We reduced our cloud bills from +$7,789.45 → $2,115.11 (on average).
But most importantly: The less we depend on specific vendors, the less locked-in we become. +80% of our infrastructure can now be hosted practically anywhere.
Biggest misconceptions I see:
❌ Self-hosting is extremely complex. Reality: It's less complex and "scary" than you think.
❌ AWS, Azure GCP make hosting easy. Reality: They can get very complex too. And large bills can hit hard and unexpectedly.
I solved OpenClaw's memory issue. (At least, this is the best solution I tried so far).
And yes. Big surprise. I solved it using RAG.
All major models can now take millions of tokens as context. But the issue remains...
The undisputed number #1 reason your agent is stupid and forgets simple things is because you're bloating context.
So here's how I solved it with OpenClaw.
1️⃣ I installed PostgreSQL + pgvetor I run my OpenClaw on a Hetzner server, so I installed the database + pgvector extension directly on here.
2️⃣ Create a search tool Ask your agent to create a search tool for itself.
Every time you ask it to remember something, it should: - Label the memory - Create a vector from the label - Store the label, vector, and raw text in the database
Every time it's asked something it doesn't know, the FIRST thing it should always do is to use the search tool.
3️⃣ Memory CRON/heartbeat The agent can write to its memory file on-the-go. Consider this short-term memory.
On a scheduled CRON (or heartbeat), it should "flush" its own short-term memory and store it in the database.
Now, on every new session, the agent has very little context. The most important one is the description of using the search tool to enhance itself based on the task it's given.
✨ Major upsides
- MUCH better memory - MUCH smarter - MUCH less token-greedy
👎 Major downsides
- More moving parts - Complex for non-devs - Ongoing maintenance
Still. Benefits outweigh the cons here.
If you REALLY want to use OpenClaw professionally, I recommend that you use this 3-tool combo as a base:
I have a secret Discord. In it, I have a full team of AI executives.
Working together. Taking initiative. Executing 24/7.
I worked most of 2025 patching this together. OpenClaw finally made it very real.
Here's what it handles right now:
1️⃣ Accounting Files bookings, handles payroll, and prepares taxes through Bexio's API.
2️⃣ YouTube Knows my entire channel history, helps outline and write scripts, and syncs everything with Notion (the collaborative editing experience in Notion with OpenClaw is unreal!)
3️⃣ Blogs Does keyword research, drafts SEO‑optimized articles, and publishes them. I alrady had this in n8n, but it's 100x better now.
4️⃣ Retention Connects Stripe with Aidbase to find churn patterns in support tickets and suggests actions. It catches details and connects the dots in ways I never could.
5️⃣ Management Reads my Todoist, prioritizes tasks, and helps me plan the day. Nudges me and reminds me when I'm late.
6️⃣ Sleep & Health Uses WHOOP data to decide how hard I should push based on recovery and sleep quality.
7️⃣ Content Creates a weekly content plan that matches my brand. Coordinates between blogs, YouTube, and social media.
8️⃣ Groceries Tracks what we need at home, builds grocery lists, and even orders items online. (Warning, this is experimental and lowkey dangerous. I use a designated, prepaid debit card for this.)
This is not "AI will do everything for you". It is "AI turns into a team of specialists you can orchestrate".
If you're a SaaS founder and you are still treating AI like a toy, you are missing out on the biggest leverage upgrade of your career.
OpenClaw + n8n. This is an extremely powerful combination.
People thought OpenClaw (or Claude Code) would replace n8n, but here's what they're missing.
n8n - as an application layer between the agent and the tool - still makes perfect sense for at least 3 reasons.
1️⃣ Observability
OpenClaw can write its own skills. But instead of letting it script away, ask it to create n8n workflows for itself to use.
It is much easier for you to investigate and see what the agent built for itself in n8n, rather than looking through 100 ugly written JavaScripts.
2️⃣ Security
After the agent is done creating its n8n workflow, you can lock in place. Making it read-only from that point.
Instead of adding API keys to .env.local for the agent to use (and abuse) in any way it likes, you can now add the credentials securely to n8n.
From here, you can also easily add any additional safeguarding step, making sure your agent doesn't make a mess by mistake.
3️⃣ Performance
An agent adds value when it needs to make decisions. But a lot of work is still predictable and deterministic.
Turn it into a workflow. It's faster and you save tokens.
🔁 The flow
- The agent needs access to an API. - It writes an n8n workflow with incoming webhook. - You lock the workflow and add the API key. - You add extra safeguarding steps.
The agent now proxies all calls through n8n. It never sees the API key. It's prevented from making crucial mistakes.
I know... It feels addictive to let OpenClaw do everything.
I'm trying my best not to hype AI. But it's just one of those moments!
Last week, I installed OpenClaw - and it's taking over my life.
I spent most of 2025 stitching together AI tools.
- n8n - TypingMind - OpenAI - ...a lot of duct tape
And this project is EXACTLY what I was trying to achieve.
OpenClaw now runs on a VPS with Hetzner. Connected to my personal Discord server.
It has access to everything it needs. Calendar, Todoist, Notion, Whoop, Aidbase, and much more...
It even uses the browser very intuitively, so it can find contractors on Upwork and YTJobs when it realizes its limitations as an AI.
- It runs in the background 24/7. - It's taking initiative and checking in with me. - It creates its own skills and tools when needed.
It TRULY feels autonomous. We're getting there 💪
---
⚠️ A few important caveats!
- Security! This agent is autonomous. It can ruin your life if you let it. Use with utmost care. Run in a sandbox first, on a separate device, with limited internet connection, and very gradually introduce privileges.
- Cost! For me, it only truly works well with Claude Opus 4.5. And it's very expensive. If you use "dumber" models, you'll likely get a very limited experience.
Simon Høiberg
n8n is a free SaaS idea catalog.
There is an arbitrage opportunity that can take you from idea to profitable SaaS with near-zero risk and a very low budget.
Here's how it works:
1️⃣ Browse n8n workflows
Stop "imagining" what your SaaS is going to be about. Look for real problems people are solving with n8n.
Look at n8n's website.
Search on YouTube.
Find 1-2 popular workflows.
2️⃣ Break it down
Essentially, an n8n workflow is a problem that's being solved. Break it down.
The basic mechanism.
The outcome.
3️⃣ Open Lovable
n8n is awesome. But it's not for everyone. Recreate the idea behind the workflow in a much more user-friendly way.
Vibe code the front end.
Ask AI to analyze the workflow.
Recreate the backend.
4️⃣ Charge
Put your solution behind a paywall using Stripe checkout.
5-minute implementation.
Start charging from day one.
Now start marketing your solution.
- Run ads.
- Use social media.
- Use cold outreach.
Worst-case: If this doesn't work, you've spent a few hundred dollars and some hours of work.
Best-case: You'll have a SaaS solving a real problem for real users paying real money.
1 week ago | [YT] | 239
View 8 replies
Simon Høiberg
This year, we started moving off "the cloud" - 80% of our infrastructure now runs on bare-metal servers.
AWS → Hetzner.
Instead of:
- DynamoDB
- ElastiCache
- ECS/Fargate
- SQS
- SNS
- CloudWatch
- X-Ray
We use:
- PostgreSQL
- Redis
- Kubernetes
- RabbitMQ
- BullMQ
- Grafana
- Prometheus
We reduced our cloud bills from +$7,789.45 → $2,115.11 (on average).
But most importantly:
The less we depend on specific vendors, the less locked-in we become. +80% of our infrastructure can now be hosted practically anywhere.
Biggest misconceptions I see:
❌ Self-hosting is extremely complex.
Reality: It's less complex and "scary" than you think.
❌ AWS, Azure GCP make hosting easy.
Reality: They can get very complex too. And large bills can hit hard and unexpectedly.
Here's the breakdown 👇
1 week ago | [YT] | 349
View 14 replies
Simon Høiberg
I solved OpenClaw's memory issue.
(At least, this is the best solution I tried so far).
And yes. Big surprise.
I solved it using RAG.
All major models can now take millions of tokens as context. But the issue remains...
The undisputed number #1 reason your agent is stupid and forgets simple things is because you're bloating context.
So here's how I solved it with OpenClaw.
1️⃣ I installed PostgreSQL + pgvetor
I run my OpenClaw on a Hetzner server, so I installed the database + pgvector extension directly on here.
2️⃣ Create a search tool
Ask your agent to create a search tool for itself.
Every time you ask it to remember something, it should:
- Label the memory
- Create a vector from the label
- Store the label, vector, and raw text in the database
Every time it's asked something it doesn't know, the FIRST thing it should always do is to use the search tool.
3️⃣ Memory CRON/heartbeat
The agent can write to its memory file on-the-go.
Consider this short-term memory.
On a scheduled CRON (or heartbeat), it should "flush" its own short-term memory and store it in the database.
Now, on every new session, the agent has very little context. The most important one is the description of using the search tool to enhance itself based on the task it's given.
✨ Major upsides
- MUCH better memory
- MUCH smarter
- MUCH less token-greedy
👎 Major downsides
- More moving parts
- Complex for non-devs
- Ongoing maintenance
Still. Benefits outweigh the cons here.
If you REALLY want to use OpenClaw professionally, I recommend that you use this 3-tool combo as a base:
- OpenClaw itself
- PostgreSQL + pgvector
- n8n (for API proxies/security)
1 week ago | [YT] | 355
View 8 replies
Simon Høiberg
I have a secret Discord.
In it, I have a full team of AI executives.
Working together.
Taking initiative.
Executing 24/7.
I worked most of 2025 patching this together.
OpenClaw finally made it very real.
Here's what it handles right now:
1️⃣ Accounting
Files bookings, handles payroll, and prepares taxes through Bexio's API.
2️⃣ YouTube
Knows my entire channel history, helps outline and write scripts, and syncs everything with Notion (the collaborative editing experience in Notion with OpenClaw is unreal!)
3️⃣ Blogs
Does keyword research, drafts SEO‑optimized articles, and publishes them. I alrady had this in n8n, but it's 100x better now.
4️⃣ Retention
Connects Stripe with Aidbase to find churn patterns in support tickets and suggests actions. It catches details and connects the dots in ways I never could.
5️⃣ Management
Reads my Todoist, prioritizes tasks, and helps me plan the day. Nudges me and reminds me when I'm late.
6️⃣ Sleep & Health
Uses WHOOP data to decide how hard I should push based on recovery and sleep quality.
7️⃣ Content
Creates a weekly content plan that matches my brand. Coordinates between blogs, YouTube, and social media.
8️⃣ Groceries
Tracks what we need at home, builds grocery lists, and even orders items online. (Warning, this is experimental and lowkey dangerous. I use a designated, prepaid debit card for this.)
This is not "AI will do everything for you".
It is "AI turns into a team of specialists you can orchestrate".
If you're a SaaS founder and you are still treating AI like a toy, you are missing out on the biggest leverage upgrade of your career.
3 weeks ago | [YT] | 342
View 9 replies
Simon Høiberg
OpenClaw + n8n.
This is an extremely powerful combination.
People thought OpenClaw (or Claude Code) would replace n8n, but here's what they're missing.
n8n - as an application layer between the agent and the tool - still makes perfect sense for at least 3 reasons.
1️⃣ Observability
OpenClaw can write its own skills.
But instead of letting it script away, ask it to create n8n workflows for itself to use.
It is much easier for you to investigate and see what the agent built for itself in n8n, rather than looking through 100 ugly written JavaScripts.
2️⃣ Security
After the agent is done creating its n8n workflow, you can lock in place. Making it read-only from that point.
Instead of adding API keys to .env.local for the agent to use (and abuse) in any way it likes, you can now add the credentials securely to n8n.
From here, you can also easily add any additional safeguarding step, making sure your agent doesn't make a mess by mistake.
3️⃣ Performance
An agent adds value when it needs to make decisions.
But a lot of work is still predictable and deterministic.
Turn it into a workflow.
It's faster and you save tokens.
🔁 The flow
- The agent needs access to an API.
- It writes an n8n workflow with incoming webhook.
- You lock the workflow and add the API key.
- You add extra safeguarding steps.
The agent now proxies all calls through n8n.
It never sees the API key.
It's prevented from making crucial mistakes.
I know... It feels addictive to let OpenClaw do everything.
But this thing is a beast!
Use it responsibly.
3 weeks ago | [YT] | 345
View 7 replies
Simon Høiberg
I'm trying my best not to hype AI.
But it's just one of those moments!
Last week, I installed OpenClaw - and it's taking over my life.
I spent most of 2025 stitching together AI tools.
- n8n
- TypingMind
- OpenAI
- ...a lot of duct tape
And this project is EXACTLY what I was trying to achieve.
OpenClaw now runs on a VPS with Hetzner.
Connected to my personal Discord server.
It has access to everything it needs.
Calendar, Todoist, Notion, Whoop, Aidbase, and much more...
It even uses the browser very intuitively, so it can find contractors on Upwork and YTJobs when it realizes its limitations as an AI.
- It runs in the background 24/7.
- It's taking initiative and checking in with me.
- It creates its own skills and tools when needed.
It TRULY feels autonomous.
We're getting there 💪
---
⚠️ A few important caveats!
- Security! This agent is autonomous. It can ruin your life if you let it. Use with utmost care. Run in a sandbox first, on a separate device, with limited internet connection, and very gradually introduce privileges.
- Cost! For me, it only truly works well with Claude Opus 4.5. And it's very expensive. If you use "dumber" models, you'll likely get a very limited experience.
4 weeks ago | [YT] | 263
View 9 replies
Simon Høiberg
Running a lean, bootstrapped (solo) business requires using a lot of tools.
But be careful.
Don't let it get out of hand.
⚠️ You want to avoid:
- High costs
- +30 open tabs
- Overlapping functionality
- Data scattered across many tools
You can fix this by grouping your tools.
✅ Self-hosted
Prefer tools you can self-host.
- They are often free.
- Data never leaves your servers.
Tools like n8n, NocoDB, and Grafana is preferred over Zapier, Airtable, and PostHog.
✅ Automation
Prefer tools where you can automate workflows.
- Can run by itself.
- Requires you to open the tool less often.
Tools like Lindy AI, n8n, and LinkDrip are tools I use heavily.
✅ API-based
Prefer tools that have APIs you can call.
- Little need to click around a UI.
- Can be integrated easily with other tools.
I use OpenAI & Replicate heavily for AI.
And I use AWS SES to send emails.
Using a chain of tools isn't in itself a problem.
But you need to set yourself up properly and stay organized.
ℹ️ Did you know? FounderStack is an ecosystem of tools that integrates seamlessly.
You get:
- Social media management
- AI support
- Link tracking
- Graphics design
- Fake account prevention
One bundle, in multiplayer mode, for a single one-time purchase.
Get it on www.founderstack.pro/
4 weeks ago | [YT] | 221
View 1 reply
Simon Høiberg
"Wow, this actually works. I wonder why Intercom can't pull this off"
That's what a user told me after switching from Intercom to Aidbase. This happened after Intercom's CEO announced he would invest $100M into AI.
So how can we, a small 3-man team, create better value for our users than Intercom?
It's simple.
When big companies invest, what they're actually spending money on is:
- Expensive cloud infrastructure
- Expanding their teams
- Adding more mid-level managers
- More process and bureaucracy
Instead of what really matters:
- Creating value for their users.
With a small, lean team, creating value is just one click on the green "Merge" button away.
That's why I keep betting on small teams.
4 weeks ago | [YT] | 217
View 8 replies
Simon Høiberg
You can block your favorite SaaS guru.
If they talk about the latest growth hacks, they do more harm than good.
If you're bootstrapping, following the wrong playbook can break your business.
It sounds smart...
But in reality, you're just wasting time.
- Stop hacking around.
- Start doing boring stuff that works.
- Put in more reps.
Here's what you wanna avoid 👇
4 weeks ago | [YT] | 253
View 6 replies
Simon Høiberg
I don't look at my code anymore.
Instead:
- I let the AI do its thing.
- I do manual QA in staging.
- I deploy to production.
Why care about "clean code" if my AI agents are the only ones reading it?
AI-assisted coding is your biggest advantage in 2026.
If you let it!
But there are a few habits you need to unlearn 👇
4 weeks ago | [YT] | 406
View 36 replies
Load more