"It's Not the AI, It's You": Hot Takes from an AI Product Builder

"AI Product Builder" is a job title that didn't exist a couple of years ago. Grace Conard has held it at Biscred since June.

We sat down with her to find out what the role actually is, what she's shipped so far, and why she has to bite her tongue when people tell her AI isn't very good yet.

Here are five takes from someone whose entire job is figuring out what AI can — and shouldn't — do.

1. "It's an evolution of product management, where AI has supercharged the ability to experiment."

Start with the obvious question: what is this job?

"I spend about half my time talking to people and digging through data to deeply understand problems, and the other half experimenting — thinking up, building, and testing solutions."

Sometimes that's a product for customers. Sometimes it's a workflow automation for an internal team. Sometimes it's making people faster at what they already do; sometimes it's reshaping the experience entirely.

"All in all, the role spans product, design, development, strategy, and operations."

Which is a polite way of saying the job description is still being written.

2. "I built each in less than 30 minutes."

Two things Grace shipped in her first weeks: an internal dashboard so our CSMs can monitor their accounts, and a personalized networking guide that tells customers who to meet at a Bisnow event and why.

Both took under half an hour. She's aware of how that sounds.

"When I share that time, I get one of two reactions: some marvel at the magic, and others are skeptical of the slop. Both are completely valid."

Here's the caveat she insists on:

"An impressive AI build time is heavily caveated. Whether upfront or in iteration, you can't skip the critical thinking. The build is only as good as its adoption."

It can take hours of conversation to reach the clarity that lets AI nail the solution in one go. The 30 minutes is the visible part of a much longer process.

3. "I purposefully build before I fully understand what I'm building."

That is, traditionally, the wrong way to work. Grace does it on purpose.

"It's the new form of prototyping. So long as I know who I'm building for, I can use prototypes to communicate. It's not going to be perfect, but it's going to quickly give me something tangible to put in front of someone to discover the nuances of the problem."

Speed and feedback first; the engineering team hardens what survives. When building is cheap enough, a rough prototype becomes a better research tool than another discovery call.

4. "It's not the AI, it's you."

This is the one she usually keeps to herself.

"I often hold my tongue when people say 'AI's not very good, I've tried it, but it's just not there yet.'"

When someone tells her the output isn't good enough, she asks them to walk her through their process — and the culprit is almost always the same:

"One well-intentioned practice I see is hyperfixating on the prompt. 'If I just ask it perfectly, I'll get exactly what I want' is a frequent misconception."

We hear "garbage in, garbage out" and assume the input is the prompt. It isn't. It's also the AI's memory of you, the tools and data it's connected to, the conversation history, the creativity and effort settings, and the model you picked.

Her fixes: connect the AI to your actual tools and data, use one AI to prompt another, start fresh conversations for new topics.

"And the biggest unlock is working in your terminal. Tools like Claude Code or Codex can feel overwhelming or unnecessary, but they combine AI with reliable, rule-based automation — which is how you get consistent quality output."

5. "Don't spend time scrutinizing AI's limitations. Tomorrow they may well be a moot point."

We asked what AI is genuinely bad at. She declined to answer, and the reason is a take of its own.

"I've learned not to say AI is definitively bad at something, because it's getting better every single day."

Understand how it works behind the scenes, sure. But building a mental list of things AI can't do is a list with a short shelf life.

"Keep an open mind, spend time experimenting, and always keep learning."

The Human Part

One concern Grace won't wave off: the fear that leaning into AI means losing the human relationships CRE runs on.

"I share that fear, and I think about it often, working in such a relationship-driven industry."

Her reframe is the useful part:

"Embracing AI isn't about the sheer volume of tasks we can offload to it, but rather asking, 'How might we optimize for the most valuable human contributions?' If we can identify those core interactions, we can redesign everything else to support them."

Internally, that means enabling each person on the team to do what they do best. Externally, it means helping customers get more out of our data and our people.

The Bottom Line

Ask Grace what drew her to Biscred and she doesn't say the technology.

"The people. The Biscred team is incredibly dedicated to helping our customers succeed, and helping each other succeed. My role is all about building to enable others, so I become embedded in that mindset of shared success."

What she's most excited to build next is that same mindset, pointed outward:

"I'm excited for Biscred to not only help you grab contacts, but partner with you to help you land deals. What if we built Biscred just for you — what would it look like? What would it do? AI is enabling that individual level of support, and I'm excited to bring that to our customers."

Curious what that looks like for your team? Set up a demo.