Six executable frameworks that turn positioning from vague intuition into operational discipline. In the AI era, positioning beats product.
The AI market has reached an inflection point. With $127B+ poured into AI startups in 2025 alone, the competitive dynamics have shifted dramatically. Here is what defines the landscape right now.
Your position in the Defensibility vs. Incumbent Attention matrix dictates your strategy. Click each quadrant to reveal the prescriptive playbook.
Real companies positioned across the matrix. Click any card to expand the positioning analysis.
Glean has built one of the strongest integration moats in enterprise AI. Their connectors span Slack, Google Workspace, Salesforce, Jira, Confluence, and 100+ others, with deep permission-aware indexing that took years to build. Each customer deployment creates richer data flywheel effects. By February 2026, Glean has achieved $300M+ ARR and incumbents find it nearly impossible to replicate the breadth of their integration surface.
Viz.ai operates in a near-perfect Sweet Spot: FDA clearance creates massive regulatory barriers, proprietary clinical imaging data is impossible for Big Tech to access, and hospital integration creates sticky switching costs. Incumbents like Google Health have repeatedly failed to match the depth of workflow integration. By 2026, Viz.ai processes millions of scans across 1,400+ hospitals with a data advantage that compounds monthly.
Palantir's Artificial Intelligence Platform (AIP) leverages two decades of government relationships and security infrastructure that no startup or Big Tech company can easily replicate. Their ontology layer, mapped to classified and sensitive data environments, creates astronomical switching costs. AIP's 2025 revenues exceeded $3.5B, with government contracts that span 5-10 year horizons.
Veeva has layered AI across its life sciences platform, combining clinical trial management, regulatory compliance, and pharma CRM into a deeply integrated stack that no horizontal AI player can match. Their domain-specific data moat spans 1,000+ pharma companies. In 2025-2026, their AI features for drug safety signal detection and clinical trial optimization have proven impossible for Salesforce or generic AI tools to replicate.
Cursor is a textbook Battlefield position: high defensibility through a passionate developer base and UX innovations (tab-complete, inline editing, chat-with-codebase), but extremely high attention from Microsoft (GitHub Copilot), JetBrains, and Google. By 2026, Cursor has grown past $200M ARR but faces relentless feature-matching from incumbents. Their survival depends on maintaining a 6-12 month UX lead and deepening the ecosystem around their editor.
Perplexity sits in one of the most visible Battlefield positions in all of AI. They are building genuine defensibility through brand, data, and user habit formation, but they are attacking Google's $175B search business directly. Google's AI Overviews and Gemini integration are existential counter-moves. With $500M+ raised by early 2026, Perplexity has runway, but the question is whether any moat can withstand Google's full attention.
Anthropic occupies a unique Battlefield position: competing at the frontier model layer against OpenAI, Google, and Meta, but with genuine defensibility through research talent density, safety-first brand positioning, and growing enterprise trust. By February 2026, Claude has become a genuine alternative to GPT in enterprise settings. However, the capital requirements and competitive intensity at the model layer make this one of the most expensive Battlefields in tech history.
Dozens of startups built RAG-based "chat with your documents" products in 2023-2024. By 2026, most remain in the Waiting Room: they have low incumbent attention (the market is too fragmented to attract Big Tech focus), but their defensibility is near-zero because RAG has become a commodity capability embedded in every major platform. These companies must find vertical niches or build proprietary data flywheels within 12 months or face slow death.
The AI voice agent space exploded in 2025 with hundreds of startups. Most sit in the Waiting Room: the market is fragmented enough that Big Tech has not yet focused, but the technology is easily replicated. Players like Bland AI, Retell, and Vapi are competing on price and features with minimal switching costs. The winners will be those who lock into specific verticals (dental, real estate, insurance) with deep workflow integrations before Twilio, Google, or Amazon consolidate the space.
By 2026, generic AI writing tools represent the purest Kill Zone. ChatGPT, Claude, Gemini, and Copilot all offer writing assistance natively. Google Docs, Notion, Microsoft Word, and virtually every text editor now includes AI writing features. Companies like Jasper, Copy.ai, and others that built thin wrappers around GPT have seen massive churn as users realize the underlying capability is freely available everywhere. The lesson: if your value proposition is "we call an LLM API and format the output," you are in the Kill Zone.
Image generation has become a canonical Kill Zone example. With DALL-E in ChatGPT, Imagen in Google products, Grok's Aurora, and open-source models like Flux and Stable Diffusion 4, standalone image generation wrappers have no defensible position. The survivors are those who pivoted to vertical creative workflows (product photography, architectural visualization) or built genuine model differentiation (Midjourney's distinctive aesthetic). Everyone else is being crushed.
Five questions every AI startup team should answer every single week. Click each to reveal the measurement methodology.
Did our moats get deeper or shallower this week? Measure user retention trends, switching cost proxies (integrations activated, data volume ingested, workflows dependent on your product), and data advantage metrics (unique data points accumulated, model performance improvements from proprietary data).
Measure: retention, switching costs, data advantageAny signals that incumbents are noticing our space? Track competitor announcements, job postings in your niche, patent filings, conference mentions, and acquisitions of adjacent startups. In February 2026, pay special attention to OpenAI, Google, Microsoft, and Meta product launch cadences—they are shipping weekly.
Measure: competitor signals, job postings, patentsAre we moving toward the Sweet Spot or away from it? Track your quadrant position trend over a 4-week rolling average. The goal is always movement upward (more defensible) or leftward (less visible to incumbents). Any rightward or downward drift requires immediate corrective action.
Measure: 4-week rolling quadrant trendWhat percentage of our resources went to moat-building vs. feature-building this week? If you are not in the Sweet Spot, the target is 40%+ of engineering resources dedicated to moat-building activities (integration depth, data pipeline improvements, community building, regulatory positioning) rather than new features.
Target: 40%+ to moats if not in Sweet SpotIf in Kill Zone or Waiting Room, are we on track to escape within our runway? Calculate weeks to defensibility milestone vs. weeks of runway remaining. If the ratio is trending worse, escalate immediately. In the 2026 funding environment, emergency capital is available but comes with extreme dilution.
Measure: weeks to milestone vs. weeks of runwayScore each moat type on Time to Defensibility (speed) and Depth Potential (durability). Prioritize moats scoring 7+ combined, weighted toward speed if runway is limited.
| Moat Type | Description (Feb 2026) | Time to Defensibility | Depth Potential | Combined |
|---|---|---|---|---|
| Proprietary Data Flywheel | User interactions improve model; models attract more users. The dominant moat of 2025-2026. | 3 | 5 | 8 |
| Regulatory / Compliance Barrier | FDA, HIPAA, SOC2, FedRAMP certifications. Especially powerful in healthcare, finance, defense AI. | 2 | 5 | 7 |
| Deep Integration / Switching Costs | Multi-system integrations, workflow embedding, API dependencies. Each integration deepens lock-in. | 4 | 4 | 8 |
| Community / Network Effects | User-generated content, marketplace dynamics, community-contributed models and templates. | 3 | 4 | 7 |
| Vertical Domain Expertise | Deep understanding of specific industry workflows, jargon, edge cases. Hard for horizontal players to replicate. | 4 | 3 | 7 |
| Brand / Trust | Reputation for reliability, safety, accuracy in high-stakes domains. Critical for AI in healthcare, legal, finance. | 2 | 4 | 6 |
| Speed / UX Innovation | Superior product experience and iteration velocity. Necessary but insufficient alone—easily copied. | 5 | 1 | 6 |
| Proprietary Model / Architecture | Custom-trained models with unique capabilities. Increasingly rare as open-weight models improve. | 1 | 3 | 4 |
Decision Rule: Prioritize moats with combined score of 7 or above. Weight toward Time to Defensibility if runway is under 18 months. In February 2026, the highest-ROI moats are Proprietary Data Flywheels and Deep Integration / Switching Costs—both score 8 and are the primary differentiators separating survivors from casualties.
Score five factors to predict how quickly and aggressively incumbents will attack your market. Adjust the sliders to see your risk level in real time.
For startups in the Kill Zone or Waiting Room: calculate whether your runway gives you enough time to reach defensibility. Enter your numbers below.
Note: A 50% buffer is automatically applied to Time to Defensibility to account for unexpected delays.
Run every major strategic decision through this three-question filter. Select your scores to see the recommendation.
Does this decision increase or decrease your moat depth?
Does this decision increase or decrease incumbent attention on your market?
Answer the questions below to get a personalized positioning recommendation based on your current situation.