DEV Community: Rahul singh Shekhawat

Why AI Agents Are Hard to Debug (And What We’re Missing)

Rahul singh Shekhawat — Tue, 24 Mar 2026 05:52:45 +0000

AI agents are getting powerful.

We can now build systems that:

Call APIs
Use tools
Chain multiple LLM steps
Make decisions autonomously

But there’s one problem I keep running into:

When something goes wrong, it’s incredibly hard to understand why.

The Illusion of Observability

Today, we have tools that provide:

Logs
Traces
Token usage
Cost tracking

These are useful.

But in practice, they answer only one question:

👉 “What happened?”

Not:

👉 “Why did it happen?”

A Simple Example

Imagine an AI agent:

Takes user input
Calls an API
Processes the response
Generates final output

Now suppose the final answer is wrong.

Where did it fail?

Was the prompt incorrect?
Did the tool return unexpected data?
Did the model misinterpret context?
Did a previous step introduce noise?

Most of the time, you’re left manually digging through logs.

The Real Problem: Debugging, Not Logging

We don’t just need better logs.

We need:

Step-by-step replay of workflows
Visibility into intermediate decisions
Clear identification of failure points
Understanding of how context evolves

In short:

We need debugging tools for AI systems, not just observability tools.

What’s Missing Today

From my experience, current workflows rely heavily on:

Manual inspection
Trial and error
Adding more logging
Using evals to detect issues

But even then:

👉 You still don’t get a clear answer to why something failed.

A Different Way to Think About It

Instead of asking:

“How do we log more?”

We should ask:

“How do we make AI systems debuggable?”

That means:

Replaying executions like a timeline
Highlighting where things diverged
Understanding cause → effect relationships
Reducing guesswork

Where I’m Heading

I’ve been exploring this space and working on something focused on:

Debugging multi-step AI workflows
Understanding root causes of failures
Improving trust in AI systems

Still early, but the goal is simple:

Help developers understand why their AI behaves the way it does.

Open Questions

If you’re working with AI:

How do you debug failures today?
Do you feel current tools are enough?
What’s the most frustrating part of working with AI systems?

Would love to hear your thoughts.

Why your AWS bill is a "Crime Scene" (and how I built an AI to prove it)

Rahul singh Shekhawat — Tue, 10 Mar 2026 18:32:39 +0000

I spent the weekend building a fun tool that uses AI to analyze cloud spending patterns.

In this post, I break down the 3 most common 'zombie' resources I found while testing:

Idle NAT Gateways
Unattached Managed Disks
Over-provisioned RDS instances
If you want to see where you stand, I put the tool live at: https://subtrackhub.com/roast

It's free, no signup, and it's brutally honest.

How We Built SubTrackHub: Solving the $40B Cloud Waste Problem 🚀

Rahul singh Shekhawat — Sun, 15 Feb 2026 14:14:43 +0000

Why we built SubTrackHub

The "billing fog" in modern cloud infrastructure is real. We analyzed 500+ AWS accounts and found that roughly 30-40% of spend is wasted on resources no one is using.

As a developer, there's nothing more frustrating than realizing you've been paying $100/mo for a t3.medium instance you forgot to delete three months ago.

The Technical Challenge

Building a "read-only" auditor that works across multi-cloud environments (AWS, GCP, Azure) and SaaS (GitHub, Vercel, Clerk) requires a robust integration strategy.

1. The Multi-Cloud "Deep Scan"

We didn't just want to pull high-level billing data. We built deep scans that look at:

CPU Utilization: Flagging instances with <5% average use over 7 days.
Orphaned EBS Volumes: Identifying storage billed for non-existent VMs.
Network Waste: Finding idle NAT Gateways and unattached Elastic IPs.

2. The Identity Gap

We built a GitHub Seat Optimizer that checks for "Ghost Users"—people who are being paid for in your organization but haven't made a commit or logged in for 30+ days.

The "Savage" Advantage: AI-Driven Remediation

Unlike other tools that just show you a graph of your waste, we built Savage AI Roadmaps.

It’s one thing to see a leak; it's another to fix it. Our AI engine generates an exact sequence of CLI commands and terraform adjustments for every leak we find. You don't have to go digging through the AWS Console—you just copy, paste, and save.

Check it out!

If you're a dev or founder tired of paying the "Shadow IT Tax," check out our internal guides or see the tool in action:

👉 View SubTrackHub
👉 Read our Guide: How to find Idle AWS Instances

SaaS developer, DevOps engineer, or startup founder Are Bleeding Money on Subscriptions (Here’s How to Fix It)

Rahul singh Shekhawat — Tue, 20 Jan 2026 20:25:33 +0000

If you’re a SaaS developer, DevOps engineer, or startup founder, chances are you’re paying for tools you no longer use.
Cloud services, CI/CD tools, monitoring platforms, AI subscriptions — modern development stacks are powerful, but they’re also silently expensive.
In this article, we’ll break down:
Why developers lose money on subscriptions
Why existing tools fail dev teams
How SubTrackHub helps track and optimize SaaS + cloud spend

💸 The Hidden Subscription Problem in Modern Dev Teams
Today’s development workflow relies heavily on subscription-based tools:

Cloud providers (AWS, GCP, Azure)
Hosting platforms (Vercel, Netlify, Render)
Dev tools (GitHub, Jira, Datadog, Sentry)
AI tools (Copilot, OpenAI, Claude)
Team tools (Slack, Notion, Figma)
What usually goes wrong?
Free trials auto-convert
Old tools stay active after migrations
Multiple billing accounts exist
No single dashboard shows everything
📌 Result:
Developers only notice the issue when the monthly invoice spikes.

🧠 Why Most Subscription Tracking Tools Don’t Work for Developers
Most subscription management tools are:

Built for finance teams, not developers
Manual and spreadsheet-driven
Lacking real cloud or usage context
Focused on reporting, not prevention
Developers need:

Automation
Usage-aware insights
Cloud-native visibility
Minimal manual work
That’s exactly the gap SubTrackHub is designed to fill.

🚀 What Is SubTrackHub?
SubTrackHub is a developer-first subscription and cloud spend tracking platform.

Our mission is simple:
Give developers full visibility and control over SaaS and cloud subscriptions before money is wasted.

⚙️ Key Features of SubTrackHub
✅ Unified Subscription Dashboard
Track all your:

SaaS tools
Cloud services
Developer platforms
From a single dashboard.
🔍 Usage-Aware Insights
Not just what you pay, but:

Which subscriptions are actually used
Which are idle or underutilized
Where downgrades or removals make sense
🤖 AI-Powered Cost Optimization
SubTrackHub uses AI to:

Detect unused subscriptions
Identify cost anomalies
Recommend optimizations
Create a clear cost-reduction roadmap
🧩 Built for Teams & Multi-Tenant SaaS
Designed for:

Startups
Growing SaaS teams
Agencies managing multiple projects
🧪 Why We’re Building This Now
The developer ecosystem has changed:

AI tools are exploding
Cloud bills are harder to predict
Teams are smaller but tool-heavy
But cost visibility hasn’t evolved with developer workflows.

👉 Cost control should be proactive, not reactive.
👨‍💻 Built by Developers, for Developers
SubTrackHub is built by developers who:

Run SaaS products
Deploy on AWS and GCP
Manage DevOps pipelines
Have personally overpaid for tools 😅

This isn’t a finance-first product.
It’s developer-first by design.

🛣️ What’s Next for SubTrackHub?
We’re actively working on:

Deeper cloud integrations
Smarter AI-driven insights
Team-level visibility improvements
Automation for unused resources
We’re also collecting early developer feedback to shape the roadmap.
🙌 Want to Try SubTrackHub?
If you’re:

A SaaS founder
A developer juggling multiple tools
A DevOps or cloud engineer
Someone tired of surprise invoices
👉 Check out SubTrackHub
👉 Share your feedback
👉 Tell us what subscription hurts the most

💬 Discussion
What’s one subscription you forgot you were paying for?

Drop it in the comments 👇
Let’s learn from each other.

SaaS developer, DevOps engineer, or startup founder Are Bleeding Money on Subscriptions (Here’s How to Fix It)

Rahul singh Shekhawat — Tue, 13 Jan 2026 20:20:30 +0000

💸 The Hidden Subscription Problem in Modern Dev Teams
Today’s development workflow relies heavily on subscription-based tools:

Cloud providers (AWS, GCP, Azure)
Hosting platforms (Vercel, Netlify, Render)
Dev tools (GitHub, Jira, Datadog, Sentry)
AI tools (Copilot, OpenAI, Claude)
Team tools (Slack, Notion, Figma)
What usually goes wrong?
Free trials auto-convert
Old tools stay active after migrations
Multiple billing accounts exist
No single dashboard shows everything

📌 Result:
Developers only notice the issue when the monthly invoice spikes.

🧠 Why Most Subscription Tracking Tools Don’t Work for Developers
Most subscription management tools are:

Built for finance teams, not developers
Manual and spreadsheet-driven
Lacking real cloud or usage context
Focused on reporting, not prevention

Developers need:

Automation
Usage-aware insights
Cloud-native visibility
Minimal manual work

That’s exactly the gap SubTrackHub is designed to fill.

🚀 What Is SubTrackHub?
SubTrackHub is a developer-first subscription and cloud spend tracking platform.

Our mission is simple:
Give developers full visibility and control over SaaS and cloud subscriptions before money is wasted.

⚙️ Key Features of SubTrackHub
✅ Unified Subscription Dashboard
Track all your:

SaaS tools
Cloud services
Developer platforms
From a single dashboard.

🔍 Usage-Aware Insights
Not just what you pay, but:

Which subscriptions are actually used
Which are idle or underutilized
Where downgrades or removals make sense

🤖 AI-Powered Cost Optimization
SubTrackHub uses AI to:

Detect unused subscriptions
Identify cost anomalies
Recommend optimizations
Create a clear cost-reduction roadmap

🧩 Built for Teams & Multi-Tenant SaaS
Designed for:

Startups
Growing SaaS teams
Agencies managing multiple projects

🧪 Why We’re Building This Now
The developer ecosystem has changed:

AI tools are exploding
Cloud bills are harder to predict
Teams are smaller but tool-heavy

But cost visibility hasn’t evolved with developer workflows.

👉 Cost control should be proactive, not reactive.
👨‍💻 Built by Developers, for Developers
SubTrackHub is built by developers who:

Run SaaS products
Deploy on AWS and GCP
Manage DevOps pipelines

Have personally overpaid for tools 😅

This isn’t a finance-first product.
It’s developer-first by design.

🛣️ What’s Next for SubTrackHub?
We’re actively working on:

Deeper cloud integrations
Smarter AI-driven insights
Team-level visibility improvements
Automation for unused resources
We’re also collecting early developer feedback to shape the roadmap.

🙌 Want to Try SubTrackHub?
If you’re:

A SaaS founder
A developer juggling multiple tools
A DevOps or cloud engineer

Someone tired of surprise invoices
👉 Check out SubTrackHub
👉 Share your feedback
👉 Tell us what subscription hurts the most

💬 Discussion
What’s one subscription you forgot you were paying for?

Drop it in the comments 👇
Let’s learn from each other.

[Boost]

Rahul singh Shekhawat — Thu, 15 May 2025 06:57:44 +0000

Rahul singh Shekhawat

May 15 '25

Adding EC2 Start/Stop to an Uptime Monitoring Tool – The Why & How

#news #webdev #devops #aws

Comments

2 min read

Adding EC2 Start/Stop to an Uptime Monitoring Tool – The Why & How

Rahul singh Shekhawat — Thu, 15 May 2025 05:38:50 +0000

Hey Devs 👋

As I was building PulseGuard, I realized something...

Most monitoring tools focus only on uptime — but cloud control matters too.

If I could monitor my server and start or stop the EC2 instance from the same dashboard, that would save time (and cost). So I built it.

Why Add EC2 Controls?

Avoid logging into the AWS Console every time you need to reboot
Turn off staging/dev servers when not in use
Save $$$ on idle EC2 time
React quickly when something goes wrong

⚙️ How It Works (Under the Hood)

Instead of using Lambda, PulseGuard uses IAM Assume Role to securely access your AWS account.

You create a role in your AWS account with permissions like ec2:StartInstances, StopInstances, etc.
That role is assumable by PulseGuard's AWS identity
When you press the start/stop button in the dashboard, PulseGuard assumes the role, executes the action, and returns the result

No permanent credentials are shared, and all activity is logged.

Bonus: It Integrates with Monitoring

You can monitor your server's uptime and control it — from the same dashboard.

It’s especially useful for short-lived dev or test environments.

Is It Secure?

Yes. Here's how:

You control the IAM role, permissions, and which instances are allowed
PulseGuard uses AssumeRole with external ID for added security
Only minimal EC2 permissions are needed (start/stop/reboot)
Your servers are never exposed directly

If you're tired of switching between monitoring tools and AWS just to reboot or shut down your EC2 instances, give PulseGuard a try.

Would love feedback, ideas, or just to connect if you're building something similar 🚀

aws #cloud #monitoring #ec2 #devops #buildinpublic #indiehacker #nodejs #pulseguard

I Built a Smart Uptime Monitoring Tool with Plans Starting at $0 — Meet PulseGuard

Rahul singh Shekhawat — Sun, 11 May 2025 10:32:13 +0000

Hey folks!

I'm Rahul, and after months of building, testing, and tweaking, I'm super proud to share PulseGuard with you — a smart, AI-powered uptime and performance monitoring tool built with simplicity, affordability, and developers in mind.

❓ Why I Built PulseGuard
As a cloud engineer and developer, I noticed a pattern:

Most uptime monitoring tools are either overpriced or overkill for indie hackers, startups, and small businesses.

Many charge $20+/mo just to monitor a few endpoints — and that's before you get any intelligent insights or SSL monitoring.

So I built PulseGuard to solve this:
Smart monitoring, clear alerts, and pricing that doesn’t hurt.

⚙️ What PulseGuard Does
Here’s what you get with PulseGuard:

Website & API Monitoring – Get notified the moment your website or endpoint goes down

SSL Monitoring – Never miss certificate expiry alerts

AI-Powered Alerts – Fewer false positives, smarter insights

Server Performance Metrics – Track server health and CPU/memory usage

Developer API – Automate and integrate PulseGuard with your tools

Modern Dashboard – Clean, fast, and distraction-free

AWS EC2 Control – Start/Stop your EC2 instances from the PulseGuard UI

Server Monitoring with Remote Control – Track server status and control basic actions

Explore these features via our Live Dashboard and Docs.

💸 Pricing Plans (Built for Real People): https://pulseguard.in/pricing

No credit card required

Full access even on the free plan

Transparent pricing, no surprises

🌍 Who Should Use PulseGuard?
Indie Hackers & Side Project Builders

Freelancers & Web Agencies

Startups on a Budget

DevOps & Cloud Engineers needing EC2 + monitoring in one place

Developers tired of bloated tools

🚀 Try It Now (and Let Me Know What You Think!)
You can be up and running in minutes:
👉 https://pulseguard.in

No lock-in

Smart alerts

EC2 control + monitoring in a single dashboard
🧵 Follow the Journey
I’m building this in public and sharing everything I learn — the wins, the bugs, and the lessons.

Let’s connect and grow together!

buildinpublic #uptime #monitoring #webdev #saas #indiehacker #cloudengineering #ec2 #aws #developer #pulseguard