead ends. Here are the seven side hustles that consistently pay in 2026, ranked by how quickly you can expect your first payout — and what I genuinely think is worth your time.

How I ranked these

Most listicles never show the framework. Here's mine:

Time-to-first-dollar — How long from "started" to "money in the bank"?
Ceiling — Can this realistically hit $1k/month, or is it capped?
Barrier to entry — Skills, accounts, or upfront cost required.
Compounding — Does the work build an asset (audience, clients, store) or is it pure time-for-money?

The sweet spot for most people: fast first dollar, decent ceiling, and something that compounds. Let's get into it.

1. Freelance writing / content at scale

Time-to-first-dollar: 1–3 weeks | Ceiling: $3k+/month | Compounds? Yes

You don't need a degree or a portfolio of 20 polished pieces. What you need is consistency and one channel that gets you paid.

Where I'd focus in 2026:

Content marketplaces that pay per word for quick first checks.
Outreach to niche sites in personal finance, health, and SaaS — these three verticals consistently buy SEO content.
Programmatic content — set up a simple system (templates, keyword lists, AI-assisted drafting with a human editing pass). One person can put out 5–10 sellable articles a week.

Realistic math: 5 articles × $0.03/word × 1,500 words = $225 per article, or roughly $1,100/week at that pace. That's the ceiling most part-time writers hit before they stop responding to every inquiry.

The catch: most buyers still screen for genuine originality. AI-assisted + human-edited is fine; pure AI spam gets you blacklisted. The edit pass is where you earn your price.

2. Affiliate content site (personal finance niche)

Time-to-first-dollar: 4–12 weeks | Ceiling: $2k–$10k+/month | Compounds? Very much

This is the one with the slow start and the best long-term ceiling. You build a site targeting a high-CPA niche — personal finance is the standout, with $25–$200 per qualified lead on credit cards, loans, and hosting referrals.

The 2026 play:

Pick one tight niche (e.g. "emergency funds," "side income," "credit card comparison").
Build 50–100 pages targeting long-tail queries.
Get indexed (sitemap + backlinks), then scale the winning pages.
Plug in affiliate links from the highest-paying programs in that niche.

Two programs worth knowing for this specific niche:

Hostinger — hosting referrals pay real commission per signup, and hosting is a natural "side income / start a blog" product your audience already wants. My own referral link, when I recommend it:

Hostinger

Hostinger's launch plans are ~$2.39/mo, so the "get a cheap host" pitch is strong and converts well.

iHerb — a bonus channel for the wellness version of this niche. It's a rewards/referral model rather than big cash:

iHerb referral code HHD6921

Worth running alongside if your content spans wellness, since it's low-effort and compounds.

The real constraint: time-to-index. A brand-new site on a free host takes weeks to be picked up by Google and Bing. If you can front-load some traffic (Dev.to syndication, Reddit, Quora answers with links), you compress that from months to a few weeks.

3. Digital products (templates, presets, courses)

Time-to-first-dollar: 2–6 weeks | Ceiling: $500–$3k/month | Compounds? Yes

You build once, sell many. What works in 2026:

Notion/Canva templates for a specific audience ("resume templates for new grads," "meal plan template for busy parents")
Lightroom/phone photo presets
Micro-courses on Gumroad or Lemon Squeezy — even $20–50 price points move volume

The win: no customer-service treadmill, near-zero marginal cost. The loss: you have to be good at distribution more than creation. A product with no audience is just a file.

4. Productized services (sell the outcome, not the hours)

Time-to-first-dollar: 2–4 weeks | Ceiling: $2k–$10k/month | Compounds? Partially

This is the underrated one. Instead of "I write articles for $X/word," you sell a flat, packaged deliverable: "5 SEO articles for $300, delivered in 5 days."

Why it beats hourly freelancing:

Buyers pay for the outcome, so price rises with perceived value, not hours.
You can template and systemize it — the 6th delivery is faster than the first.
You scale by delegating production to other writers or automation.

Where I'd start: pick one narrow service ("I write personal finance blog posts") and one narrow price. Don't build a menu of services yet.

5. Paid newsletter in a profitable niche

Time-to-first-dollar: 6–12 weeks (subscriber ramp) | Ceiling: $1k–$8k/month | Compounds? Very

Substack and beehiiv make the plumbing free. The money isn't the subscription itself (most readers stay free); it's:

A premium tier once you hit ~1,500–3,000 engaged subscribers
Sponsorships (a single $200–500 sponsorship post, a few times a month, out-earns most people's whole subscription revenue)
Affiliates woven into the content (hosting, courses, tools)

The 2026 sweet spot: a niche with a buying audience. "Money," "AI for work," and "health" all do it. General-interest newsletters are harder to monetize.

6. High-skill freelancing (AI + your existing expertise)

Time-to-first-dollar: 1–3 weeks | Ceiling: $500–$5k/month | Compounds? No (pure time)

If you already have a day-job skill that pays — design, coding, video, data — that's your fastest cash. The 2026 twist: pair it with AI tooling to 3× your throughput. You still sell your skill, but you deliver faster than a solo human could.

This isn't compounding, so treat it as your bridge income while the asset-building hustles above mature.

7. E-commerce / print-on-demand

Time-to-first-dollar: 4–10 weeks | Ceiling: $500–$5k/month | Compounds? Yes (if you find a winning product)

Print-on-demand and dropshipping still work, but the 2026 reality is:

Discovery is the whole game. A winning product + paid or viral marketing beats 50 average products with no marketing.
Platform fees and ad costs eat margins; the "passive income" claim is mostly a marketing hook.

I'd only do this if you're willing to test 5–10 products before finding one that sticks.

What I'd actually do if I had to start Monday

Starting from zero in early 2026 with a goal of $1k/month within 90 days, this is the exact stack:

Weeks 1–3: Freelance writing / content, aimed at 3–5 niche sites that buy finance/health content. Fast first dollar.

Weeks 2–12 (parallel): One affiliate site in a tight personal-finance or health niche, using free hosting, targeting 100 long-tail pages. Plug in Hostinger for the "build a blog / side income" angle and iHerb for wellness. Syndicate the top 10 guides to Dev.to to buy index speed and backlinks.

Weeks 4–16 (parallel): Package the content work into a productized service ($300 for 5 articles) and start pitching the same niche sites.

By week 12 you'd have: some freelance income coming in, an asset (the site) that compounds monthly, and a higher-priced service that scales without proportional effort.

The one thing that decides which of these you should pick

The right choice isn't the highest ceiling — it's the one where you'll still be doing it in week 8. If you hate writing, pick #4 or #6. If you love the slow compound build, pick #2 + #5 and stay for 12 months. If you need cash this month, #6 is your bridge while the rest matures.

The mistake is picking based on the screenshot of someone else's revenue, not on your own appetite for the actual daily work.

Building a Self-Sustaining AI Side Income in 2026: What Actually Works

Amr Mohamed — Fri, 16 Oct 2026 18:18:14 +0000

Building a Self-Sustaining AI Side Income in 2026: What Actually Works (And What Wastes Your Time)

The internet in 2026 is flooded with low-effort AI hype: "Make $10,000 in 24 hours with ChatGPT," "Auto-generate 1,000 YouTube shorts with one click," or "Sell AI-generated coloring books on Amazon."

If you've actually tried any of these, you already know the truth: low-effort spam gets suppressed by algorithms, ignored by buyers, or blacklisted by platforms within weeks.

Real online income in 2026 doesn't come from pushing a single magic button. It comes from building lightweight automated systems that solve genuine problems and compound over time.

In this comprehensive guide, we break down the exact operational models that are generating verifiable income today, the tech stacks required to run them with zero ongoing operational cost, and the pitfalls that separate profitable operators from frustrated dabblers.

The Reality of AI Monetization in 2026

Before picking a model, understand how the market has shifted:

Information arbitrage is dead. Nobody pays for generic text summaries or basic prompt outputs that any user can get from a standard LLM interface.
Distribution is 80% of the game. An automated asset with zero traffic earns exactly zero dollars. Systems must be engineered with built-in discoverability (SEO, backlinks, programmatic directory listings, or marketplace search volume).
Execution beats novelty. Simple workflows executed consistently—like automated daily sitemap submissions, targeted programmatic pages, or niche affiliate comparisons—vastly outperform over-engineered multi-agent experiments that break every other day.

Here are the 4 models delivering consistent returns right now.

1. Programmatic SEO & High-CPA Affiliate Stacks

Time-to-First-Dollar: 4–8 weeks

Monthly Ceiling: $2,000 – $10,000+

Intervention Level: Low (once built, fully autonomous)

Compounding Potential: High

The Playbook

Instead of manually typing single blog posts, programmatic SEO generates hundreds of hyper-targeted, long-tail landing pages answering high-intent transactional questions:

"Best AI tools for freelance copywriters"
"How to build a faceless content site on a budget"
"Cheapest reliable cloud hosting for AI automation scripts"

These queries have low individual keyword difficulty, meaning they can rank on search engines without requiring expensive domain authority. When multiplied across 200–500 structured pages, the aggregate organic impressions generate steady, recurring referral clicks.

The Monetization Engine

Pair the traffic with high-converting, proven affiliate networks:

Web Infrastructure & Hosting: Services like Hostinger offer aggressive CPA commissions ($60–$150+ per sale or high percentage tiers) because web hosting has near-zero churn and high customer lifetime value. Recommending budget-friendly launch tiers (~$2.39/month) delivers conversion rates far superior to expensive enterprise alternatives.
Physical Products / Recurring Essentials: For health, wellness, and lifestyle niches, referral programs like iHerb (Rewards Code: HHD6921) provide immediate credit and discounts that resonate with wellness seekers.

The Zero-Cost Architecture

Static Site Hosting: Cloudflare Pages (Free tier: unlimited bandwidth, instant global CDN, 500 automated builds/month).
Generation Engine: Python script running templated markdown generation against structured JSON keyword clusters.
Continuous Deployment: Cron tasks running in bash or Windows Task Scheduler that regenerate, deploy via Wrangler CLI, and trigger indexing pings daily.

2. Productized Micro-Services & Automated Lead Generation

Time-to-First-Dollar: 1–3 weeks

Monthly Ceiling: $1,500 – $5,000

Intervention Level: Medium (client onboarding, then templated execution)

Compounding Potential: Moderate

The Playbook

Local businesses, niche e-commerce brands, and agency owners need automated workflows but lack the technical expertise to set up API keys, Docker containers, or web scrapers.

Instead of selling open-ended "AI consulting," package a specific, repeatable micro-service:

Automated Competitor Price Tracker: A weekly Python crawler that emails a local retail shop or Amazon seller a PDF comparison of competitor pricing shifts.
Local Business Review Responder: A lightweight webhook listening to Google Business reviews and drafting personalized, professional draft replies for the manager to approve in 1 click.
Programmatic Directory Listing Builder: Setting up a niche listing portal for local contractors or service providers.

How to Charge

Offer this as a flat setup fee ($300–$500) plus an ongoing monthly maintenance retainer ($49–$99/month). Three to five local clients easily cover foundational living expenses with minimal weekly maintenance.

3. Digital Asset Stores (Notion Templates & Automation Workflows)

Time-to-First-Dollar: 2–4 weeks

Monthly Ceiling: $500 – $3,000

Intervention Level: Very low after launch

Compounding Potential: High

The Playbook

Digital products have 99% gross margins and zero fulfillment overhead. The top-selling digital assets in 2026 focus on productivity and automation:

Specialized Notion Systems: Freelance CRM boards, crypto portfolio rebalancing calculators, agency project management templates.
Ready-to-Deploy Code Starter Kits: Pre-configured Flask/FastAPI boilerplates with OAuth, Stripe billing, and Cloudflare Pages deployment ready to go.
Prompt & Workflow Libraries: Domain-specific libraries tested for actual output reliability (e.g., legal document summarizers, medical transcript formatting).

Distribution Channels

Publish your assets across:

Gumroad & Lemon Squeezy: Frictionless checkout with automated payout handling.
Etsy: Surprisingly strong organic search intent for printable templates, budgeting spreadsheets, and Notion dashboards.
GitHub Public Gists & Repositories: Providing a free, lightweight version in a public Gist or repo with a link to the complete premium pack captures high-intent technical users directly from Google search results.

4. Curated Niche Hubs & Paid Newsletters

Time-to-First-Dollar: 6–12 weeks

Monthly Ceiling: $1,000 – $8,000+

Intervention Level: Low-to-Medium

Compounding Potential: Exceptional

The Playbook

Audience attention has fragmented. Readers are actively looking for curated filters that cut through the endless noise.

Pick an underserved intersection (e.g., "AI in Renewable Energy", "Biohacking & Sleep Optimization", or "Solopreneur Automation Tactics").
Use automated collectors to scan RSS feeds, arXiv papers, and public repositories daily.
Synthesize the top 5 developments once a week with concise, actionable commentary.

Once your subscriber list passes 1,000 active readers, monetize through:

Direct sponsor placements ($50–$150 per send).
Recommended tools and hosting providers like Hostinger for readers launching their own digital sites.
Premium community or deep-dive quarterly industry reports.

Recommended Starter Stack: The Autonomous 3-Site Blueprint

If you are starting today with zero upfront capital, here is the exact infrastructure blueprint we operate:

┌────────────────────────────────────────────────────────┐
│             Autonomous Income Portfolio                │
├──────────────────────────┬─────────────────────────────┤
│ 1. Personal Finance Hub  │ https://financeedge.pages.dev│
│    Monetization:         │ Hostinger Referral + Cash   │
├──────────────────────────┼─────────────────────────────┤
│ 2. Health & Wellness Hub │ https://vitalpath-d2c.pages.dev│
│    Monetization:         │ iHerb Rewards (Code HHD6921)│
├──────────────────────────┼─────────────────────────────┤
│ 3. AI & Online Income    │ https://sidestack.pages.dev │
│    Monetization:         │ Hostinger + iHerb Combined  │
└──────────────────────────┴─────────────────────────────┘
                           │
       ┌───────────────────┴───────────────────┐
       ▼                                       ▼
Automatic Daily Cron             Public Syndication Engine
(Cloudflare Pages + Wrangler)    (GitHub Gists + Web Archives)

Deploy 3 Free Niche Sites: Use Cloudflare Pages to host lightweight HTML sites targeting distinct high-intent clusters (Finance, Health, Side Income).
Embed Proven Affiliate Links:
- Hostinger Referral: hostinger.com/?REFERRALCODE=OVDAMRMASPIM (high-yield cash payouts on hosting packages).
- iHerb Rewards: iherb.com/?rid=HHD6921 (health and wellness rewards credit).
Automate Daily Refresh & Backlink Signals:
- Run a scheduled bash script to update page counts, refresh sitemaps, and deploy via the Cloudflare Pages API.
- Publish high-authority long-form guides as public GitHub Gists to pass real page authority and drive referral discovery.
- Ping Wayback Machine endpoints on high-intent money pages to anchor crawler records without manual intervention.

3 Critical Pitfalls That Kill AI Income Projects

Relying Exclusively on AdSense: Display ads pay pennies ($1–$3 RPM) on early-stage sites. You need hundreds of thousands of views to make coffee money. High-CPA affiliate referrals ($20–$150 per transaction) make your first 1,000 visitors genuinely profitable.
Letting Automated Builds Drift Stale: Search engines reward active, updated sites. A simple automated daily cron script that adds fresh long-tail pages signals continuous active maintenance to web crawlers.
Failing to Anchor Authoritative Backlinks: A static site with zero external links takes months to index organically. Syndicating real, value-dense guides to authoritative platforms (like GitHub Gists) cuts the initial indexing delay down from months to days.

This guide is part of our research at SideStack. Explore our sister research hubs at FinanceEdge and VitalPath for in-depth data on automated personal finance and wellness systems.

The 10 Best AI Affiliate Programs in 2026 (Real Rates, No Stale Data)

Amr Mohamed — Fri, 16 Oct 2026 18:13:05 +0000

The 10 Best AI Affiliate Programs in 2026 (With Real Commission Rates)

Most "best AI affiliate programs" articles copy each other and list tools whose programs have quietly closed. This one is different — every rate below is pulled from the merchant's own terms, and I've flagged the two big traps that quietly kill most affiliate income in this category.

I run programmatic content in the AI-tools niche (you can see the working system on ToolRadar), so I evaluate these the way a publisher does: what will this actually pay me per 1,000 visitors?

The two traps that make a high percentage worthless

Before you pick a program, understand the two things that decide real income:

1. Duration, not headline rate. The category standard is 20-30% recurring for 12 months. A 50% rate that pays for 3 months is worth less than a 30% rate that pays for the customer's entire life. Always multiply rate × months before you compare.

2. Retention is the whole game. AI tools have unusually high churn — people try one, love it for a month, cancel. A 30% commission on a $10 tool people quit in 6 weeks loses to a 20% commission on a $199 tool people keep. Pick a tool your audience will still be using next year.

With that in mind, here's the ranking.

The 10 I'd actually promote

1. CapCut Pro — 35% recurring, lifetime

The terms are the best in the entire category. 35% of every payment, with no stated end date. It's a video-editing tool with a huge free audience, so conversion is strong and retention is decent. If you only join one program, make it this one.

2. ElevenLabs — 22% recurring, 12 months

The AI-voice leader. 22% on Starter through Scale plans for the first year, 90-day cookie. Read the fine print: it pays 11% on Business plans and nothing on enterprise. Don't build content that pushes enterprise buyers — they'll be your best leads and your $0 earnings.

3. Synthesia — 25% recurring, 12 months

AI video for business. Pays up to $267 per customer referral and gives you a free Starter account after your 10th sale. Strong for a B2B or creator audience.

4. InVideo — 50% one-time, 120-day cookie

The long 120-day cookie is huge — it gives a slow-deciding buyer four full months to convert. 50% of the first payment is a strong one-time payout for a video audience.

5. Jasper — 25% recurring, 12 months (rises to 30% at 100 referrals)

The AI-writing name brand. Big audience recognition means high click-through. The rate ticks up to 30% after your 100th converted lead, so volume rewards you.

6. Jotform — 30% recurring, 12 months

Underrated. Forms + AI for a huge business audience, quick approval, and a very sticky product (teams don't churn off a form they've already built). 30% on a product that stays.

7. Fireflies.ai — 30% recurring, 12 months, 90-day cookie

AI meeting notes. The 90-day cookie is the longest in the AI-tools group. Great fit for a productivity or remote-work audience.

8. Writesonic — 20% recurring, 12 months

Lower rate but a mature product with strong retention. 60-day cookie. A safe, reliable earner even if not the headline-grabber.

9. Murf.ai — 20% recurring, 24 months

The longest duration of any major program I track: it pays for a full two years, not one. On retention math, 20% × 24 months out-earns 30% × 12 months for any customer who sticks around.

10. HeyGen — 35% recurring, 3 months

High rate but only for 3 months. I rank it #10 despite the 35% because the duration caps it hard — it's a one-time spike, not recurring income. Good for a quick campaign, bad to build a site around.

Programs that look great but are a trap

Notion — appears on every "best AI affiliate" list but has repeatedly closed to new applicants. Don't build content around a program that may not approve you.
Canva — replaced its affiliate program with the "Canvassador" ambassador scheme, which pays a one-time bounty, not recurring. Not what it used to be.
Grammarly — strong brand, unpublished rates, and a keyword-bidding-friendly policy that can burn you in paid search. Skip for SEO content.

The realistic math (so you don't chase the wrong one)

Let's say a niche site does 5,000 visits/month and 2% convert:

CapCut (35% lifetime, ~$10/mo avg): 100 signups × $3.50 = $350/month recurring, and it grows as the customer base compounds.
Jasper (25%, 12mo, ~$40/mo avg): 100 × $10 = $1,000/month for the first year of each cohort — the ceiling is higher, but you must keep referring new customers.
HeyGen (35%, 3mo): the same 100 signups pay out over just 3 months, then stop.

The pattern: a mid-rate program with lifetime or multi-year duration on a product with good retention beats a high-rate, short-duration one. That's why CapCut and Murf out-earn flashier names over time.

How I'd actually run this (zero upfront cost)

Pick 3 programs, not 10. Focus where your audience converts. For an AI-tools site I'd run CapCut + ElevenLabs + Jotform.
One program per content cluster. A "CapCut vs Descript" page promotes CapCut. A "meeting notes for remote teams" page promotes Fireflies.
Host on free infrastructure. Static pages on Cloudflare Pages (free, unlimited bandwidth), regenerated daily by a scheduled script, with the affiliate links wired in — that's the exact system running on ToolRadar.
Pair hosting with an affiliate. Tools like Hostinger both host your site and pay you per signup, so one link does double duty.

Recurring + lifetime + retention is the only combination that turns a 5,000-visit site into real passive income rather than a monthly spike.

This guide is part of the research running on ToolRadar. We cover these tools across 200+ pages — see the full AI-tools hub, or grab a cheap host to build your own via Hostinger referral OVDAMRMASPIM.

Hermes Agent: A Practical Beginner's Guide to Doing Real Things With It (2026)

Amr Mohamed — Fri, 16 Oct 2026 17:01:14 +0000

Hermes Agent: A Practical Beginner's Guide to Doing Real Things With It (2026)

Hermes Agent is an open-source AI agent framework by Nous Research that runs in your terminal, your messaging apps, and your IDE. Most of the time "AI agent" means a chatbot with some extra steps. Hermes is different: it ships with the actual building blocks that make an agent useful — persistent memory, scheduled jobs, subagents that run in parallel, and a plugin system you can extend without forking the code.

This guide is the version I wish existed when I started. It skips the marketing and focuses on: what Hermes can actually do for you, how to get running in about two minutes, and the four features that separate it from a plain chat assistant.

What Hermes Agent is (in one paragraph)

It's a self-hosted agent that keeps a persistent model of you and your environment. It can run terminal commands, edit files, browse the web, search its own past conversations, schedule itself to do jobs on a timer, and spawn parallel subagents. The same agent can answer you in the terminal, on Telegram, on Discord, in Slack, or over email — and it uses the same memory and tools on every platform.

That last part is the big deal. Most chat tools start from zero every conversation. Hermes remembers your preferences, your projects, and what it has learned.

Getting started (about two minutes)

On Linux, macOS, or WSL:

curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
source ~/.bashrc
hermes

On native Windows (PowerShell):

iex (irm https://hermes-agent.nousresearch.com/install.ps1)
hermes

That drops you into an interactive chat. A few commands matter:

hermes model — pick your LLM provider and model
hermes setup — the full setup wizard (config, tools, messaging)
hermes doctor — diagnose install or config problems
hermes update — pull the latest version

If you want the desktop app after a CLI-only install, just run hermes desktop.

The four features that actually matter

1. Persistent memory

Hermes keeps a curated memory file that carries across every session. Tell it your preferences once ("I use a Mac and deploy with Cloudflare") and it doesn't forget. It also can search its own past sessions — ask it to recall something you discussed weeks ago and it can find it. This is what turns a tool you re-brief every day into something that builds up context over time.

2. Skills (its self-improvement loop)

Skills are small documents that encode a procedure — "how I do X in this environment" — that the agent loads when a matching task comes up. The important bit: when Hermes works out a non-trivial workflow, it can save that as a skill so it reuses the approach next time. It's a genuine learning loop: it gets better at your specific recurring tasks the more you use it.

3. Cron / scheduled jobs

You can schedule the agent to run on a timer — "every morning at 6, check these things and send me a summary to Telegram." Jobs can be agent-driven (it figures out what to do) or just plain scripts (run this and only wake me if the output changed). This is how people turn Hermes into an overnight worker rather than something you have to sit in front of.

4. Subagent delegation

For a big task, Hermes can spawn isolated child agents that each get their own context and tools and run in parallel, then report back. Three at a time by default. It's how you get a team of workers instead of one slow conversation: one researches, one drafts, one reviews — and you only see the results.

Where people actually use it

Personal automation — a morning briefing, a nightly backup, a "watch this price and tell me when it drops" job.
Self-hosted assistants — the same agent reachable from your phone over Telegram or WhatsApp.
Developer workflows — it reads your code, runs your tests, opens pull requests, and remembers your repo's conventions.
Multi-agent setups — a team of agents with isolated profiles and shared skills for a bigger project.

A note on running it seriously

Hermes is free and open source, but you still need a place for it to live if you want it running 24/7 — a always-on machine, a VPS, or a container. For a personal setup most people keep it on a home machine; for a team setup it runs on a small server. If you're building a project site or dashboard around an agent, cheap VPS hosting covers it — you can start a Hostinger VPS for a few dollars a month: Hostinger (referral code OVDAMRMASPIM).

Keep one rule: the memory and skill files are yours and live on your machine, so back them up. They're the part of Hermes that is actually irreplaceable.

Bottom line

If you want a chatbot, Hermes is more than that. If you want an agent that does things — runs your commands, schedules itself, remembers you, and learns your workflows — this is one of the most complete open-source options available. Start with the two-minute install, point it at a real recurring task, and let the skills and memory accumulate. That's where it goes from "a tool" to "something that actually works for you."

This guide is part of the how-to collection on HermesGuide — 90+ guides on installing, configuring, automating, and extending Hermes Agent. Hosting your agent deployments? Grab a cheap VPS via Hostinger referral OVDAMRMASPIM.

MCP Explained: The 10-Minute Guide to Connecting AI Agents to Real Tools (2026)

Amr Mohamed — Fri, 16 Oct 2026 16:52:55 +0000

MCP Explained: The 10-Minute Guide to Connecting AI Agents to Real Tools (2026)

Every AI agent demo eventually hits the same wall: the model can reason brilliantly, but it can't touch anything — not your database, not your files, not your CRM. MCP (the Model Context Protocol) is the fix that became the de facto standard in 2026. This is the plain-language guide I wish existed when I first hit the term.

What MCP actually is

MCP is an open, standard protocol that defines how an AI application (the "client") connects to external tools and data (the "server") in a consistent way.

The one idea to grab: it's a plug-in standard for AI tools. Before MCP, every tool integration was a bespoke, one-off API hookup. With MCP, any tool exposes itself once as an MCP server, and any MCP-aware assistant can use it. Think of it as USB-C for AI capabilities.

The three pieces (this is 80% of the guide)

MCP Client — the AI app that uses tools: Claude Desktop, Cursor, your OpenClaw or Hermes agent, or your own agent code.
MCP Server — the program that exposes a capability: "read a Postgres database," "search the web," "control a Home Assistant device." A server is just a small program implementing the MCP spec.
The tools/resources it exposes — a server advertits a set of named tools (e.g. query_database, get_weather), each with a JSON-schema description of what it takes and returns. The client hands that description to the model; the model picks a tool; the client runs the server; the result comes back.

That's the whole loop: describe → pick → execute → return.

Why it changed the game

Before MCP, wiring one tool into one agent was a project. Now:

Reuse: one MCP server for Slack works across Claude, Cursor, your OpenClaw, and n8n — no rework.
Composability: your agent gets more capabilities by adding servers, not by forking the agent.
The ecosystem exploded: within months there were hundreds of community MCP servers (databases, browsers, file systems, GitHub, Stripe, Home Assistant...) — most of them free.

The standard effect is real: MCP stopped being "a protocol" and became the way agent-tool integration is described.

Setting up an MCP server (the realistic path)

You rarely write your own MCP server from scratch — you usually install one and point your client at it.

1. Pick a server

Start with one that solves a real problem in your workflow:

Filesystem — let the agent read/write files in a folder
Postgres / SQLite — query your data
GitHub / Stripe / Notion — act on SaaS you already use
Browser (playwright) — let the agent drive a webpage

2. Configure the client

MCP clients hold a small config listing each server and how to launch it. The pattern is always the same — a server name, a command (or a URL for remote servers), and optional env vars:

{
  "mcpServers": {
    "postgres": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-postgres", "postgresql://localhost/app_db"],
      "env": { "DATABASE_URL": "postgresql://..." }
    },
    "filesystem": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-filesystem", "/home/you/projects"]
    }
  }
}

Restart (or reconfigure) the client, and the tools show up. That's it.

3. Run remote servers on a VPS

Here's where self-hosting earns its keep. Many MCP servers do useful but heavy work (web scraping, database access, home-automation control). You can host the server on a cheap always-on VPS and point any client at it over a remote/HTTP transport. For the box itself, a low-cost VPS is all you need — e.g. Hostinger (referral code OVDAMRMASPIM).

Writing your own MCP server (15 lines of the idea)

When no existing server fits, you write one. A server is just a program that, on request, returns the tool list and executes a chosen tool. In TypeScript (MCP's official SDK), the shape is:

import { McpServer } from "@modelcontextprotocol/sdk/server/mcp.js";
const server = new McpServer({ name: "weather", version: "1.0.0" });

server.tool("get_weather", "Get the weather for a city.",
  { city: "string" },
  async ({ city }) => { /* call your API */ return { content: [{ type: "text", text: "18°C" }] }; }
);

// start over stdio (local) or http (remote)

The discipline that matters: describe tools precisely. The model chooses tools by reading their names + descriptions + schemas. Vague descriptions = the agent picks the wrong tool or hallucinates an argument. Treat the tool description like you'd write a function's docstring for a junior dev.

The part everyone skips: security

An MCP server is code your agent can run with your credentials. That's powerful and dangerous. Three rules:

Least privilege. Give each server only the credentials it needs (a read-only DB user, not a superuser). Scope a filesystem server to one folder, not the whole disk.
Audit what your agent can do. List the tools each server exposes and ask: "if the model is wrong, what's the worst it can do?" A read-only tool has a boring worst case; a write tool with your AWS key does not.
Don't expose servers publicly. Keep local servers on stdio (they run in-process, never on a port). If a server must be remote, put it behind auth + a firewall — never an open port with a global database key in the env.

For team or production use, add logging + rate limits on every remote server, and treat MCP config the same as any secrets file.

MCP + n8n: the no-code path

If you'd rather not write config, n8n (the visual automation tool) has native MCP integration: you point an AI agent node at an MCP server and build your automation visually instead of in JSON. That's how non-developers get the same capability — "when X happens, have my AI agent look up Y via MCP, then do Z." If you're self-hosting n8n for that, it's another small VPS — cheap on Hostinger OVDAMRMASPIM.

The honest bottom line

MCP is the piece that turned AI agents from "they can chat" into "they can do." If you're building anything agentic in 2026, MCP is not optional — it's how your agent connects to the real world. Start by installing one server your agent actually needs (filesystem or your database are the best first picks), keep its permissions tight, and scale up from there.

This guide is part of the MCP & AI-agent collection on AgentStack — 75+ pages on MCP, n8n, and building agents. Hosting your MCP servers or n8n instance? A cheap always-on VPS via Hostinger OVDAMRMASPIM covers it.

How to Run OpenClaw on a VPS (Self-Hosted AI Assistant, Step by Step)

Amr Mohamed — Fri, 16 Oct 2026 16:52:19 +0000

How to Run OpenClaw on a VPS (Self-Hosted AI Assistant, Step by Step)

OpenClaw is the open-source, self-hosted AI assistant that hit 380k+ GitHub stars in 2026 — it turns your machine (or a server) into a personal assistant you can message from WhatsApp, Telegram, Slack, Discord, and 20+ other apps, powered by whatever model you want, with all data staying on your hardware.

The one question everyone asks: where do I actually run it so it's always on? Not your laptop — that goes to sleep. The answer is a small, cheap VPS. This is the exact setup I use, written so you can replicate it in about an hour.

Why a VPS (and not your laptop)

A personal assistant is only useful if it's there 24/7. Your laptop sleeps, reboots, and takes your Wi-Fi with it. A VPS:

Stays on 24/7 regardless of your devices
Gives OpenClaw its own persistent home (config, memory, sessions)
Can sit in a datacenter close to your AI model provider

All you need is roughly 2 GB RAM and a couple of vCPUs — OpenClaw itself is light; your AI provider does the heavy lifting. That's a low-cost VPS tier, which is the whole point.

Step 1 — Get the hardware

Pick a small VPS. For a personal OpenClaw setup, a single 2 GB / 1-2 vCPU instance is plenty:

What to grab: 1 vCPU, 2 GB RAM, 20+ GB SSD, an NVMe volume, and an IPv4 address.
Where: any provider with a low entry tier works. I run mine on Hostinger's VPS — their launch VPS plans start around $5-6/mo, which is the cheapest honest price for a machine that never sleeps.
OS: I use a plain Linux image (Ubuntu 22.04/24.04 or Debian 12). Skip "managed" stacks for the base — you'll be installing Node anyway.

After provisioning you get a root/SSH login. That's all you need for this step.

Step 2 — Open the right port and lock the firewall

OpenClaw's Gateway listens on a local port. You only ever connect through SSH, so keep the default SSH-only exposure:

In your provider's control panel, open one port for SSH (22).
Do not open OpenClaw's port to the public internet. Run it behind an authenticated reverse proxy (see Step 5) instead.

Then on the box, install ufw and lock it down:

sudo apt update && sudo apt install -y ufw
sudo ufw default deny incoming
sudo ufw allow 22/tcp
sudo ufw enable

Step 3 — Install Node and OpenClaw

OpenClaw is built on Node. Install a current LTS Node, then OpenClaw:

# Node LTS (via NodeSource)
curl -fsSL https://deb.nodesource.com/setup_22.x | sudo -E bash -
sudo apt install -y nodejs

# OpenClaw (follows its documented install)
curl -fsSL https://openclaw.ai/install.sh | bash

Confirm the version, then start the setup wizard:

openclaw --version
openclaw setup

The setup wizard walks you through: model provider, API key, the channels you want (Telegram / WhatsApp / etc.), and where OpenClaw keeps its workspace (its config, memory, and operating instructions). Let it use the default ~/.openclaw workspace on the VPS.

Step 4 — Connect your AI model (bring your own key)

OpenClaw is model-agnostic — Claude, GPT, Gemini, or fully local via Ollama. For a 24/7 personal assistant I recommend a hosted model provider (reliable, no GPU cost on the VPS):

Get an API key from your chosen provider.
Put it in OpenClaw's config / environment — never hardcode it in a file you'll commit.
Test with a one-shot prompt to confirm the round-trip works.

Cost control: the VPS is a few dollars/month; your AI usage is the variable cost. Pick a model tier you'll actually use, and set provider spend limits so an agent that loops can't surprise you.

Step 5 — Keep it running 24/7 (systemd)

Don't just run it in a terminal that dies when you log out. Make OpenClaw a service:

# /etc/systemd/system/openclaw.service
[Unit]
Description=OpenClaw Gateway
After=network-online.target
Wants=network-online.target

[Service]
User=youruser
ExecStart=/usr/local/bin/openclaw gateway
Restart=always
RestartSec=5
Environment=NODE_ENV=production

[Install]
WantedBy=multi-user.target

Then:

sudo systemctl enable --now openclaw
sudo systemctl status openclaw

Restart=always means if it ever crashes, systemd brings it right back — the whole point of a 24/7 assistant.

Step 6 — Reach it from your phone (reverse proxy + auth)

The Gateway has a browser Control UI. To reach it from your phone over the internet safely:

Put it behind Caddy or nginx with a free SSL cert.
Front it with basic-auth or OAuth so the world can't open your assistant.

Caddy makes the TLS part automatic:

openclaw.example.com {
    basic_auth {
        youruser <hash>   # run: caddy hash-password
    }
    reverse_proxy 127.0.0.1:<openclaw-port>
}

Point a cheap domain (or a subdomain of one you have) at the VPS's IP, and Caddy issues the cert. Now your phone reaches the assistant at https://openclaw.example.com — encrypted, and only you can in.

(A simpler alternative: SSH tunnel — ssh -L 8080:localhost:<port> your-vps and open localhost:8080 in your phone browser. Zero extra setup, no public exposure at all.)

Step 7 — Point your channels at it

Now the satisfying part. Connect the chat apps you already use:

Telegram: create a bot via BotFather, drop the token into OpenClaw. Message the bot — that's your assistant.
WhatsApp: link a dedicated number (the two-phone setup OpenClaw recommends: one number just for the assistant, so your personal chats stay private).
Slack / Discord / Signal / iMessage: OpenClaw supports these through its channel plugins.

From that moment, your "assistant" lives on the VPS and is reachable from anywhere.

Keeping it healthy

Back up the workspace (~/.openclaw) — that's the memory and config you can't get back. A nightly rsync or tarball to object storage is enough.
Watch disk: logs and sessions grow; add a small rotation.
Update deliberately: OpenClaw moves fast, but test an update in a clone before rolling it onto the box your assistant runs on.
Credentials: keep API keys in env vars or a secrets file with 600 perms, never in the repo.

The honest bottom line

The VPS itself is the cheapest part — a few dollars a month on a provider like Hostinger (referral code OVDAMRMASPIM). The real work is the setup: Node, systemd, a locked-down firewall, and a reverse proxy in front of the Control UI. Do it once and you have an always-on, model-flexible, private assistant that meets you in the apps you already use. That's the whole pitch of self-hosting OpenClaw, and it's genuinely achievable in an afternoon.

This guide is part of the OpenClaw how-to collection on ClawDesk — 90+ guides on installing, configuring, and hosting OpenClaw. Ready to host it? A cheap VPS via Hostinger OVDAMRMASPIM is all you need.

A Realistic Guide to Better Sleep in 2026 — What Actually Works

Amr Mohamed — Wed, 14 Oct 2026 20:08:19 +0000

A Realistic Guide to Better Sleep in 2026 — What Actually Works (and What's Hype)

"Fix your sleep" is the most repeated health advice on the internet and also the least useful, because it's almost always vague. "Go to bed earlier," "turn off your phone," "take magnesium" — all of it, none of it contextualized.

Here's the honest version: what's proven, what's marketing, and where I'd actually spend your money.

The 4 levers that matter more than anything else

If you're going to do just four things, these are them, in order of impact:

1. Consistent wake time (yes, wake time, not bed time)
This is the #1 lever and most people ignore it. The sleep literature is very clear: fixing your morning anchor time synchronizes your circadian rhythm far more than any bedtime ritual. Pick a wake time and hold it on weekends too, at least 70% of the week.

2. Light exposure in the morning
10–15 minutes of outdoor light within an hour of waking sets your clock. This is free, takes no willpower, and beats most supplements I could list.

3. Caffeine cutoff
Not "no coffee" — a real cutoff. Caffeine has a half-life of 5–7 hours, meaning a 4pm coffee is still 25–50% in your system at 1am. Move your last cup earlier than you think you need to.

4. Consistent room temperature
Your body needs to drop ~1°F to initiate sleep. A bedroom that's 65–68°F (18–20°C) is the single most-cited environmental factor in sleep-quality studies. This one is cheap and immediate.

If you do all four, you're ahead of most people with no spending required. The rest is optimization.

The supplements that are worth it (and the ones that aren't)

The sleep-supplement market is full of overpriced marketing. Here's the honest tier list:

Worth trying (evidence + effect size):

Magnesium glycinate (200–400mg, 1–2 hours before bed). The most consistently beneficial mineral for sleep quality, especially if you're deficient.
L-theanine (100–200mg). Promotes relaxation without the sedation crash that leaves you groggy.
Melatonin (0.5–1mg, used sparingly). The doses sold in most stores (3–10mg) are 3–20× higher than needed. Use a small dose, short-term, only for jet-lag or clock-shift — not as a nightly crutch.

Mostly marketing, skip:

"Melatonin + 7 other things" blends. Usually the melatonin is a trace dose and the rest is proprietary fillers. You're paying for the label.
Gummies. You're buying the candy; absorption is worse than a capsule and you're often taking sugar right before bed.

Where I keep my shopping: iHerb has the widest range of independent brands at prices well under US pharmacy markups, and it's a clean, no-frills way to stock up on magnesium, L-theanine, and quality omega-3s. My own referral link, if you want a small bonus on your first order:

iHerb referral code HHD6921

iHerb's rewards structure means you earn points on orders, so if you're going to build a supplement habit, this is the most cost-efficient way to do it.

The "20-minute rule" for falling asleep

If you can't fall asleep, don't stay in bed fighting it. Here's the protocol:

If you haven't fallen asleep in ~20 minutes (don't check the clock; estimate), get up.
Go to a dim, different room. Do something boring — read, stretch, listen to calm music. No screens.
Return to bed only when you feel drowsy.
Repeat if needed.

This trains your brain to associate your bed with sleep, not frustration. It's a version of stimulus control from CBT-I (cognitive behavioral therapy for insomnia), which is the gold-standard non-drug treatment for chronic insomnia.

Track, don't guess

One "tracker" that actually pays for itself: just log three things every night — time in bed, time to fall asleep, and morning wake time. A week of data tells you more than any app.

If you want a device-based tracker, the 2026 options split into:

Ring-based (Oura) — best for a long-term, low-disruption signal.
Wrist-based (Apple Watch, Whoop) — already on your wrist, good enough for trends.
Under-mattress sensor (Eight Sleep, Withings) — no wear, good for partners who resist devices.

Don't overthink the tracker. The signal you need is: is my time-to-sleep trending down, and is my wake time staying consistent? Two numbers, weekly.

What I'd actually do this week

If I were starting fresh on sleep in 2026, here's the exact 7-day reset:

Day 1: Pick your anchor wake time. Set it on your watch.
Day 2: Set a 65–68°F thermostat. Get up, get 10 minutes of outdoor light within 30 min.
Day 3: Move your last caffeine to 12pm or earlier.
Day 4: Add magnesium glycinate in the evening.
Day 5–7: Hold the same wake time every day, even weekends. Log bed-time, sleep-onset, wake-time in a notes app.

By day 7 you'll have data. Most people are already 30–45% further along by the time they see it.

The one thing I'd skip entirely

"Sleep optimization" courses, cold exposure, and the 50 other "biohacks" that get pushed each year. None of them have a clean effect size that beats the boring four: consistent wake time, morning light, earlier caffeine, cooler room. Master those first; everything else is noise on top.

Sleep is the highest-leverage thing you can do for energy, mood, weight, and longevity. But it's boring by design — the protocols that work don't feel exciting. That's the whole game: the unglamorous four, executed daily.

7 Side Hustles That Actually Pay in 2026 — Ranked by Time-to-First-Dollar

Amr Mohamed — Wed, 14 Oct 2026 20:08:17 +0000

7 Side Hustles That Actually Pay in 2026 — Ranked by Time-to-First-Dollar

Most "make money online" lists are recycled. They tell you to start a blog, do surveys, or drop-ship, but they skip the real question: how fast does the money actually land?

How I ranked these

Most listicles never show the framework. Here's mine:

Time-to-first-dollar — How long from "started" to "money in the bank"?
Ceiling — Can this realistically hit $1k/month, or is it capped?
Barrier to entry — Skills, accounts, or upfront cost required.
Compounding — Does the work build an asset (audience, clients, store) or is it pure time-for-money?

The sweet spot for most people: fast first dollar, decent ceiling, and something that compounds. Let's get into it.

1. Freelance writing / content at scale

Time-to-first-dollar: 1–3 weeks | Ceiling: $3k+/month | Compounds? Yes

You don't need a degree or a portfolio of 20 polished pieces. What you need is consistency and one channel that gets you paid.

Where I'd focus in 2026:

Content marketplaces that pay per word for quick first checks.
Outreach to niche sites in personal finance, health, and SaaS — these three verticals consistently buy SEO content.
Programmatic content — set up a simple system (templates, keyword lists, AI-assisted drafting with a human editing pass). One person can put out 5–10 sellable articles a week.

The catch: most buyers still screen for genuine originality. AI-assisted + human-edited is fine; pure AI spam gets you blacklisted. The edit pass is where you earn your price.

2. Affiliate content site (personal finance niche)

Time-to-first-dollar: 4–12 weeks | Ceiling: $2k–$10k+/month | Compounds? Very much

The 2026 play:

Pick one tight niche (e.g. "emergency funds," "side income," "credit card comparison").
Build 50–100 pages targeting long-tail queries.
Get indexed (sitemap + backlinks), then scale the winning pages.
Plug in affiliate links from the highest-paying programs in that niche.

Two programs worth knowing for this specific niche:

Hostinger

Hostinger's launch plans are ~$2.39/mo, so the "get a cheap host" pitch is strong and converts well.

iHerb — a bonus channel for the wellness version of this niche. It's a rewards/referral model rather than big cash:

iHerb referral code HHD6921

Worth running alongside if your content spans wellness, since it's low-effort and compounds.

3. Digital products (templates, presets, courses)

Time-to-first-dollar: 2–6 weeks | Ceiling: $500–$3k/month | Compounds? Yes

You build once, sell many. What works in 2026:

Notion/Canva templates for a specific audience ("resume templates for new grads," "meal plan template for busy parents")
Lightroom/phone photo presets
Micro-courses on Gumroad or Lemon Squeezy — even $20–50 price points move volume

The win: no customer-service treadmill, near-zero marginal cost. The loss: you have to be good at distribution more than creation. A product with no audience is just a file.

4. Productized services (sell the outcome, not the hours)

Time-to-first-dollar: 2–4 weeks | Ceiling: $2k–$10k/month | Compounds? Partially

This is the underrated one. Instead of "I write articles for $X/word," you sell a flat, packaged deliverable: "5 SEO articles for $300, delivered in 5 days."

Why it beats hourly freelancing:

Buyers pay for the outcome, so price rises with perceived value, not hours.
You can template and systemize it — the 6th delivery is faster than the first.
You scale by delegating production to other writers or automation.

Where I'd start: pick one narrow service ("I write personal finance blog posts") and one narrow price. Don't build a menu of services yet.

5. Paid newsletter in a profitable niche

Time-to-first-dollar: 6–12 weeks (subscriber ramp) | Ceiling: $1k–$8k/month | Compounds? Very

Substack and beehiiv make the plumbing free. The money isn't the subscription itself (most readers stay free); it's:

A premium tier once you hit ~1,500–3,000 engaged subscribers
Sponsorships (a single $200–500 sponsorship post, a few times a month, out-earns most people's whole subscription revenue)
Affiliates woven into the content (hosting, courses, tools)

The 2026 sweet spot: a niche with a buying audience. "Money," "AI for work," and "health" all do it. General-interest newsletters are harder to monetize.

6. High-skill freelancing (AI + your existing expertise)

Time-to-first-dollar: 1–3 weeks | Ceiling: $500–$5k/month | Compounds? No (pure time)

This isn't compounding, so treat it as your bridge income while the asset-building hustles above mature.

7. E-commerce / print-on-demand

Time-to-first-dollar: 4–10 weeks | Ceiling: $500–$5k/month | Compounds? Yes (if you find a winning product)

Print-on-demand and dropshipping still work, but the 2026 reality is:

Discovery is the whole game. A winning product + paid or viral marketing beats 50 average products with no marketing.
Platform fees and ad costs eat margins; the "passive income" claim is mostly a marketing hook.

I'd only do this if you're willing to test 5–10 products before finding one that sticks.

What I'd actually do if I had to start Monday

Starting from zero in early 2026 with a goal of $1k/month within 90 days, this is the exact stack:

Weeks 1–3: Freelance writing / content, aimed at 3–5 niche sites that buy finance/health content. Fast first dollar.

Weeks 4–16 (parallel): Package the content work into a productized service ($300 for 5 articles) and start pitching the same niche sites.

By week 12 you'd have: some freelance income coming in, an asset (the site) that compounds monthly, and a higher-priced service that scales without proportional effort.

The one thing that decides which of these you should pick

The mistake is picking based on the screenshot of someone else's revenue, not on your own appetite for the actual daily work.