From: aidotengineer
Munir, CEO and co-founder of Porsche AI, shares his journey and insights from transitioning from various product roles at large tech companies like Stripe, Google, and Amazon to founding his own AI startup [00:00:09]. This presentation aims to provide useful notes for those looking to make a similar leap or who are in a similar early-stage AI startup phase [00:00:28].
Munir’s Background
Munir has a background in big tech, always based in London [00:00:49]. His previous roles include:
- Launching Amazon’s deals program in the UK, used for events like Black Friday and Prime Day [00:00:56].
- Launching and growing Google Pay in 30 markets across EMEA [00:01:02].
- Leading Bank as a Service at Stripe for EMEA [00:01:06].
He always aspired to start his own venture, reaching a point where “nothing else made sense anymore” [00:01:11].
Porsche AI
Porsche AI was founded in June of the previous year (approximately a year prior to the presentation) with co-founder Emma, whom Munir met at Stripe [00:01:28]. It is an open-source SDK for building production agents, with a particular focus on regulated industries [00:01:38].
Lessons Learned in an AI Startup
Munir highlights several key lessons and observations from his experience:
1. User Problems and Discovery
The process of identifying user problems and conducting user discovery in an AI startup is “freaking different” from big tech [00:02:11].
- Emergent Problems: Unlike established product areas where problems have shape and adjacency to known issues (e.g., payment products at Stripe) [00:02:31], in AI, user problems are often an emergent property of the space and not yet defined [00:03:33].
- Rapidly Changing Technology: What is possible with AI technology changes very quickly [00:03:41].
- Hypothesis-Driven Iteration: Traditional three or six-month roadmaps are less effective [00:03:45]. Instead, it’s necessary to make directional bets and iterate very fast by putting products out and observing user reactions [00:03:52]. The focus for the team becomes hypothesis-driven, deliberate iteration [00:04:06].
Soulslike Game Analogy
Understanding user problems in AI feels like the first time playing a “Soulslike game” such as Nioh, where configuring a character is challenging because “nothing makes sense at first” [00:03:00]. Users know they want to use AI and see opportunity, but they don’t know their specific problems [00:03:25].
No Man's Sky Analogy
Reassuring concepts like user storyboards and critical user journeys “go out the window” [00:04:21]. It’s like starting No Man’s Sky without a clear story arc or mission [00:04:41]. Founders must stretch the muscle of helping users overcome inertia, creating narratives, and anchoring on a use case [00:04:54]. This gradually leads to more familiar territory of iterating on a defined use case, akin to colonizing a planet in No Man’s Sky where problems become tangible [00:05:33].
2. Product Development Velocity
The product development process in an AI startup is described as “gratifying” [00:06:01].
- Rapid Iteration: Unlike big tech where product launches require numerous departmental sign-offs, in a startup, one can go from an idea to a spec, design, testing, and release within hours or a few days [00:06:06]. This speed from concept to code and deployment is “exhilarating” [00:06:42].
- Velocity as “Table Stakes”: In AI, velocity is not a “nice to have” but a fundamental requirement; inability to release quickly means being “out of the game” [00:06:52]. Opportunities for differentiation and gaining “power ups” emerge and disappear very rapidly [00:07:07], such as new protocols (MCP, agent-to-agent) or models (diffusion models) [00:07:20]. Reacting quickly to these breakthroughs is crucial for brand growth and cementing oneself in the community [00:07:31].
Mario Analogy
Swift opportunities that provide an advantage are compared to “power ups” in Mario [00:07:12].
3. Outreach, Awareness, and Adoption Challenges
This aspect has been the toughest and steepest learning curve [00:07:59].
- Lack of Scaffolding: Startups lack the built-in scaffolding for outreach, awareness, traffic, and adoption that big brands provide [00:08:11]. The AI space is saturated with “hype and meme wars,” making it hard to distinguish signal from noise [00:08:25].
Crash Bandicoot / Mario Kart Analogy
This experience is akin to playing Crash Bandicoot without boosters, or Mario Kart without red and green turtles to fire [00:08:41].
- Importance of Advocates: Unlike a big tech company where a tweet from a leader can generate instant awareness [00:09:09], a startup must find advocates and credible industry voices to speak on its behalf and act as “force multipliers” [00:09:21].
- Finding Allies: To build brand awareness and traffic without a large company’s established channels, startups must find allies [00:09:33]. These can be other startups at a similar stage or slightly ahead, who see synergy in the space [00:09:51]. Strategies include partnering, cross-marketing, cross-selling, and even getting products listed in the integration documentation of other companies (e.g., BrowserBase) to drive traffic to their own properties like websites and GitHub repos [00:09:58].
Munir acknowledges that this area has been challenging, but also a massive learning curve, which he embraces as part of his personal growth [00:10:24].
Munir is still learning and does not claim to be an expert [00:11:18]. He encourages viewers to reach out on LinkedIn for discussions and to consider starring Porsche AI’s GitHub repository [00:11:00].