# AI Radar

URL: https://www.onsomble.ai/tools/ai-radar

AI search moves fast. This helps you keep up.

A quarterly read on the answer engines, signals, and tactics shaping how AI talks about your business, rated by what's worth acting on now.

## Ratings

- Proven — Battle-tested in production. Build on it with confidence.
- Promising — Strong signal and real results. Worth committing a pilot to.
- Emerging — Interesting and early. Worth a spike or exploration session.
- Watch — Early or unproven. Follow the space but don't commit yet.

## Models & Foundations

The foundation models that power AI assistants. What answer engines build on.

- [Claude](https://www.onsomble.ai/tools/ai-radar/claude-opus-4) — Proven; movement: none. Anthropic's flagship model family. When Claude searches the web, the answer comes back with the URLs it pulled from, which makes it the easiest answer engine to audit if you want to know what content of yours it actually trusts.
- [Gemini 3.1 Pro](https://www.onsomble.ai/tools/ai-radar/gemini) — Proven; movement: up. Google's flagship multimodal model family. Gemini 3.1 Pro is the model behind Google AI Overviews and Google AI Mode, and via Apple's January 2026 deal it's also the model behind the rebuilt Siri. That makes it the engine increasingly answering questions that used to go through Google Search, plus the one that will mediate iPhone queries.
- [GPT-5 Family](https://www.onsomble.ai/tools/ai-radar/gpt-5-family) — Proven; movement: none. OpenAI's current model family, behind ChatGPT and the OpenAI API. ChatGPT Search runs on GPT-5 and is the highest-volume AI surface where consumer questions about businesses get answered.
- [Reasoning Models](https://www.onsomble.ai/tools/ai-radar/reasoning-models) — Proven; movement: up. Models that perform extended chain-of-thought before answering. Includes OpenAI's o-series, Claude's extended thinking mode, Gemini Deep Think, and DeepSeek R1. Increasingly the default for complex queries about businesses and products.
- [DeepSeek](https://www.onsomble.ai/tools/ai-radar/deepseek-r1) — Promising; movement: up. A Chinese open-weight model family with strong reasoning capabilities. DeepSeek R1 and successors power a meaningful slice of Asian-market answer engines and cost-sensitive enterprise deployments globally.
- [Llama](https://www.onsomble.ai/tools/ai-radar/llama-4) — Promising; movement: up. Meta's open-weight model family. Llama matters for AEO because Llama-derived models power a long tail of vertical search engines, enterprise assistants, and embedded chat experiences that aggregate to non-trivial citation volume.
- [Mistral](https://www.onsomble.ai/tools/ai-radar/mistral) — Promising; movement: none. French AI lab shipping open-weight and proprietary models, plus the Le Chat consumer answer engine. The default frontier-grade option for EU enterprises with sovereignty or data-residency constraints, and a meaningful European AEO surface in its own right.
- [Amazon Nova](https://www.onsomble.ai/tools/ai-radar/amazon-nova) — Watch; movement: none. Amazon's Bedrock-native model family, increasingly deployed inside Amazon's own consumer surfaces (Rufus shopping assistant, Alexa+) and AWS enterprise contexts.

## Answer Engines

The AI assistants people search with. Where your business shows up, or doesn't.

- [ChatGPT](https://www.onsomble.ai/tools/ai-radar/chatgpt) — Proven; movement: new. OpenAI's consumer answer engine. Powered by GPT-5 plus live web retrieval via ChatGPT Search. The single highest-volume AI surface where consumer questions about businesses get answered.
- [Gemini.app](https://www.onsomble.ai/tools/ai-radar/gemini-app) — Proven; movement: new. Google's standalone consumer Gemini app and the Gemini integrations across Google Workspace, Android, and now Apple Intelligence (via the January 2026 partnership). A distinct AEO surface from AI Overviews and AI Mode in Search.
- [Microsoft Copilot](https://www.onsomble.ai/tools/ai-radar/microsoft-copilot) — Proven; movement: new. Microsoft's family of AI assistants spanning consumer chat (copilot.microsoft.com), Microsoft 365 (Word, Excel, Teams, Outlook), and developer (GitHub). Powered by GPT-5 family with Bing-grounded retrieval. The only major answer engine with publisher-side AEO tooling baked in.
- [Perplexity](https://www.onsomble.ai/tools/ai-radar/perplexity) — Proven; movement: new. A citation-first AI answer engine. Asks a question in natural language, retrieves the web in real time, and returns an answer with inline numbered sources. 45M+ monthly users, over 1 billion queries per month, $450M ARR (March 2026). Now extended via the Comet AI browser.
- [Claude.ai](https://www.onsomble.ai/tools/ai-radar/claude-app) — Promising; movement: new. Anthropic's consumer answer engine, separate from the Claude API. Smaller user base than ChatGPT or Gemini, but the most transparent citation UX and a technical-skewed audience that disproportionately influences B2B and engineering buying decisions.
- [Grok](https://www.onsomble.ai/tools/ai-radar/grok) — Emerging; movement: new. xAI's answer engine, deeply tied to X (formerly Twitter). Grok 4.20 features a 2M token context window and four-agent collaborative architecture. Real-time access to roughly 68 million English-language posts per day from X.

## AI Search Behaviour

How discovery is shifting: AI Overviews, zero-click, conversational query patterns.

- [Google AI Overviews](https://www.onsomble.ai/tools/ai-radar/ai-overviews) — Proven; movement: up. Google's generated summaries that appear above traditional Search results. Powered by Gemini, expanded to 200 countries and 40 languages as of May 2025, and now triggering on a meaningful share of informational queries. The single biggest reshaping of search behaviour in 2026.
- [Zero-Click Search](https://www.onsomble.ai/tools/ai-radar/zero-click-search) — Proven; movement: up. Searches where the user gets the answer they need without clicking through to any source page. Up from roughly 50% pre-AI to roughly 69% of all queries by mid-2025 (Similarweb). The structural reason traditional SEO traffic is declining and AEO has emerged as a discipline.
- [Google AI Mode](https://www.onsomble.ai/tools/ai-radar/google-ai-mode) — Promising; movement: up. Google Search's conversational tab, powered by Gemini. Multi-step Q&A, follow-up questions, multimodal input, and inline numbered citations. Rolled out broadly to US users in May 2025 without Labs sign-up. The long-form successor to traditional Search results.
- [Deep Research Mode](https://www.onsomble.ai/tools/ai-radar/deep-research-mode) — Emerging; movement: new. The long-form synthesis mode now offered by ChatGPT, Claude, Gemini, and Perplexity. Browses many sources, takes minutes to complete, and produces report-style outputs with rich citations. The AEO surface for considered-purchase, comparison, and decision-stage queries.
- [Query Fan-Out](https://www.onsomble.ai/tools/ai-radar/query-fan-out) — Emerging; movement: new. The AI search behaviour where one user query decomposes into multiple parallel sub-queries (often 8-10) before retrieval and synthesis. The reason topical authority and adjacent-content depth matter more in 2026 than they did under traditional SEO.

## AEO Foundations

Technical signals AI uses to find and cite you. The AEO equivalent of on-page SEO.

- [E-E-A-T signals](https://www.onsomble.ai/tools/ai-radar/e-e-a-t-signals) — Proven; movement: none. Google's framework for evaluating content quality (Experience, Expertise, Authoritativeness, Trust) has migrated from being a Search ranking input to being an AEO citation input. Gemini's AI Mode and AI Overviews lean on E-E-A-T signals when deciding which sources to cite.
- [Schema.org structured data](https://www.onsomble.ai/tools/ai-radar/schema-org-structured-data) — Proven; movement: up. JSON-LD structured data using Schema.org vocabulary. The highest-leverage AEO investment available in 2026: Gemini's AI Mode uses schema to verify claims and establish entity relationships during answer synthesis, and other engines increasingly weight schema-backed sources higher in retrieval.
- [AI crawler controls](https://www.onsomble.ai/tools/ai-radar/ai-crawler-controls) — Promising; movement: up. The technical controls publishers use to manage AI crawler access: robots.txt directives for GPTBot, ClaudeBot, Google-Extended, PerplexityBot, and others, plus emerging paid-access models like Cloudflare's Pay-Per-Crawl (private beta, Q1 2026 public launch). The publisher-side lever in the AEO equation.
- [Content freshness signals](https://www.onsomble.ai/tools/ai-radar/content-freshness-signals) — Promising; movement: up. The technical and on-page signals that tell AI engines (and Search) when content was last updated. `dateModified` schema, sitemap `lastmod` values, and visible publish dates. AI engines weight freshness when retrieving sources, which means stale content can fall out of citation pools even if it still ranks on traditional Search.
- [Entity authority](https://www.onsomble.ai/tools/ai-radar/entity-authority) — Promising; movement: up. Establishing your business as a verified, well-linked entity in the knowledge graphs that AI engines lean on. Wikidata is the primary input to Google's Knowledge Graph. Schema's `sameAs` property links your entity across authoritative external sources. Entity authority is the AEO investment that pays off across every engine, not just one.
- [llms.txt](https://www.onsomble.ai/tools/ai-radar/llms-txt) — Promising; movement: up. A proposed standard for a Markdown file at `/llms.txt` that gives LLMs a clean, structured summary of your site's most important content. Adopted by Anthropic, Cloudflare, Vercel, and the Mintlify-hosted ecosystem. Not (yet) recognised by Google as a ranking signal.

## Content & Knowledge

The substrate AI pulls from: content patterns, knowledge sources, brand SERPs.

- [Reddit & community signals](https://www.onsomble.ai/tools/ai-radar/reddit-community-signals) — Proven; movement: up. Reddit is the most cited domain across major AI engines, with around 40% citation frequency. Google's $60M/year licensing deal and OpenAI's roughly $70M/year deal give those engines real-time access to Reddit's user-authored content. The dominant citation substrate for community-validated questions and brand opinions.
- [Wikipedia presence](https://www.onsomble.ai/tools/ai-radar/wikipedia-presence) — Proven; movement: none. Wikipedia is foundational training material for AI engines. Independent research places it at 26-48% of ChatGPT's top citations (47.9% of ChatGPT's top citations overall). Less prominent in Google AI Mode (~2%), but still load-bearing for any business notable enough to qualify for an article.
- [Brand SERP control](https://www.onsomble.ai/tools/ai-radar/brand-serp-control) — Promising; movement: up. What users (and AI engines) see when they search your brand name on Google. Encompasses your Knowledge Panel, top organic results, social profiles, news mentions, review platforms, and any owned properties. The substrate AI engines use to build their model of who your business is.
- [Topical authority](https://www.onsomble.ai/tools/ai-radar/topical-authority) — Promising; movement: up. The depth and breadth of content you publish across a topic family. Because AI engines decompose user queries into 8-10 sub-queries via fan-out, comprehensive topical coverage outperforms single high-quality pages for AEO. The content investment that compounds across every AI surface.
- [YouTube & video transcripts](https://www.onsomble.ai/tools/ai-radar/youtube-video-transcripts) — Promising; movement: up. YouTube is the dominant video citation source for AI answer engines, cited around 200× more than any other video platform. AI engines retrieve and read transcripts (not videos), which makes transcript quality, video metadata, and structured descriptions the load-bearing AEO signals for video content.

## Visibility & Measurement

How presence and sentiment are measured across answer engines.

- [AI traffic attribution](https://www.onsomble.ai/tools/ai-radar/ai-traffic-attribution) — Promising; movement: up. Measuring referral traffic and conversion behaviour from AI engines. ChatGPT drives 87.4% of AI referral traffic, and AI-referred visitors convert at 14.2% versus Google's 2.8% (averi.ai). GA4 underreports the true volume because free ChatGPT users don't pass referrer data. Setting up correct attribution is the first AEO measurement task most teams skip.
- [Citation tracking](https://www.onsomble.ai/tools/ai-radar/citation-tracking) — Promising; movement: up. Per-URL measurement of which pages get cited when AI engines answer brand-relevant queries. The diagnostic layer underneath share of voice. Reveals which content is actually doing the AEO work, which is rarely obvious without measurement.
- [Share of voice](https://www.onsomble.ai/tools/ai-radar/share-of-voice) — Promising; movement: up. The percent of brand-relevant queries across major AI engines where your business is mentioned. The headline AEO metric. Useful for benchmarking against competitors and tracking the directional impact of AEO work, but coarse on its own.
- [Hallucination & accuracy auditing](https://www.onsomble.ai/tools/ai-radar/hallucination-accuracy-auditing) — Emerging; movement: new. Validating whether AI engines describe your business accurately. Wrong pricing, phantom features, confused competitor comparisons, and outdated company descriptions are the norm for brands that haven't actively monitored AI accuracy. Most general-purpose AEO platforms don't validate accuracy, which is why specialised hallucination-auditing tools have emerged.
- [Sentiment monitoring](https://www.onsomble.ai/tools/ai-radar/sentiment-monitoring) — Emerging; movement: up. Automated classification of the tone in AI mentions about your brand. Most AEO platforms now ship sentiment analysis alongside share of voice and citation tracking. Useful for catching reputation drift in AI answers before it shows up in customer behaviour, but model-dependent and noisy.

## AEO Platforms

The competitive set: tools helping businesses optimise for AI search.

- [AthenaHQ](https://www.onsomble.ai/tools/ai-radar/athenahq) — Promising; movement: new. GEO platform with autonomous content agents. Founded by ex-Google Search and DeepMind engineers (Andrew Yan, Alan Yao). $2.2M from Y Combinator and SEO industry veterans. Tracks citations across ChatGPT, Perplexity, Claude, Gemini, Copilot, and Google AI Overviews; uses agents to identify content gaps and draft optimisations.
- [Otterly AI](https://www.onsomble.ai/tools/ai-radar/otterly-ai) — Promising; movement: new. Established AI search monitoring platform. 20,000+ marketing professionals; G2, OMR, and Gartner recognition. Tracks AI Overviews, AI Mode, ChatGPT, Gemini, Copilot, and Perplexity. Differentiators: GEO Audit (25+ on-page factors that affect AI citation), Prompt Research (transforms keywords into real prompts), Brand Reports, citation + sentiment analysis, and a track record of credible AEO research.
- [Peec AI](https://www.onsomble.ai/tools/ai-radar/peec-ai) — Promising; movement: new. AI search analytics platform for marketing teams and SEO agencies. $21M raised. Customers include Axel Springer, Chanel, n8n, ElevenLabs, and TUI. Adding ~300 customers per month in early 2026. Focuses on visibility tracking, prompt analysis, citation tracking, sentiment, and competitive benchmarking across ChatGPT, Perplexity, Gemini, and Google AI Overviews.
- [Profound](https://www.onsomble.ai/tools/ai-radar/profound) — Promising; movement: new. Enterprise-grade AI visibility platform tracking ChatGPT, Gemini, Claude, Perplexity, and others. Features include share of voice, sentiment, competitive benchmarking, plus differentiators: Agent Analytics (how AI views your site), Prompt Volumes (what your audience actually asks), Shopping Insights (product representation in ChatGPT shopping), and Agents (automated content campaigns).
- [Goodie AI](https://www.onsomble.ai/tools/ai-radar/goodie-ai) — Emerging; movement: new. GEO platform tracking 11+ AI models including ones other platforms often skip (Grok, DeepSeek, Meta AI, Amazon Rufus, AI Mode). Built for GEO from day one rather than retrofitted SEO tooling. Combines AI Visibility Monitoring, AI Optimization Hub (schema + semantic content suggestions), AEO Content Writer, Topic Explorer, and Traffic + Attribution.
- [Scrunch AI](https://www.onsomble.ai/tools/ai-radar/scrunch-ai) — Emerging; movement: new. AEO platform with a distinctive infrastructure layer (Agent Experience Platform / AXP) that sits between your website and AI crawlers, serving compressed, structured content optimised for LLM consumption. Standard measurement features (SOV, citations, sentiment, prompt-level reporting) plus GA4 integration for AI referral traffic. Pricing from $300/mo brands, $500/mo agencies. Enterprise SOC 2.

## Agent Protocols

Open standards letting AI assistants act on behalf of users. The agent layer.

- [Model Context Protocol](https://www.onsomble.ai/tools/ai-radar/model-context-protocol) — Proven; movement: none. An open protocol for connecting AI assistants to external tools and data sources. The de facto standard for letting models access live business systems (bookings, inventory, eligibility checks) instead of just describing them.
- [A2A Protocol](https://www.onsomble.ai/tools/ai-radar/a2a-protocol) — Promising; movement: none. An open protocol for agent-to-agent communication and delegation. Originally developed by Google, donated to the Linux Foundation. Where MCP connects AI assistants to tools, A2A connects AI assistants to other AI assistants, including specialist agents representing specific businesses.
- [Computer Use](https://www.onsomble.ai/tools/ai-radar/computer-use) — Promising; movement: new. Anthropic's Computer Use lets Claude control a computer the way a human does: looking at the screen, moving a cursor, clicking, and typing. Generally available in Claude Cowork, Claude Code, and as an API tool. Includes prompt-injection scanning, user confirmation, and a zoom feature for viewing screen regions at full resolution. Available in Opus 4.7, 4.6, 4.5, and Sonnet 4.6.
- [ACP — Agent Communication Protocol](https://www.onsomble.ai/tools/ai-radar/acp) — Emerging; movement: new. Open standard for inter-agent communication, originally introduced by IBM's BeeAI project, now under Linux Foundation governance alongside A2A. Lightweight, HTTP-native, JSON-RPC over HTTP/WebSockets. Recent specification additions include Trajectory Metadata, Distributed Sessions, Citation Metadata, and High Availability Support.
- [Agent discovery manifests](https://www.onsomble.ai/tools/ai-radar/agent-discovery-manifests) — Watch; movement: new. The proposed standards for declaring an agent's identity, capabilities, and trust signals at a well-known URL. Multiple candidates compete: ai-agent.json (Aiia), agent-card.json (A2A v1.0), MCP Server Cards (proposed for the MCP roadmap), and others. None has emerged as the default; the eventual winner shapes how agent-mediated commerce is discovered.
- [OAGP — Open Agent Gateway Protocol](https://www.onsomble.ai/tools/ai-radar/oagp) — Watch; movement: new. A protocol architecture that treats HTML as the universal agent interface and the web as the agent-to-service communication layer. Created by Onsomble (the publisher of this radar, disclosed). Differs from MCP and A2A in being semantic-first: an agent that can read a webpage can use the service at the basic level, with progressive enhancement via JSON-LD metadata and capability tokens.
