https://learn.microsoft.com/en-us/users/yasirali-3997/transcript/vj58xiyxkkoj12x

Yasir Ali, chakwal, Chakwal (2026)

08/02/2026

Training an AI agent is less about code,and more about psychology.

Recently, I worked on setting up Retell AI with n8n, and the experience completely changed how I think about conversational AI.

The technical setup itself was smooth:
connecting Retell AI with n8n workflows, handling triggers, managing API calls, and orchestrating logic felt like assembling a well-designed automation puzzle. But the real learning began after the integration was complete.

What stood out most was training the agent on behavior.

I learned that building an effective AI voice or chat agent isn’t just about prompts—it’s about:

Defining intent clearly

Anticipating edge cases in real conversations

Teaching the agent how to respond, pause, clarify, and escalate

Designing conversations that feel natural, not robotic

Every small change in instructions had a visible impact on how the agent behaved. It felt less like programming and more like coaching a digital team member.

The most interesting part?
Seeing how workflow automation (n8n) and conversational intelligence (Retell AI) come together to create agents that can actually understand context, follow logic, and deliver real value.

This experience reinforced one key insight for me:

> The future of AI isn’t just smart models—it’s well-designed behavior.

If you’re working with AI agents, voice automation, or workflow orchestration, I highly recommend exploring this combination. The learning curve is worth it.

Curious to hear—how are you designing behavior and personality in your AI agents?

05/02/2026

n8n vs GoHighLevel: What Should Technical Teams & Startups Choose?

When you’re building fast and scaling smarter, the tools you choose matter. For technical teams and startups, automation isn’t just about convenience — it’s about control, flexibility, and long-term scalability.

Two platforms often compared are n8n and GoHighLevel (GHL). While both automate workflows, they’re built for very different mindsets.

🧩 n8n — Built for Builders

Pros:
✅ Open-source & self-hostable — Full control over infrastructure, data, and costs.

✅ Highly customizable workflows — Build exactly what your product or internal tools need.

✅ Developer-friendly — API-first, supports custom code, and complex logic.

✅ Scales with your architecture — Ideal for microservices and modern stacks.

Cons:

⚠️ Higher learning curve — Requires technical expertise to unlock full value
⚠️ Less “plug-and-play” — You design the system instead of relying on templates

👉 Best for: Startups with engineering teams, CTO-led decisions, and products requiring custom integrations.

📊 GoHighLevel — Optimized for Speed, Not Depth

Pros:
✅ Fast setup — Prebuilt CRM, marketing automations, and pipelines.

✅ Minimal engineering effort — Great for quick go-to-market needs.

✅ Centralized tools — Useful if sales and marketing are the priority.

Cons:

⚠️ Limited customization — Workflows are constrained by the platform.

⚠️ Not engineering-first — Less suitable for complex logic or product-level integrations.

⚠️ Vendor lock-in risk — Harder to decouple as your stack matures.

👉 Best for: Early-stage startups focused on sales ex*****on rather than deep technical automation.

---

⚖️ Which One Is Better for Startups & Technical Teams?

✔️ Choose n8n if you’re building custom systems, value architecture flexibility, and want automation that evolves with your product.
✔️ Choose GHL if you need to validate fast, run lean marketing ops, and avoid engineering overhead early on.

💡 Final Takeaway:
Startups often outgrow tools that prioritize convenience over control. If automation is part of your core infrastructure, n8n is the stronger long-term choice. If automation is just supporting sales, GHL can get you moving quickly.

💬 What’s your priority right now — speed to market or technical control?
Let’s discuss.

01/02/2026

Hidden costs in n8n don’t come from bugs.
They come from bad architecture.

Most unexpected bills in n8n automation are not caused by traffic spikes.
They come from silent mistakes that slowly burn APIs, tokens, and compute.
If you’re running AI workflows, API integrations, or agent-based automations, these practices matter.
1️⃣ Treat API calls like money
Every node ex*****on costs something.
Unfiltered triggers and looping workflows can multiply calls without you noticing.
✔ Add conditions before expensive nodes
✔ Validate input early
✔ Stop workflows fast when data is invalid
2️⃣ Control AI token usage explicitly
LLMs don’t know when to stop. You must tell them.
✔ Limit context size
✔ Avoid passing full ex*****on data into prompts
✔ Summarize once, reuse many times
✔ Cache AI responses where possible
Tokens are infrastructure cost, not usage cost.
3️⃣ Use batching instead of per-item ex*****on
Calling an API 1,000 times vs batching 1,000 items is the difference between a clean system and a surprise invoice.
✔ Use Split In Batches
✔ Aggregate data before AI or API nodes
4️⃣ Add hard ex*****on limits
Runaway workflows are real.
✔ Set max ex*****on time
✔ Add safety counters in loops
✔ Use error workflows to stop retries from looping endlessly
5️⃣ Log smarter, not louder
Excessive logging increases storage and slows ex*****on.
✔ Log only what you need
✔ Remove debug nodes in production
✔ Store summaries, not raw payloads
6️⃣ Secure secrets properly
Leaked keys don’t just risk security.
They create uncontrolled cost.
✔ Use environment variables
✔ Rotate keys regularly
✔ Never hardcode secrets in workflows or GitHub repos
Good automation is predictable automation.
If you can’t predict cost, the system isn’t production-ready.
n8n gives you power.
Architecture decides whether that power saves money or burns it.

31/01/2026

AI News to Content Automation is an end-to-end automated pipeline that transforms real-time AI news into ready-to-publish LinkedIn content. The system continuously monitors trusted AI news sources, intelligently selects the most relevant story, generates SEO-optimized post copy, creates a context-aligned visual, and delivers the final output automatically. Built with a modular and scalable architecture, this workflow eliminates manual research, writing, and design, enabling consistent, high-quality content generation with zero human intervention

30/01/2026

End-to-End WhatsApp Connection in n8n (Latest Graph API v24)

If your WhatsApp token expires every day while using n8n,
don’t panic. This is a very common issue.

Most people don’t realize that:

The temporary WhatsApp access token is valid for only 60 minutes

Once it expires, workflows fail

This creates the impression that the WhatsApp or n8n connection is unstable

In reality, this is expected behavior.

Temporary tokens are meant for testing only.
For production-level WhatsApp automation, you need a permanent system user token.

The good news 👇
With the latest Meta Graph API v24, WhatsApp now supports a proper method to generate a one-time, long-lived access token for stable integrations.

I can guide you step by step through:

Meta Business Manager configuration

System User creation

Correct WhatsApp permission scopes

Permanent token generation

Secure WhatsApp integration in n8n

Eliminating token expiry issues permanently

If your WhatsApp token:

Expires daily

Becomes invalid after 60 minutes

Stops working after a Graph API update

The issue is not n8n.
It’s the type of token being used.

With WhatsApp Cloud API and Graph API v24,
stable and production-ready WhatsApp automation is absolutely possible
when the setup is done correctly.

If you need help fixing token issues or building
end-to-end WhatsApp automation in n8n, feel free to reach out.

28/01/2026

Creating AI content isn’t hard. Doing it manually is.

Every single day, high-quality AI news is published on platforms like:
1. Google AI Blog,
2. MIT Technology Review
3. ArtificialIntelligence-News.

But here was the real problem 👇
❌ Manually visiting multiple websites
❌ Scraping or copy-pasting articles
❌ Separating signal from noise
❌ Choosing one strong topic
❌ Writing a LinkedIn-ready post
❌ Designing or sourcing an image
❌ Saving or sharing everything manually

This process used to take hours, and consistency was almost impossible.

⚙️ The Solution: End-to-End AI News Automation
I built a fully automated workflow in n8n that now:
✅ Pulls live AI news automatically from:
1. Google AI Blog (RSS)
2. MIT Technology Review
3. ArtificialIntelligence-News
✅ Uses AI to shortlist the most impactful topic
✅ Generates a SEO-optimized, human-sounding LinkedIn post
✅ Creates an AI-ready image prompt and generates the image
✅ Delivers the final post + image automatically to WhatsApp
🔥 What changed?
• No more manual scraping
• No repetitive writing
• No context switching
• Fully consistent, daily AI content
Content creation is now strategic and creative.
Everything repetitive is handled by automation.

If you’re serious about:
AI content creation
LinkedIn growth
Or building real-world AI workflows
Automation is no longer optional.
This is how modern content engines are built.

27/01/2026

ClawDBot: Hype or the Next Shift in Automation?

This small lobster is doing big things.

ClawDBot launched in late January 2026 and crossed 9,000 GitHub stars in 24 hours. It is now sitting above 30,000 stars, with people literally buying Mac minis just to run it 24/7.

That alone tells you something changed.

What actually changed in automation?

In 2025, we already had powerful tools:

n8n for connecting apps

Claude Code for writing logic

APIs for ex*****on

But there was still a gap.

Planning was intelligent.
Ex*****on was automated.
Ownership of the task was still manual.

You had to trigger workflows, glue steps together, and babysit long-running tasks.

What ClawDBot is trying to solve

ClawDBot is not just a chatbot.

You can message it from WhatsApp, Telegram, or iMessage, and it does not stop at replying. It:

Browses websites

Manages inboxes

Writes and edits code

Builds new skills for itself

Executes multi-step tasks end to end

From an automation lens, this is important.

It shifts automation from workflow-driven to agent-driven.

The real breakthrough: persistent memory

The most interesting part is not tools or integrations.
It is persistent memory.

ClawDBot remembers context from weeks ago and uses it actively. That turns it from a task runner into a long-lived digital operator.

Reportedly, one user burned through 180 million Anthropic tokens in a week, using it for:

Daily briefings

Calendar management

Ongoing task ex*****on

Self-modifying capabilities

That is not a bot. That is an always-on system.

Why this matters for builders and teams

ClawDBot highlights where automation is heading:

Less rigid workflows

More autonomous agents

Memory as infrastructure

Ex*****on without constant human triggers

Tools like n8n are still critical. But they may become ex*****on layers, while agents like this handle intent, context, and decision-making.

So is it overhyped?

Maybe.
But dismissing it would be a mistake.

Even if ClawDBot itself evolves or fades, the model it represents is here to stay.

Automation is no longer just about connecting apps.
It is about delegating responsibility.

And that is a real shift.

26/01/2026

WhatsApp automations breaking on n8n? It’s not a bug. It’s Meta.

Meta has updated WhatsApp Graph API policies, and from Jan 15, 2026, general-purpose AI chatbots on WhatsApp are being restricted.
If your workflow looks like this:
n8n → LLM → open-ended replies
⚠️ You’re at risk.
What still works 👇
✅ Task-based business automations
✅ Customer support, order updates, reminders
✅ Template-based messaging that follows Graph API rules
The API didn’t really change.
The intent did.
If it behaves like a business system, you’re fine.
If it behaves like ChatGPT on WhatsApp, you’re not.

👉 If you’re facing issues with n8n, WhatsApp, or the Graph API, I can help you fix them.

26/01/2026

Most websites still use contact forms — and 70% of visitors never fill them.
So I tried replacing one with an AI chat agent using n8n + OpenAI. 👇

Today I built a simple chat workflow:

🔹 Trigger: “When chat message received”
🔹 Sends message to OpenAI Chat Model
🔹 Uses Simple Memory to keep conversation context
🔹 Replies instantly inside the website chat widget
🔹 Fully tested live with screenshots below

Why this matters:
Instead of waiting for someone to submit a form, an AI agent can:
• Answer FAQs
• Qualify leads
• Book services
• Collect customer info politely
• Work 24/7 without supervision

Next upgrades I want to build:

Save chat logs into Sheets/DB

Add routing to a human agent

Train the bot on custom business data

Just trying to learn automation through real mini-projects — step by step.

Screenshots attached ↓

25/01/2026

𝗦𝘁𝗼𝗽 𝘄𝗮𝘀𝘁𝗶𝗻𝗴 𝗵𝗼𝘂𝗿𝘀 𝗼𝗻 𝗿𝗲𝗽𝗲𝘁𝗶𝘁𝗶𝘃𝗲 𝘁𝗮𝘀𝗸𝘀!🛑 𝗥𝗲𝗮𝗱𝘆 𝘁𝗼 𝗽𝘂𝘁 𝘆𝗼𝘂𝗿 𝘄𝗼𝗿𝗸𝗳𝗹𝗼𝘄 𝗼𝗻 𝗮𝘂𝘁𝗼𝗽𝗶𝗹𝗼𝘁?
As an aspiring Azure AI & ML Engineer, I’m always looking for ways to connect smart models with real-world actions. That’s where n8n comes in! 🔗
n8n is a powerful, fair-code workflow automation tool that lets you build complex automations visually. Whether it's syncing data, sending automated alerts, or connecting Python scripts to your favorite apps, n8n handles it all with ease.

𝗪𝗵𝘆 𝗜’𝗺 𝗰𝗵𝗼𝗼𝘀𝗶𝗻𝗴 𝗻𝟴𝗻 𝗳𝗼𝗿 𝗺𝘆 𝗷𝗼𝘂𝗿𝗻𝗲𝘆:

𝗦𝗲𝗹𝗳-𝗛𝗼𝘀𝘁𝗲𝗱 & 𝗦𝗲𝗰𝘂𝗿𝗲: You have full control over your data.
𝗩𝗶𝘀𝘂𝗮𝗹 𝗪𝗼𝗿𝗸𝗳𝗹𝗼𝘄: Build logic using a "Node-based" interface—no more messy code for simple triggers.
𝗜𝗻𝗳𝗶𝗻𝗶𝘁𝗲 𝗙𝗹𝗲𝘅𝗶𝗯𝗶𝗹𝗶𝘁𝘆: From Webhooks to Python snippets, it’s a playground for engineers.

I’m currently exploring how to bridge Azure AI services with n8n to create smarter, autonomous workflows. Stay tuned as I share my builds!

Are you using any automation tools like 𝗭𝗮𝗽𝗶𝗲𝗿 or 𝗠𝗮𝗸𝗲? Or are you a fan of self-hosted solutions like n8n? Let’s discuss in the comments! 👇

Yasir Ali

08/02/2026

05/02/2026

01/02/2026

31/01/2026

30/01/2026

28/01/2026

27/01/2026

26/01/2026

26/01/2026

25/01/2026

Address

Website

Alerts

Shortcuts

Share

Category