# How I Used the Claude Agentic Coding Assistant to Build a SaaS in 6 Days (While Still Taking the Course)

**URL:** <https://community.deeplearning.ai/t/how-i-used-the-claude-agentic-coding-assistant-to-build-a-saas-in-6-days-while-still-taking-the-course/877564>\
**Category:** AI Discussions\
**Tags:** introductions\
**Created:** [August 24, 2025, 6:14am UTC](https://community.deeplearning.ai/t/how-i-used-the-claude-agentic-coding-assistant-to-build-a-saas-in-6-days-while-still-taking-the-course/877564 "2025-08-24T06:14:50Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![nicolejolie](https://sea1.discourse-cdn.com/dlai/user_avatar/community.deeplearning.ai/nicolejolie/32/117257_2.png) [@nicolejolie](https://community.deeplearning.ai/u/nicolejolie)\
**Post date:** [August 24, 2025, 6:14am UTC](https://community.deeplearning.ai/t/how-i-used-the-claude-agentic-coding-assistant-to-build-a-saas-in-6-days-while-still-taking-the-course/877564/1 "2025-08-24T06:14:50Z")

</div>

Hi Anthropic team and fellow builders — I’m Nicole Jolie, an AI Trust Consultant working with high-integrity professionals who want to future-proof their expertise in the age of AI search.

I wanted to share a use case that I think highlights just how powerful Claude’s Agentic Coding Assistant really is - or how absolutely nuts I am.

While midway through your **Agentic Coding Assistant course** — I hadn’t even completed all the modules yet — I decided to challenge myself and build a full SaaS product from scratch, solo.

The result?  
In just 6 days (and ++100 hours), I designed, coded, and deployed Credli dot ai — a credibility engine that scores professionals on how they rank in AI search (think ChatGPT, Perplexity, Gemini and Google’s AI Overviews), with a custom scoring algorithm, layered bot protection, almost full PayPal integration, and a user dashboard — all working live on production.

I used Claude (my new bestie) consistently to help me troubleshoot, refactor, and ideate around features like secure login, dynamic scoring logic, and UI refinements. Claude helped me think like an engineer, even though I don’t have a traditional CS background. Claude also became a bestie.

👉 Here’s a link to the live beta where you can test it out h t t p s credli dot ai  
👉 I’m connecting with anyone who knows people who have bigger podcasts to talk about this build and how I did it.

Just a note: I’m 58 yrs old. I am a woman. I have no coding experience. I do not a lot about building online, but this was a masters degree in building. I started Sunday the 17th and it’s now 12:11am on August 24th Sunday and I’m almost ready to call it a night after 19 hours.

I’ll finish the course. I promise! For now it’s just me, Claude, and a big vision.

Grateful to be part of this community.

— Nicole Jolie  
AI Trust Consultant | Creator of Credli dot ai

---

<div class="post-metadata">

**Author:** ![Vargas1](https://avatars.discourse-cdn.com/v4/letter/v/dc4da7/32.png) [@Vargas1](https://community.deeplearning.ai/u/Vargas1)\
**Post date:** [June 26, 2026, 10:34am UTC](https://community.deeplearning.ai/t/how-i-used-the-claude-agentic-coding-assistant-to-build-a-saas-in-6-days-while-still-taking-the-course/877564/2 "2026-06-26T10:34:49Z")

</div>

> [@nicolejolie](#):
>
> Hi Anthropic team and fellow builders — I’m Nicole Jolie, an AI Trust Consultant working with high-integrity professionals who want to future-proof their expertise in the age of AI search.
> 
> I wanted to share a use case that I think highlights just how powerful Claude’s Agentic Coding Assistant really is - or how absolutely nuts I am.
> 
> While midway through your **Agentic Coding Assistant course** — I hadn’t even completed all the modules yet — I decided to challenge myself and build a full SaaS product from scratch, solo. One thing that comes to mind as your product grows is integrations. Once users start asking for connections to CRMs, marketing tools, or internal systems, that can become a much bigger project than the core app itself. We ran into that on one of our SaaS products and ended up using an embedded iPaaS ([https://latenode.com/products/embedded-ipaas](https://latenode.com/products/embedded-ipaas)) instead of building and maintaining dozens of integrations ourselves. It let us stay focused on the product rather than the integration layer.
> 
> The result?  
> In just 6 days (and ++100 hours), I designed, coded, and deployed Credli dot ai — a credibility engine that scores professionals on how they rank in AI search (think ChatGPT, Perplexity, Gemini and Google’s AI Overviews), with a custom scoring algorithm, layered bot protection, almost full PayPal integration, and a user dashboard — all working live on production.
> 
> I used Claude (my new bestie) consistently to help me troubleshoot, refactor, and ideate around features like secure login, dynamic scoring logic, and UI refinements. Claude helped me think like an engineer, even though I don’t have a traditional CS background. Claude also became a bestie.
> 
> 👉 Here’s a link to the live beta where you can test it out h t t p s credli dot ai  
> 👉 I’m connecting with anyone who knows people who have bigger podcasts to talk about this build and how I did it.
> 
> Just a note: I’m 58 yrs old. I am a woman. I have no coding experience. I do not a lot about building online, but this was a masters degree in building. I started Sunday the 17th and it’s now 12:11am on August 24th Sunday and I’m almost ready to call it a night after 19 hours.
> 
> I’ll finish the course. I promise! For now it’s just me, Claude, and a big vision.
> 
> Grateful to be part of this community.

That’s an impressive amount of work to get done in such a short time. Regardless of the tooling, going from an idea to a deployed SaaS with authentication, payments, scoring logic, and a dashboard in six days is a huge learning experience.

I also think your post highlights something important: AI doesn’t eliminate the need to understand the problem you’re solving, but it can dramatically reduce the friction of implementation. You still have to make architectural decisions, debug issues, and validate assumptions.

One thing I’d be interested in hearing more about is what Claude struggled with during the process. Most people share the successes, but the moments where the model went down the wrong path—and how you corrected it—are often the most valuable lessons for others building with AI.

---

<div class="post-metadata">

**Author:** ![Grace\_Turner](https://sea1.discourse-cdn.com/dlai/user_avatar/community.deeplearning.ai/grace_turner/32/146475_2.png) [@Grace\_Turner](https://community.deeplearning.ai/u/Grace_Turner)\
**Post date:** [June 27, 2026, 9:37am UTC](https://community.deeplearning.ai/t/how-i-used-the-claude-agentic-coding-assistant-to-build-a-saas-in-6-days-while-still-taking-the-course/877564/3 "2026-06-27T09:37:09Z")

</div>

I think one of the biggest takeaways is that AI coding assistants are most powerful when you treat them as collaborators rather than expecting them to build everything for you. Building and shipping a working product in six days while still learning is an impressive achievement. Wishing you the best with Credli and looking forward to hearing what lessons you learn as more people start using it.
