From: aidotengineer
Munir, CEO and co-founder of Porsche AI, shared insights from his journey as a relatively recent founder of an AI startup [00:00:03]. Porsche AI focuses on building production agents for regulated industries through an open-source SDK [00:01:38]. The following outlines key challenges and observations from his experience, particularly for those looking to start their own AI company [00:00:28].
User Problems and Discovery
In AI, user problems and their discovery are starkly different from conventional product development [00:02:11].
Ill-defined Problems
Unlike established industries where user problems have a clear shape and adjacency to known issues (e.g., payment products at Stripe) [00:02:31], AI presents a scenario where users don’t fully know their problems [00:03:25]. User problems are often an “emergent property” of the space [00:03:33], and what is technologically possible changes very quickly [00:03:41].
Iterative Approach to Roadmaps
Traditional three-to-six-month product roadmaps are replaced by a need to make directional bets and iterate very fast [00:03:45]. Focus is maintained through “hypothesis-driven deliberate iteration” rather than conventional roadmaps [00:04:06].
Lack of Pre-defined User Journeys
Reassuring concepts like user storyboards and critical user journeys largely “go out the window” in AI development [00:04:35]. Founders must help users overcome initial inertia and work with them to create narratives and anchor on specific use cases [00:04:51]. This process gradually leads to more familiar territory and scoped problem spaces [00:05:17].
Velocity as Table Stakes in AI Production
The product development process in AI can be highly gratifying, allowing rapid movement from an idea to a released product within hours or days [00:06:01]. However, this velocity is not merely a “nice-to-have” but is “table stakes” in the AI industry [00:06:55].
Criticality of Speed
A startup cannot remain in the game if it cannot release quickly and effectively [00:07:02]. Opportunities to differentiate and gain advantage (e.g., breakthroughs like MCP or agent-to-agent protocols) appear and disappear rapidly [00:07:07]. Being able to react and build products that are at the forefront of these technological breakthroughs is crucial for brand growth and cementing one’s position in the developer or user community [00:07:31].
Outreach and Awareness
One of the toughest and most underestimated challenges in AI development is the minimal scaffolding available for outreach, awareness, traffic, and adoption when starting from scratch [00:08:08].
Navigating Hype and Noise
The AI space is characterized by “full-on hype and meme wars,” making it difficult to distinguish signal from noise [00:08:25]. Without the advantages of a larger brand, gaining traction feels like playing a game without any boosters or power-ups [00:08:36].
Building Advocacy
A key learning is that people primarily follow other people [00:09:04]. Unlike launching a product at a large company where a tweet from a prominent leader can generate instant awareness [00:09:09], a startup needs to find credible industry voices and advocates to speak on its behalf, using their platforms as force multipliers [00:09:21].
Strategic Partnerships
To build brand awareness and traffic without the broad channels of a larger company [00:09:33], startups must find allies [00:09:49]. These can be other startups at a similar stage or slightly ahead, interested in the same space, where synergies exist [00:09:51]. Strategies include:
- Partnering [00:10:00]
- Cross-marketing [00:10:00]
- Cross-selling [00:10:02]
- Getting products listed in the integration documentation of other companies (e.g., BrowserBase) [00:10:08], which can be a significant channel for driving traffic to one’s own properties (website, GitHub repo) [00:10:17].
This area represents a steep learning curve and a massive challenge for new AI ventures [00:10:45].