From: aidotengineer
Munir, CEO and co-founder of Porsche AI, shares insights from his journey of transitioning from big tech companies like Stripe, Google, and Amazon to founding an AI startup [00:00:09]. His experience is particularly useful for individuals looking to start their own companies in the current AI space [00:00:30]. Porsche AI, founded in June of the previous year with co-founder Emma, is an open-source SDK for building production agents with a focus on regulated industries [00:01:28].
User Problems and User Discovery in AI
User problem definition and discovery in the current AI landscape feel significantly different from traditional product roles in big tech [00:02:11].
The “Soulslike Game” Analogy
In conventional product development, user problems typically have a discernible shape and adjacency to familiar issues [00:02:31]. For example, a user at Stripe wanting card issuance for their balance is a problem anchored in existing payment products [00:02:41]. However, in an AI startup, especially when building from scratch, it feels like playing a “Soulslike game” for the first time [00:02:58]. Users don’t necessarily know what their problems are; instead, they sort of know they want to use AI and recognize an opportunity, but the specific user problems emerge as a property of the space itself [00:03:25]. The rapid pace of technological change further complicates this, as what is possible with AI is constantly evolving [00:03:39].
Instead of conventional three-to-six-month roadmaps, companies must make directional bets and rapidly iterate on them by releasing products and observing user reactions [00:03:45]. The primary anchor for the team becomes a hypothesis-driven, deliberate iteration process, providing necessary focus in a rapidly changing environment [00:04:00].
The “No Man’s Sky” Analogy
Traditional user storyboards and critical user journey definitions become largely irrelevant in this context [00:04:24]. This environment is akin to starting “No Man’s Sky,” where there’s no clear story arc or mission [00:04:39]. Instead, the product team must create their own narrative [00:04:51].
A crucial aspect of AI Accelerated Need Finding Process is helping users overcome their initial inertia when trying AI for specific use cases [00:04:58]. This involves working collaboratively with users to create these narratives and help them anchor on a specific use case [00:05:12]. Once a use case is established, it becomes more familiar territory, allowing for iteration and refinement of the product against defined problems and challenges [00:05:17]. This stage is like “colonizing the planet in No Man’s Sky,” where actual problems become clearer and the problem space is more scoped [00:05:33].
Rapid Product Development in AI
The product development process in AI is highly gratifying due to its speed [00:06:01]. Unlike the lengthy sign-off processes in big tech, an AI startup can go from an idea to a released product (including spec, design, testing) within hours or a few days [00:06:21].
However, velocity is a fundamental requirement, not a bonus, in AI [00:06:52]. Opportunities that allow products to stand out, like new standards (e.g., MCP), agent-to-agent protocols, or diffusion models, emerge and disappear rapidly [00:07:07]. Being able to react quickly and build products at the forefront of these technological breakthroughs is essential for brand growth and cementing presence in the developer or user community [00:07:31].
Challenges in Outreach and Awareness
Gaining outreach, awareness, traffic, and adoption presents a significant challenge for new AI startups [00:08:11]. In the current AI space, which is characterized by hype and “meme wars,” it’s difficult to distinguish signal from noise [00:08:25]. Without the brand advantages of a larger company, it feels like playing Crash Bandicoot or Mario Kart without any boosters or power-ups [00:08:32].
Key learnings for gaining visibility include:
- People follow people: Unlike big tech, where a tweet from a founder can launch a product, startups need to find credible advocates and industry voices to act as force multipliers [00:09:04].
- Finding allies: To compensate for the lack of brand association and established channels, startups should find allies [00:09:33]. These can be other startups at a similar stage or slightly ahead, interested in synergistic partnerships for cross-marketing and cross-selling [00:09:51]. Getting a product listed in the integration documentation of a larger company (like Browserbase) can also serve as a significant channel for driving traffic to a website or GitHub repository [00:10:04].
This area has been a steep learning curve and a massive challenge, but also a significant source of learning [00:09:56].