Will the real AI agent please stand up?

Will the real AI agent please stand up?

The hype

I was on a call with a client 4 months ago. The client was a large, multi-billion-dollar IT services company that has 100+ recruitment vendors.

The recruitment lead started the call by saying the only reason she was talking to us was that we claimed to have a GenAI product. She said she really didn’t care about the value we brought unless we were using GenAI. So, before we start, can you confirm this?

My guess is that using GenAI was a mandate passed down by someone. This was the theme of most conversations a few months ago. But I think that period is over.

Everyone is now realizing that AI is just hype unless you prove real business value.

Companies built on the hype will now face ground realities. GenAI is cool, but what problem does it solve for customers?

From hype to value

We're well and truly past the "wow" phase of GenAI and into the phase where we need to solve real customer problems and prove why our AI product is better than what they're doing today. Not only better, but much, much better, so we can overcome the inertia that exists in enterprise companies.

Customers are no longer asking if we tick the GenAI checkbox but instead, what metrics we have to prove our AI solution works. What tangible results can we show? How can we ensure usability and consistency at scale?

Hopefully, these are questions every builder asked themselves before boarding the GenAI gravy train.

It's not all or nothing with AI

We have been building agentic AI products, and our learnings stem from that. Very quickly, agentic AI refers to AI that can take actions, solve problems, and achieve goals with minimal human input.

From the outset, our intention was to build agentic AI that can:

  • Conduct live code pairing interviews autonomously.
  • Interact naturally with candidates, adapting to their responses.
  • Provide informed, unbiased insights (backed by proof) into candidate performance, enabling humans to make the right decisions.

Over the last few months, we've learned that building fully autonomous agentic AI products capable of holding 60 to 90-minute conversations with humans is incredibly hard. Not impossible, but hard.

We have to chip away slowly at the hundreds of issues that arise. Not all these issues are due to AI. Stitching together these agents requires regular software development craftsmanship and AI is just one piece among many that need to come together.

Until we achieve full autonomy for the agent, we are providing human support by having an expert in the loop. This expert is available on demand to address any issues that may arise during the interview.

What matters is that we automate processes reliably, with explainability and great UX built in. No customer cares whether you are using AI alone to solve their problem. Unless we work closely with customers, we are in danger of not solving real customer problems. Building in isolation with GenAI is a hole you cannot dig yourself out of.

Showing value sells

This is one of the things we need to remind ourselves as we build with AI. I’ve seen quite a few funded startups building GenAI products that made no business sense to me. It seemed like if you picked a hot space, plugged in GenAI, and added a founder from a renowned college, you were in hot demand.

As much as we initially focused on selling our product's AI features, our sales became easier when we showed the outcomes our product could achieve.

And we showed this in tangible, real-world terms.

  • A senior developer who could run 1 interview in 60 minutes could now read interview reports, watch videos, and make decisions on 10 candidates in the same amount of time.
  • This way, they could review the outcomes (reports + videos) of our AI agents and decide whether to move the candidate directly to Round 2 or not.
  • AI outcomes + expert in the loop = immediate value for our clients

We saw that when we started showing interview reports to our clients, they wanted more. This was the affirmation we needed to know we’re on the right track.

GenAI needs to earn its place in the real world. Customers want value, not AI. So, we as GenAI app builders need to make sure the real AI agents stand up.

If you're interested to know what we're building, check our AI product video - https://www.youtube.com/watch?v=KIwKcALoF54

To know more check out -> https://www.geektrust.com/assemble








Er. Shikhar Rathore

Founder and CEO of The Unifed | Former Project Engineer at RTNS Technology Pvt Ltd | Former C.S. Associate at Tech Mahindra Limited

3 个月

Third party or business companies want commission but AI doesn't. As per 2024, yeah the statement is correct that AI is not as proficient as the Carnot engine's efficiency, but the time will be soon. Humanity is depreciating day by day due to the business nature of a few companies for decades. I hope that the arrival of AI can disrupt the third party commission and the re-establishment of humanity.

回复

要查看或添加评论,请登录

Krishnan Nair的更多文章

社区洞察