Gap Analysis
The gap is not features. The gap is the difference between demonstrating a booking and completing one.
INFRASTRUCTURE
No Booking Rails
Miso has no disclosed GDS/NDC partner. The AI can't book autonomously—hence the human concierge hires. This is the last-mile problem Reddit identified: "Show me one that can actually book a flight without redirecting to Expedia." Otto's Spotnana integration is the only credible answer in this cohort, and even it hasn't disclosed booking volume.
TRUST
Hallucination Tax
"AI travel" carries negative brand equity. BBC reported tourists in Peru seeking a nonexistent "Sacred Canyon of Humantay." Reddit users cite AI-planned itineraries with closed museums and 8-activity days. The category has a credibility deficit that every new entrant inherits. Specific, verifiable savings ($2,300 rebook) cut through where general claims don't.
POLICY
No Corporate Intelligence
None of the three offer policy engines that understand travel programs, approval chains, or negotiated rates. Otto allows "corporate policy upload"—a file, not an integration. Miso targets corporate group travel in marketing but provides no policy tooling. This is Navan's deepest moat: 10+ years of enterprise travel program data.
DISRUPTION
No Crisis Response
Flight delayed. Connection missed. Hotel overbooked. None of the three handle disruption in production. Otto lists it as "coming soon." Miso claims "delay resolution" but behind a waitlist. Business travelers—Otto's explicit target audience—need this most and have it least.
DEPTH
Shallow Personalization
All three claim personalization. None have the data to deliver it. Otto has 9 App Store ratings—the personalization loop requires volume they don't have. Miso is behind a waitlist. Gondola's personalization is structural (loyalty tier changes prices) rather than behavioral. Navan has millions of historical bookings to train on.
ECONOMICS
Tarpit Dynamics
"Travel has the worst unit economics for AI. High CAC, low margin, massive regulatory surface." $11M combined is a rounding error against the infrastructure cost. Manus AI consumed $11 for a single travel request. The gap between "impressive demo" and "viable business" is where these startups die.
Implications for Edge 2.0
Not what to copy—what to exploit. These startups' failures and users' articulated wants point to five specific design priorities for Edge 2.0.
Show Transactional Depth Immediately
Users have been burned by demos that "spit out errors at monetization time." Edge 2.0 must demonstrate real booking capability in the first interaction—not after onboarding, not after 5 conversations. The confidence gap between "I can search" and "I can book" is the entire value proposition. Gondola's $2,300 rebook savings resonates because it's specific and verifiable.
Lead with Corporate Policy as a Moat
None of the three competitors have policy engines. Otto targets the SMB "unmanaged" segment precisely to avoid this complexity. Edge 2.0 should make policy compliance visible and effortless—show when a booking is in-policy, surface the savings vs. out-of-policy alternatives, auto-route approvals. This is infrastructure no startup can replicate.
Disruption Handling as a Differentiator
Zero competitors handle flight disruptions in production. This is the highest-stress, highest-value moment in business travel. Edge 2.0 that proactively rebooks a missed connection before the traveler reaches the gate would be the single most defensible feature in the category. Users want "an AI EA for taking things off your plate"—this is the purest expression of that.
Don't Build Another ChatGPT Wrapper
"If ChatGPT can do what an AI wrapper can do, then the wrapper is useless." Reddit and Twitter sentiment is clear: the category has a ChatGPT-wrapper credibility problem. Edge 2.0 must feel fundamentally different from a chatbot—through structured UI, proactive intelligence, and visible access to systems a general LLM can't reach (GDS, corporate rates, approval workflows).
Earn Trust Per-Interaction, Not Per-Brand
Hallucination fatigue means "AI travel" is a negative signal. Users won't trust the brand—they'll trust the result. Every interaction must include a verifiable signal: real prices from real GDS, policy compliance confirmation, savings vs. alternatives quantified. Gondola wins on this axis with specific dollar amounts. Miso loses with unsubstantiated claims.