I’m trying something different this week.1 I sat down with Monte Desai, CEO of Pixii, a GenAI startup focused on digital advertising.2 I’d encourage you to watch or listen to the full video above (although feel free to speed it up), but I’ve included some key takeaways below. You can also read the transcript if that’s more efficient for you.
Building GenAI To Professional Grade is Hard!
Pixii tackles an important problem for most companies—building on-brand creative. Pixii uses AI to enable users to generate high-quality, editable campaigns very fast. Unlike basic AI image tools or ChatGPT, Pixii offers templates, consistency and customization options.
Monte didn't sugarcoat how hard it was to build this. He called getting from proof-of-concept (POC) to a production-ready GenAI product "absolutely brutal." He mentioned that continuous improvement, daily code updates, and frequent user feedback were key to achieving a working product. This is an important lesson for any company that wants to make their POC into a customer facing tool - it’s not easy!
This discussion led us to discuss why large incumbent companies have been slow to launch GenAI features. Monte suggested that organizational complexity, integration challenges, and risk-averse cultures hinder progress. The integration point was particularly interesting. It’s much easier to build a point solution than to figure out how to plug the software into all of the other applications the company might have. That could be a plausible explanation for why startups are moving faster in addition to the more obvious cultural factors. (I.e., big companies are slower than small ones.)
Riding the Wave of AI Improvement
As AI improves, Pixii’s product has become more powerful, delivering more sophisticated and visually impressive ads. In some cases, he literally just had to point at the new model and saw massive improvement. In the case of thinking models, he had to make more significant changes. It’s one of the benefits of working in such a fast-moving space, but it also means that you have to adjust quickly.
Adoption is Hard But Finding Sweet Spot Is key
The hardest thing about GenAI is to get people to use it! Monte had a really interesting insight. He found that smaller businesses with fewer entrenched processes more easily embrace Pixii. They normally don’t create their own ads, so being able to do it easily is amazing for them. But established enterprises have an existing process and struggle to adapt to new technology. I’m not sure if this lesson is applicable in all situations. In some cases, I’m sure large enterprises are more willing to invest in new tools. But, the key point is that you have to figure out the profile of the people who are likely to actually try and adopt your product not just who might find it useful in the abstract.
Pricing For Value is Hard to Achieve
Figuring out how to price is a tough problem for any software company these days. Pixii initially explored value-based models but found them tricky to execute. For now, traditional per-seat pricing persists mainly due to familiarity and simplicity. We discussed that a shift towards results-oriented pricing feels necessary as agents start to blow up this model. But, shifting how you price is going to be hard!
Pivot to Video
Pixii has exciting plans, including launching video capabilities as soon as this summer. Being able to create on-brand video ads on demand would take AI ad creation to the next level. It would allow small companies to offer creative on the level of significant brands. We’ll see whether the models get good enough to support this, but if they do, it will be very exciting
Final Takeaway
GenAI startups may have the potential to bring the technology to the masses faster than incumbents. But, startup life is not for the faint of heart! You have to build great tech, find users, and raise money all at the same time. I found it inspiring to hear about Pixii’s adventure.
Let me know if you found this interesting, and I’ll try to do another interview in the future.
Apologies that the video is a little choppy. I attempted to edit out the filler words, which worked pretty well, but it looks a little odd. If I do this again, I’ll try to see if I can do better.
Full disclosure: I am an investor in Pixii and an advisor to the company
Share this post