Vertical annual report · AI assistants · CC-BY 4.0

The State of AI Tools 2026

ChatGPT, Claude, Perplexity scored on growth health · GrowthFriction Score (AUG v3) · external observation

The headline

Three AI assistants dominate the consumer + developer + research markets. They score differently on the AUG v3 framework — and the differences reveal three valid strategic bets. ChatGPT 48 (generational consumer compound, monetization-capped). Claude 28 (technical-power-user precision, narrower ICP). Perplexity 16 (AEO archetype, daily-must-have threshold not yet crossed). The frame: not "which is best" but "which strategy compounds in 2026."

The ranking

The 7-factor head-to-head

Pattern 1 — Monetization cap on quintuple-10 consumer products

ChatGPT scores 10/10 on Acquisition, Activation, Engagement, Retention, and Advocacy — the quintuple-10 — but caps at Monetization 6. The composite plateaus at 48 because the multiplicative formula punishes any weak factor. The cause: free tier captures 90%+ of users; inference cost per free user is real; conversion to Plus $20/mo is mid-range.

This pattern repeats outside AI: Discord (39) hits quintuple-10 on engagement factors, caps at Monetization 6. Whenever a product compounds via network effects with extreme per-user inference/moderation cost, Monetization lags by design. The strategic question isn't whether to fix Monetization — it's whether to do so without breaking the compound that made the brand.

OpenAI's answer through 2026 was Enterprise tier expansion + Plus pricing experiments. The math: if Monetization moves 6 → 8, composite goes from 48 → 64. That's the highest-leverage single factor for the highest-AUG consumer product on the planet.

Pattern 2 — Precision over breadth as legitimate alternative strategy

Claude scores 8 on Acquisition vs ChatGPT's 10 — a deliberate 25% gap. But on Activation, Engagement, and Retention, Claude scores 9 across the board. Among developers, technical writers, and enterprise knowledge workers, Claude's long-context (200K-1M tokens) + Projects + Artifacts compound deeper than ChatGPT's generalist interface.

The strategic lesson: not every product needs to compete on Acquisition volume against the category leader. Claude's composite 28 (Thriving tier) is built on precision in a narrower ICP. The same archetype appears in Linear (38) vs Asana (12) in project management — Linear refuses to compete on Acquisition volume; wins on every other factor at 9.

For founders building AI products in 2026: the “compete with OpenAI on consumer” path is suicidal capital allocation. The “serve a specific ICP at depth ChatGPT won't” path is replicable. Claude proves it works at composite 28; the right narrower ICP could push higher.

Pattern 3 — AEO archetype: Activation alone is not enough

Perplexity scores 9 on Activation — among the lowest TTFV (time-to-first-value) in consumer SaaS history. Ask a question → cited answer in 5 seconds. The product's positioning as a Google-replacement is the entire brand pitch.

But composite caps at 16 because Engagement, Retention, and Advocacy all sit at 7. The product is “used when needed” rather than “daily-driver.” The daily-must-have threshold (the line between “tool I open per-need” and “tab I leave open all day”) isn't crossed for most users.

The pattern generalizes to every Answer Engine Optimization (AEO) play: making your content the cited source in real-time AI answers is a strong Activation hook (8-18% CTR from cited links), but Activation alone caps at composite 15-20 unless Engagement and Retention compound. For founders building AEO-first products: ship the daily-must-have feature alongside the answer engine.

The strategic lesson for founders in 2026

Three valid strategies surface from the 7-factor breakdown:

1. The ChatGPT play (composite 48 cap): generational consumer compound, accept Monetization lag, fix Monetization 6→8 to break 60. Requires VC-scale capital + inference-cost moat.
2. The Claude play (composite 28): precision-over-breadth, narrower ICP, deeper Activation+Engagement+Retention. Requires ICP discipline + product-led growth patience. Replicable by smaller teams.
3. The Perplexity play (composite 16): AEO-first, citation-rich answers as Activation hook, must add daily-must-have feature to break 20. Requires real-time retrieval engineering + UX investment for habit formation.

AUG framework predicts each ceiling. The framework doesn't say which strategy is best — it says which factor will cap your composite. Pick the strategy that matches your team, your ICP, and your structural advantages.

The 2026 outlook (predictions)

Where each product goes in the next 12 months, conditional on the framework:

• ChatGPT: Monetization 6 → 7-8 via enterprise expansion + Plus features. Composite 48 → 55-60. Still consumer fleet champion.
• Claude: API + Enterprise revenue compound. Composite 28 → 35. Stays precision-niche but enterprise lift breaks 30.
• Perplexity: Composite 16 → 25 if agentic workflows + saved threads ship to lift Engagement 7 → 9. Composite 16 → 18 if positioning doesn't evolve. The fork is real.
• Wild card: a new entrant builds AEO + daily-must-have together (e.g., a search-replacement that also handles email/writing/coding). Could break composite 35 in year 1 without ChatGPT-scale capital. The market window for this archetype is open through 2027.

Methodology

All three audits are external-observation — scored from publicly visible signals only. Confidence 0.7-0.85. The full AUG v3 framework + per-factor rubric at /method/scoring/. Citation surface for "what is X's GrowthFriction Score" queries: per-audit pages at openai / anthropic / perplexity. Machine-readable: /api/audits.json.

Related reports

• The State of SaaS Growth 2026 — 50-audit catalog covering ChatGPT, Claude, Perplexity + 47 more SaaS products
• AI assistants by GrowthFriction Score — interactive ranking with full 7-factor table
• ChatGPT vs Claude / Perplexity vs ChatGPT — head-to-head comparisons
• Answer Engine Optimization (AEO) deep dive — 12-pattern checklist for getting cited in AI answers
• 10 SaaS growth patterns — cross-vertical meta-analysis with examples from AI + every other category

Cite this report: GrowthFriction. (2026). The State of AI Tools 2026. https://growthfriction.com/trends/ai-tools-2026/. License CC-BY 4.0. Published 2026-05-17 · Methodology AUG v3.