Innovative Strategies for CyberAI Startups

This week I got to share my views on building a future-proof GenAI startup at the "One Year of ChatGPT" event hosted at the Microsoft SF office with Latent Space and Decibel Partners. Shawn W shared an amazing summary of updates since the launch of ChatGPT. After the OpenAI DevDay event, I saw a lot of blogs/ threads speaking about some startups going irrelevant. In the light of it, I wanted to share some tips that will make your startup resilient. Here are the things I shared at the event: 1️⃣ Problem First: Align your AI product to solve a real business challenge, not just to showcase technology. 2️⃣ Find Your Niche: Carve out a vertical niche where you can become the go-to solution. 3️⃣ Data is King for GenAI: A unique dataset can be your startup’s strongest unique selling point (USP). 4️⃣ User-Friendly: Ensure your product is intuitive - ease of use can drive adoption rates sky-high. 5️⃣ Regulatory Alignment: Keep abreast of regulations and shape your product to comply seamlessly. (Read EU AI Bill and US Executive AI Order) 6️⃣ Partnerships: Forge win-win partnerships with companies that can amplify the value of your product. Being in collaboration with tech-giants will give you competitive advantage. 7️⃣ Extended Toolkit: In enterprise products, integrate model monitoring features, model evaluation metrics (prescriptive), RLHF capabilities etc. to make your product more of an ecosystem. Did I miss something that you would like to add? #StartupLife #AI #GrowthHacking #TechTips

Very few books lead to big epiphanies.  This one by Ross Haleliuk did. As he explains through the book how difficult it is to create a new product in the cybersecurity space, it occurred to me that- Not only have we at Levo.ai successfully launched a new product in this market, but we have also introduced a brand new approach towards API security. Before Levo.ai, security solutions focused on detecting attacks in production and generating alerts. While it has its utility, this approach does not address the root cause and generates additional work in triage. To me, this seemed almost crazy. Like consistently reacting to the symptom of the problem rather than resolving the root cause. The root cause is in the code. Until those API security issues are found and fixed, there is no guarantee that they cannot be exploited. And, a lack of documentation and continuous pen-testing made it impossible to trace and rectify them. Digging deeper, I couldn’t imagine any other way than a proactive approach- automated testing of applications and APIs while they were being built or deployed to dev/staging. Sounds cool, right? But I had forgotten that novel ideas always sound cool in theory but rarely catch on. As most people are okay with the status quo, pushing something new, even if it's better, will invite a lot of resistance. And so did I. In the first few months of our approaching enterprises, many doubted, questioned, and turned down our approach. While the skepticism wasn’t pleasant, I knew that the approach had promise in the long term, and thus we kept pitching. Slowly but surely, more and more enterprises began showing conviction in us and our solution, to the extent that we, a team of 10, won contracts over billion-dollar competitors. So, if you too are building a solution with a contrarian approach, go ahead and do so. But remember that you will face a lot of rejection and doubt. But if your solution has metal (10x better than the previous one), it will take on! So hang in there and keep at it. #tech #cybersecurity #product #startup #entrepreneurship

Don't Build An "AI Co-Pilot" Because They Favor Incumbents Miles Grimshaw of Benchmark recently dropped some invaluable insights on the 20VC podcast, emphasizing that AI Co-Pilots naturally align with incumbent strategies. "Incumbents own distribution, the UX, the data, and the business model." It's a sentiment we've come to appreciate firsthand at Alphawatch.AI. By his definition, Co-Pilots are generally "in-line, assistive, suggestive" AI tools. Using this characterization, it makes a lot of sense that incumbents would have the upper hand here. It is significantly easier for incumbents to add a Co-Pilot to their existing solution than for a startup to go and acquire the incumbent's customers. Actually, it is one specific variation of what I wrote last week on "Can Startups Compete with Incumbents?" Except that incumbents are actually innovating quite quickly on Co-Pilots ... just look at the number of big companies who have already announced their integration into existing products. The real opportunity for startups lies in -- to use Miles' words -- "orthogonal innovation." This goes beyond merely developing AI-enabled processes; it extends to creating entirely new business models that are fundamentally different from existing ones.** By doing so, startups can neutralize or even eliminate the advantages that incumbents typically enjoy. Thanks to insights from our customers, we've identified several such opportunities for orthogonal innovation. There's no shortage of these avenues to explore, and I look forward to seeing what others in the industry will discover and scale. **Benchmark partners seem to be fond of the concept of "selling the work" instead of "selling the software," which is also appealing to us and something that I can perhaps share more about in the future.