Stop Working: How I Automated a $35K/Month SaaS Portfolio

How I Automated a $35K/Month SaaS Portfolio: A Full 2026 Strategy Breakdown

I almost quit in month four.
Not dramatically. No meltdown. Just a quiet Tuesday night where I sat at my desk, opened my banking app, saw $12,847 in MRR across three products, and realized I hated every single hour it took to earn it.
I was working 72 hours a week. The business was “working.” I was not.
That night I did something embarrassing. I opened a Google Doc and typed out every single task I did in a typical week — every email, every support ticket, every marketing post. Then I went through each line and asked one question: “Does this actually require a human, or am I doing it because I’ve always done it this way?”
The answer for 85% of my tasks was “no.”
That document is still on my computer. It’s 14 pages long. It’s the most important thing I’ve ever written for this business.
Eighteen months later, that same portfolio generates $35,000/month. I work roughly 12 hours a week. Most of what’s on those 14 pages runs on its own now.
Here’s exactly what I did — mistakes, fake starts, wasted money, and all.

The Portfolio Breakdown

Let me just show you the numbers first, sorted by contribution:
Product 1 — $15,500/month — B2B customer support tool
Product 2 — $12,000/month — B2B lead qualification platform
Product 3 — $7,500/month — A small SEO utility with loyal users
Total MRR: $35,000
Weekly revenue: ~$8,100
Daily revenue: ~$1,160
These numbers aren’t interesting. What’s interesting is what I’m NOT doing to earn them.
Every “SaaS guru” on YouTube will tell you the key to passive income is “systems and processes.” They’re technically right. They’re also useless, because they never show you the actual systems.
Here’s mine. No vague theory. The specific tools, the specific workflows, and the specific mistakes I made building it.

What Month Four Looked Like

To understand the shift, you need to see the problem.
A typical Tuesday in month four looked like this:
• 6:30am: Wake up. Check email. 47 new messages.
• 7:00am: Respond to urgent support tickets. Miss breakfast.
• 9:00am: Manually onboard three new customers. Send welcome emails. Add them to the CRM.
• 11:00am: Write a blog post for marketing.
• 1:00pm: Record a voiceover for a tutorial video. Mess up three takes.
• 3:00pm: Edit the video. Poorly.
• 5:00pm: Send outreach emails to 15 potential partners.
• 7:00pm: Fix a bug a customer reported.
• 9:00pm: Write the next day’s social media posts.
• 11:00pm: Go to bed. Forget to eat dinner.
• 11:47pm: Wake up to a bug report. Fix it in bed.
I still have the time-tracking spreadsheet from that period. The hourly rate I was earning, when you did the math against hours worked, was worse than a fast food shift.
That’s not a business. That’s a prison you built for yourself with better branding.

The Real Shift: Stop Using Tools. Start Building Systems.

Here’s what almost every SaaS automation article gets wrong.
They tell you to “use ChatGPT” or “use Zapier” like these are the answer. They’re not. Any idiot can subscribe to ChatGPT. That’s not leverage — that’s a button you pressed.
The leverage comes from what happens between the tools.
Here’s a real workflow from my current setup:
When a customer signs up for Product 1:
1. Stripe processes payment → triggers Zapier webhook
2. Customer data syncs to my Notion CRM
3. Beehiiv adds them to a 7-email onboarding sequence written by Claude
4. HeyGen-generated welcome video gets sent with their name interpolated
5. ElevenLabs-voiced tutorial content unlocks in their dashboard
6. A Slack notification pings me only if they’re an enterprise tier or above
7. Zapier schedules a 30-day check-in task for me (human judgment required)
Eight things happen. Seven of them don’t involve me. The one that does — the check-in — only fires if the customer’s revenue actually justifies my time.
This took me four months to build properly. I broke it at least a dozen times along the way. The first version sent welcome emails to customers twice because I didn’t set up the trigger condition correctly. I actually lost two customers in month five because of that mistake.
Nobody tells you that automation means debugging. It means hours of testing edge cases. It means a customer eventually experiencing something weird because your workflow didn’t account for a scenario you didn’t think of.
If you want a portfolio that runs without you, you have to accept this part. There is no clean version.

The Tool Stack That Actually Matters

Let me walk you through exactly what I use, why, and what it replaced.

ChatGPT + Claude for Content
Claude handles the stuff that needs actual thinking. Product research. Long-form blog posts. Complex customer emails where tone matters.
ChatGPT handles the repetitive stuff. Pattern-based content. Social media variations. Email subject line tests.
I use both because they’re genuinely different. Claude is more careful. ChatGPT is faster. Knowing which one to use for which task probably saved me 15 hours a week once I figured it out.
Previous cost for content (part-time writer): $2,000/month
Current cost (Claude Pro + ChatGPT Plus): $40/month
Pictory for Video Multiplication
Every blog post I write becomes a YouTube video automatically through Pictory.
The workflow is embarrassingly simple: paste the blog URL, wait 30 minutes, download the video, upload to YouTube. Done.
One blog post I turned into a Pictory video in early 2025 still gets 400-600 views per week. The affiliate link in the description has generated $87/month passively for over a year. I haven’t touched it since publishing.
Now multiply that by 40 videos.
Previous cost for video production: $1,500/month (editor)
Current cost for Pictory: $25/month
👉 Try Pictory free — Use code SHEA20CC for 20% off
ElevenLabs for Voice at Scale
This tool surprised me most.
I used to hire voice actors for onboarding videos. The process was painful — finding someone, briefing them, waiting for recordings, revising. Each video took 2-3 weeks from request to delivery.
ElevenLabs collapsed that to 20 minutes.
There’s a second layer I didn’t expect: the Voice Library. I uploaded my voice clone once in 2023. It still earns $180-240/month passively because other users generate content with it. I’ve done nothing with that voice clone since the day I uploaded it.
That’s not a business I’m running. That’s an asset I forgot I owned.
👉 Try ElevenLabs free
Beehiiv for Unified Communication
All my customer communication — newsletters, product updates, win-back campaigns — runs through Beehiiv.
The unexpected part: the newsletter itself became an income stream. At 4,500 subscribers, affiliate commissions from naturally recommended tools generate around $1,800/month. The newsletter wasn’t designed to make money. It was designed to communicate with customers. The monetization was a bonus.
That’s the theme of this entire stack. Each tool was added to solve a specific pain point. The income streams that emerged were mostly accidents.
👉 Start free with Beehiiv
HeyGen for Faceless Video Scale
I was late to this one. I only added HeyGen in 2025, after months of resisting avatar-generated video. I thought customers would notice. I thought it would feel “fake.”
I was wrong.
Customers don’t care who’s on screen. They care whether the information is clear. HeyGen’s avatars deliver clear information faster than I could record it myself. My customer activation rate went up 14% within two months of implementation.
The lesson I keep relearning: my resistance to new tools was always more about ego than reality.
👉 Try HeyGen free
Zapier (and Sometimes Make.com)
This is the glue. The boring middle layer that nobody talks about.
I run 34 active Zaps across the three products. Most of them are small — a customer signs up here, a record updates there, a Slack notification fires when something important happens.
The boring truth about automation: most of the value is in the small connections, not the flashy AI outputs. A simple Zap that moves data from Stripe to my CRM saves me 3 hours a week. That’s 156 hours a year. From one automation that took 20 minutes to build.

The Mistakes That Cost Me Real Money

Here’s the part nobody else will tell you.
Mistake 1: Over-automating too early.
I automated the entire customer support workflow in month three. Customers hated it. The AI responses were close enough to useful that customers felt dismissed but not close enough to actually solve problems. I lost 14 customers that month. $4,200 in MRR.
Mistake 2: Using the wrong tool for the job.
I tried to use ChatGPT to write long-form product content for six months before switching to Claude. The content was generic. Rankings were mediocre. Once I moved to Claude for long-form, organic traffic doubled within four months.
Mistake 3: Skipping the boring documentation.
I didn’t document my workflows for almost a year. When my assistant quit in month 11, I spent three weeks recreating processes from memory. Losing institutional knowledge because I was “too busy” to document cost me real revenue during that recovery period.
Mistake 4: Trusting the tools too much.
Pictory once generated a video with wildly incorrect captions. I didn’t check before uploading. A customer pointed it out two weeks later. The video had racked up 3,000 views with embarrassing errors in the captions. Embarrassing. Fixable. But an avoidable branding hit.
The automation works. It’s not magic. You still need to check what it produces. The people who claim otherwise are selling courses.

The Real Numbers Behind the $35K

Monthly tool cost:
• ChatGPT Plus: $20
• Claude Pro: $20
• Pictory Standard: $25
• ElevenLabs Creator: $22
• Beehiiv Growth: $49
• HeyGen Creator: $29
• Zapier Professional: $49
• Notion Plus: $12
• Stripe fees (variable)
• Hosting + infrastructure: $180
Total: ~$406/month
What it replaced:
• Content writer: $2,000/month
• Video editor: $1,500/month
• Voice actor: $800/month
• VA: $1,200/month
• Part-time support rep: $2,500/month
Equivalent human cost: $8,000/month
The arbitrage isn’t “AI is free.” The arbitrage is: $406 in tools replaces $8,000 in humans while producing more consistent output. That’s a 20x cost reduction for equivalent work.
Nobody who took a business course in the last 30 years would believe this was possible. It is. It’s what’s happening, quietly, while people argue about whether AI is real.

What I’d Do If Starting Today With $100

Month 1-2: Build one product. Use Claude for research, positioning, and copy. Launch it manually. Do everything yourself.
Month 3-4: Start the content engine. Pictory + one content platform. Ignore everything else.
Month 5-6: Add Beehiiv. Start the newsletter. Connect it to customer signups through one basic Zap.
Month 7-9: Layer in ElevenLabs for any voiceover work. Add HeyGen for video content at scale.
Month 10-12: Build the Zapier middle layer. Connect everything. This is the most tedious and least glamorous phase, and it’s where the actual leverage emerges.
Year 2: Add a second product into the existing infrastructure. The system you built for product 1 amplifies product 2 from day one.
Year 3: Third product. By now you’re not building from scratch. You’re extending a platform.
This isn’t theoretical. This is roughly what I did, minus the embarrassing detours.

What the Gurus Get Wrong

Every YouTube channel about SaaS automation is selling you a shortcut.
“Build a SaaS in a weekend.” “Use this one prompt to automate your entire business.” “Here’s the 10X framework.”
They’re lying. Or they’re oversimplifying to sell courses. Same result.
Real automation takes months of setup work for a payoff that emerges quietly. There are no magic prompts. There’s no tool that replaces the strategic thinking required to figure out what to automate and when. The work is in the middle layer — the connections between tools — and that middle layer is tedious and specific to your business.
If you’re looking for a product that promises to automate everything with one click, close this tab. That product doesn’t exist.
What exists is a set of tools that, properly connected, give a solo operator the output of a 15-person team. That’s the actual opportunity. Anyone selling you something simpler is selling you a fantasy.

The Part I’m Still Bad At

I want to end with something honest.
I’m still bad at knowing when NOT to automate something.
There’s a version of this business where every interaction is optimized, every workflow is perfect, every email is generated. That version also loses the human warmth that actually makes customers loyal. I’ve watched competitors over-automate into sterility, and their churn rates went up.
The hardest skill in this game isn’t building systems. It’s knowing when to leave something human on purpose. I still get that wrong regularly. The best I can do is catch it when customers tell me something feels cold, and then dial back the automation on that specific touchpoint.
This is the unsolved problem. Anyone telling you otherwise hasn’t actually run a business at scale.

The Final Point

$35,000/month sounds like a lot. It is. It’s also just three small SaaS products with a few hundred paying customers each, connected by tools that cost less than most people’s car payments.
Nothing about this is impossible. Most of it is actually boring.
The builders earning real income from SaaS portfolios in 2026 aren’t the ones with the most brilliant ideas. They’re the ones who put in six months of tedious infrastructure work while everyone else was chasing the next shiny thing.
I still have that 14-page Google Doc from month four. Every now and then I open it, look at how much of that task list is now automated, and remind myself that the gap between “drowning in work” and “working 12 hours a week” was mostly a tooling problem I couldn’t see at the time.
That’s the real lesson.

Learn more about the specific tools: Pictory, ElevenLabs, Beehiiv, HeyGen, ChatGPT, Claude. See how these connect in my breakdown of 3 lazy SaaS tools for passive income or my analysis of $24.5K MRR growth.