Experimenting With a Conversational AI Layer in My Portfolio

I recently tried a small experiment on my portfolio called:

“Talk with Akshay.”

And before anyone imagines some futuristic AI portfolio revolution — no, I don’t think people normally open a portfolio and immediately think:

“Let me have a conversation with this chatbot.”

That would be a little too much.

So this wasn’t designed as a serious replacement for case studies, navigation, or a traditional portfolio experience.

It was mostly an experiment. A playful one.

I just wanted to explore what an AI-assisted portfolio interaction could feel like.

The Idea

Most portfolios work the same way.

You open them. Scroll through projects. Skim a few sections. Maybe open one case study. Then leave.

That interaction model has worked for years, and honestly, it still works perfectly fine.

But while working on my portfolio, I kept thinking about one small thing:

What if someone doesn’t want to read everything?

What if they only want one answer?

Like:

• “What was your role in this project?”
• “What problem were you solving?”
• “Why did you make this design decision?”
• “What tools do you use?”
• “Tell me about your AI experiments.”
• “What kind of designer are you?”

Normally, the answers already exist somewhere inside the portfolio. But visitors still have to search for them manually.

So I wondered:

What if the portfolio could simply respond?

Not intelligently in some magical sci-fi way. Just conversationally.

How It Works

The interaction flow is actually pretty simple:

1. The visitor speaks into the portfolio
2. Audio gets converted into text
3. AI processes the question using my portfolio content as context
4. The response gets converted back into speech
5. A side drawer updates with relevant visual information

The interesting part is that I also experimented with voice cloning.

So instead of using a generic AI voice, the responses come back sounding like me.

Which honestly felt a little strange the first time I tried it.

But also very exciting.

Because suddenly the portfolio stopped feeling static.

It started feeling responsive.

Why the Visual Drawer Mattered

One thing I realized very quickly: voice alone is limiting.

People don’t always want to sit and listen to a long response.

Sometimes they want to skim. Sometimes they want to scan visually. Sometimes they just want to jump directly to the relevant project.

So while the assistant responds through voice, the interface also updates a right-side drawer with contextual information:

• Relevant projects
• Skills
• Experience
• AI experiments
• Related sections from the portfolio

That combination made the experience feel much more usable.

The voice creates interaction. The visuals create clarity.

The Important Part: This Is Not Me Saying “Every Portfolio Needs AI”

I think this is where a lot of AI experiments go wrong.

People add AI just because they can.

Suddenly every product has:

• a chatbot
• a copilot
• an assistant
• an AI button floating somewhere

Even when it doesn’t actually improve the experience.

So I want to be very clear:

I don’t think portfolios suddenly need conversational AI.

I don’t think recruiters are waiting to interview a chatbot version of me.

And I definitely don’t think this replaces strong case studies or good storytelling.

This experiment was never about replacing the fundamentals.

It was more about exploring interaction design in a different format.

A portfolio is usually passive.

This experiment explored: What happens when it becomes slightly conversational?

Even if only for a moment.

What I Learned

The most interesting part wasn’t the AI itself.

It was how different the portfolio felt once interaction became dynamic.

A few things stood out:

1. Voice changes the emotional feel of an interface

Reading and listening create very different experiences.

Even simple responses felt more personal when spoken aloud.

2. Conversational interfaces need visual support

Without the visual drawer, the experience felt slow and frustrating.

The UI still needs to help people navigate quickly.

3. AI interactions work better when scoped tightly

The assistant only answers using my portfolio context.

That limitation actually made the experience feel more focused and useful.

4. Experimental interfaces reveal new design questions

Not necessarily solutions. But definitely new questions.

And honestly, that’s what made this fun.

Final Thoughts

The feature is still rough. Still experimental. And probably more “interesting interaction layer” than genuinely useful product feature right now.

But I enjoyed building it.

It pushed me into exploring:

• voice interfaces
• AI response design
• conversational interactions
• contextual UI updates
• voice cloning workflows

And more importantly, it made me think differently about what a portfolio can feel like.

Not smarter. Not more “AI-powered.” Just slightly more responsive.

And sometimes, small experiments like that are worth trying purely because they make you curious.