Recently I caught up with Tom Rudczynski, one of my former coworkers from my digital ad agency days at Huge. Tom has a long history in ecommerce, and has been imagining the future of ecommerce with AI through his company Prompt Engineering, and its primary product, XTAL Search. We've talked several times about AI prototyping, and in this conversation, I wanted to focus on how AI prototyping has shifted in the last 7-9 months in how he pitches and builds product in his business. We also get into the impact AI will have on junior roles, and even the impact on the future of the ad agency model.
For his company Prompt Engineering, as recently as 7 months ago, they would go into sales meetings with a pitch deck to explain what their product can do, but in the last few months, they've become super users of Bolt (bolt.new) so they can show, not just tell.
"As CEO and lead pitcher, being able to go in and have a bunch of ideas and creative ways of applying some of the Gen AI technology that we want, or that we see as a good fit for the customer and to actually show it."
Tom and I remembered the days when it would be "two weeks just to have a design that we could put together for a prototype." For a product company, that's great news, but for agencies, it's a double-edged sword. While it saves them time on the initial pitch to build an interactive prototype, the agency model is based on billable hours. And while AI can't build pixel-perfect designs on its own as of yet, the process of getting to those designs and a final product is shorter because AI speeds up the time it takes to validate initial assumptions and direction. Overall, this will likely mean fewer overall billable hours for agencies. It could also mean the "collapse of the talent stack" (I believe I first heard this from Clair Vo, CPO of LaunchDarkly) within the agency world, especially if product people within agencies also become responsible for an account management role that sells in work by working directly with the client to pitch solutions in prototype form vs having an intermediary to interpret and manage the relationship.
While we talked a great deal about the potential of AI prototyping, we also talked about some of the drawbacks we've faced with AI prototyping tools in their infancy. Tom expressed my sentiments well that "you can ruin something that's been working well by trying to change a small element just on the UI." I've experienced this several times, and it's incredibly frustrating. Many of the AI prototyping tools have ways to roll things back now, but it's not always perfect, and I'm looking forward to when this is a bit smoother.
But what Tom has been impressed by is how easy it is to make something quickly with the look and feel of a specific potential customer to show them how the idea works.
"I was not only able to come up with a few ideas, I was able to show them live how content on their site would be changed using just Bolt and just prompts with a live integration with open AI."
There are a lot of dimensions that AI prototyping is changing how we work, and selling and pitching are only one piece of the larger puzzle. It levels the playing field in some ways in terms of the level of functionality you can show potential customers early, but it raises the bar on being able to demonstrate your ability to understand that customer and what they actually need. It will be amazing to see where things are even 6 months from now.
Stay curious.
P.S.: I’m a fractional product leader and consultant. If your business is building its first AI features, I have an AI audit offering to show businesses where their best (and worst) are opportunities to implement AI. You can reach out here.
P.S.S.: I recently launched Building Successful Product with AI, a workshop for product teams to learn how to leverage AI in their processes to 10x themselves and focus on the highest leverage work. It’s only available for companies right now, and I’m teaching the first one in April. If you want to accelerate your team while teaching them not to give away their thinking, reach out here.
Share this post