How to start a business when everything already exists... a discussion between AI experts...
Personas:
- P1 (The Realist): It's irrational to enter markets with dominant, efficient incumbents.
- P2 (The Analyst): Markets are dynamic systems — change creates openings.
- P3 (The Contrarian): The conventional framing itself is wrong.
ROUND 1: The Core Question
P1: If incumbents have scale, brand, distribution, data, and talent — your entry is value-destroying. You'll spend more acquiring customers than they spend retaining them. Efficiency gaps don't exist long enough to exploit. The rational move is don't play.
P2: You're assuming a static snapshot. Markets aren't equilibria — they're processes. Incumbents optimize for current demand curves, cost structures, and regulatory environments. When any of those shift, their optimization becomes a liability. Size means inertia. Their strength is their weakness — they can't pivot without cannibalizing what works.
P3: You're both trapped in the frame of "competing." The deepest entries don't compete — they redefine the category. Incumbents aren't your enemy; their customers' unspoken frustrations are your opening. The question isn't "how do I beat them" — it's "what job are customers hiring this product for that nobody's actually solving well?"
ROUND 2: Why Incumbents Are Vulnerable
P1: Even granting shifts happen — incumbents see them too. They have more resources to respond. First-mover advantage in response belongs to whoever has the most capital and talent. You think Walmart didn't see e-commerce? They just built their own.
P2: Seeing ≠ acting. Three structural reasons incumbents fail to respond:
- Margin architecture — they can't adopt lower-margin models without board revolt.
- Organizational antibodies — middle management kills internal disruption to protect their units.
- Customer capture paradox — their best customers demand improvements to the existing product, not a new one. Clayton Christensen's core insight still holds.
P3: Add a fourth: identity lock-in. Companies become their narrative. "We are a [X] company." That sentence is a prison. When the world needs Y, they'll spend billions trying to make X look like Y rather than just building Y. IBM was a "hardware company." Kodak was a "film company." The identity kills them slower than any competitor could.
ROUND 3: So What Actually Works?
P1: Fine. Give me the mechanism. Concretely — how does a nobody with no resources, no brand, no data actually take share from someone who has all of it?
P2: Seven proven vectors:
- Serve the overserved — incumbents add features, raise prices, and overshoot what most customers need. Enter below with "good enough" at radically lower cost. (Christensen's low-end disruption.)
- Serve the non-consumer — find people who can't access the current solution at all. Build for them. Incumbents ignore them because the unit economics don't work at scale. (Mobile banking in Africa.)
- Ride a technology shift — new tech resets cost structures. If cloud, AI, mobile, or a protocol change makes something 10x cheaper, the entrant who builds natively on the new stack beats the incumbent who's porting legacy.
- Exploit regulatory change — deregulation, new mandates, trade shifts create windows. Incumbents lobby to preserve the old regime; entrants build for the new one.
- Unbundle — incumbents bundle because it's efficient for them. But customers pay for things they don't use. Unbundle and sell only the piece people actually want, cheaper.
- Rebundle differently — take pieces from multiple incumbents' products and combine them into a new integrated offering no single player provides.
- Change the business model — same product, different monetization. Subscription vs. purchase. Freemium vs. enterprise. Commission vs. license. The model shift changes who can afford it and how they experience it.
P3: Those are tactical. I'll add the strategic layer most people miss:
- Asymmetric commitment — incumbents allocate 5% attention to your space; you allocate 100%. You will always out-learn, out-iterate, and out-care them in your niche. This compounds.
- Taste and opinion as moat — in commoditized markets, having a point of view is differentiation. Incumbents committee-design for everyone; you design with conviction for someone specific. That specificity creates loyalty algorithms can't replicate.
- Community as product — the product isn't the thing, it's belonging. Build the tribe first, then give them tools. Incumbents sell products; you sell identity.
ROUND 4: Critical Stress-Test
P1: I'll push back on each:
On low-end disruption — incumbents now expect this. Amazon, Google, etc. actively cannibalize themselves. The playbook is known.
On tech shifts — if the shift is obvious, incumbents adopt it. Cloud didn't kill banks; banks moved to cloud.
On community — communities are fragile, expensive to maintain, and don't scale predictably.
P2: Fair critiques. Refinements:
- Self-cannibalizing incumbents are rare — most claim to do it, few actually do. Amazon is the exception, not the rule. Most public companies face quarterly earnings pressure that prevents true self-disruption.
- Tech adoption ≠ tech-native. Banks "moved to cloud" but kept legacy architecture, processes, compliance layers. Fintechs built cloud-native. The difference is 10x in speed and cost, even if both "use cloud."
- Communities don't need to scale to be valuable. A 1,000-person community with 80% retention and 50% word-of-mouth rate is more valuable than 100K followers with 2% engagement. Depth beats breadth at the entry stage.
P3: The meta-point P1 keeps missing: you don't need to win the whole market. Incumbents think in market share. Entrants should think in minimum viable dominance — own a micro-segment so completely that you're the default. Then expand adjacently. The error is thinking you need to be competitive across the board on day one. You don't. You need to be unchallengeable in a space too small for the incumbent to care about — until it's too late.
ROUND 5: When Should You NOT Enter?
P1: Finally my territory. Don't enter when:
- The incumbent has true network effects with high switching costs (payments rails, social graphs, developer ecosystems).
- The market is shrinking and the incumbent is the efficient last-mover.
- Regulation creates hard barriers (banking charters, spectrum licenses, pharma approvals) with no upcoming reform window.
- The product is genuinely commoditized with no under-served segment — price is the only variable and the incumbent has the lowest cost structure.
P2: Agreed — with caveats. Network effects can be beaten with interoperability plays, multi-homing strategies, or category creation that makes the old network irrelevant. Regulatory barriers fall faster than people expect when political winds shift. But yes — honest assessment of structural barriers is essential. Most failed startups die not from bad execution but from entering structurally impossible markets.
P3: I'll add one more don't-enter signal: if you can't articulate what you believe that the incumbent cannot believe. If your thesis is "we'll execute better," you'll lose. You need a belief about the world that is structurally impossible for the incumbent to hold — because holding it would require them to destroy their current business. That asymmetric belief is the only real edge.
ROUND 6: Synthesis — The Entrant's Framework
All three agree on core principles:
- Never compete on the incumbent's terms.
- The entry point must exploit a structural asymmetry — not just a gap.
- Speed of learning matters more than speed of execution.
- Start where the incumbent can't or won't respond.
- Have a belief about the future that the incumbent cannot hold.