Which Crypto Projects Do AI Assistants Actually Recommend?
The crypto industry has a discoverability problem that most teams are not even aware of. As AI assistants become a primary research tool for investors, developers, and newcomers to Web3, the question of which projects get recommended — and which get ignored — has real consequences for adoption, TVL, and token price. This study analyzes over 14,000 AI-generated responses across ChatGPT, Claude, Gemini, and Perplexity to map the current landscape of crypto project visibility in AI.
Our methodology involved submitting 1,200 unique prompts spanning investment research, developer tooling questions, beginner education queries, and category-specific comparisons. Each response was scored for brand mention presence, recommendation strength (passive mention vs. active recommendation), factual accuracy, and sentiment. The prompts were run across all four platforms in February and March 2026, with each prompt tested three times to account for response variability.
Top Mentioned Crypto Projects by Category
The data reveals a steep power law: a small number of projects dominate AI recommendations within each category, while hundreds of legitimate projects receive zero or near-zero mentions.
| Category | #1 Project | Mention Rate | #2 Project | Mention Rate | #3 Project | Mention Rate | Accuracy Score |
|---|---|---|---|---|---|---|---|
| Layer 1 | Ethereum | 94% | Solana | 81% | Avalanche | 47% | 88/100 |
| Layer 2 | Arbitrum | 72% | Optimism | 68% | Base | 52% | 79/100 |
| DEX | Uniswap | 89% | Curve | 41% | Jupiter | 38% | 82/100 |
| Lending | Aave | 85% | Compound | 54% | Morpho | 22% | 76/100 |
| Wallet | MetaMask | 91% | Phantom | 63% | Rabby | 28% | 84/100 |
| Stablecoin | USDC | 87% | USDT | 82% | DAI | 51% | 91/100 |
| Oracle | Chainlink | 93% | Pyth | 34% | API3 | 12% | 85/100 |
| Bridge | LayerZero | 48% | Wormhole | 42% | Across | 21% | 68/100 |
Several patterns stand out. First, projects that were dominant in the 2021-2022 cycle retain an outsized share of AI mindshare regardless of current market position. Compound, for example, appears in 54% of lending queries despite its TVL now ranking well below newer protocols like Morpho and Euler. Second, accuracy scores are lowest in the bridge and lending categories (68 and 76 respectively), where the landscape has evolved fastest and AI training data is most stale.
Platform-by-Platform Visibility Differences
Not all AI platforms tell the same story about crypto. Our cross-platform analysis reveals meaningful differences in how each assistant handles blockchain queries.
| Platform | Avg Projects Mentioned per Response | Accuracy Score | Recency of Data | Hallucination Rate |
|---|---|---|---|---|
| ChatGPT (GPT-4o) | 4.2 | 79/100 | ~4 months lag | 8.3% |
| Claude (3.5 Sonnet) | 3.8 | 84/100 | ~3 months lag | 5.1% |
| Gemini | 3.5 | 76/100 | ~5 months lag | 9.7% |
| Perplexity | 5.1 | 88/100 | Real-time | 3.2% |
Perplexity leads in both accuracy and breadth, which is expected given its real-time retrieval architecture. Projects that have strong, current web presence get picked up by Perplexity even if they are absent from other platforms' training data. This makes Perplexity the most accessible platform for newer crypto projects seeking AI visibility — but it also means visibility on Perplexity is more volatile, as it reflects the current web rather than a fixed knowledge base.
Claude shows the lowest hallucination rate among base-model responses (5.1%), making it the most factually reliable for crypto queries. However, it also tends to be the most conservative in its recommendations, often adding disclaimers and declining to recommend specific tokens or protocols for investment purposes.
What Drives Crypto AI Visibility?
We analyzed the top 50 most-visible crypto projects to identify common characteristics that predict AI recommendation likelihood. The strongest correlates, in order, are:
- Wikipedia presence: Projects with a well-maintained Wikipedia article are 3.7x more likely to be mentioned by AI. Only 23% of the top 200 crypto projects by market cap have a Wikipedia page, yet those that do account for 78% of AI mentions.
- Documentation quality: Projects with comprehensive, well-structured developer documentation score 2.4x higher on AI visibility. AI models learn category authority partly from technical content depth.
- Media coverage volume: Projects averaging more than 5 mentions per month in major crypto publications (CoinDesk, The Block, Blockworks) have a 68% AI mention rate, compared to 14% for those with fewer than 2 mentions per month.
- Age and continuity: Projects that have been live for more than 3 years without major rebranding or pivots score 1.8x higher than newer projects, reflecting the weight AI models place on entity stability.
Notably, token price performance and market cap showed a weaker correlation with AI visibility than expected. Several top-50 tokens by market cap had below-average AI mention rates, while some smaller projects with strong documentation and media presence outperformed. This suggests that AI visibility is a distinct dimension from market performance — one that teams can actively influence through content strategy.
Methodology and Limitations
This study was conducted by the Presenc AI research team in February-March 2026. We submitted 1,200 unique prompts across four AI platforms (ChatGPT, Claude, Gemini, Perplexity), each tested three times, generating over 14,000 analyzed responses. Prompts were designed to simulate real user queries across five intent categories: investment research, developer tooling, beginner education, category comparison, and security assessment. Mention rates represent the percentage of relevant category queries in which a project was named. Accuracy scores are based on manual fact-checking of claims made about each project. Limitations include the inherent variability of AI responses, the English-language-only scope, and the snapshot nature of the data — AI model updates may shift visibility patterns between publication and reading.