Four New AI Citation Surfaces Landed in April 2026. Most GEO Strategies Don't Track Any of Them.

On April 28, 2026, General Motors announced that Google Gemini is rolling into approximately 4 million GM vehicles in the US. Model-year 2022 and newer Cadillac, Chevrolet, Buick, and GMC vehicles replace their existing Google Assistant with Gemini. Drivers ask conversational questions. Gemini answers them. Those answers draw from the same citation pool that feeds Gemini's browser, app, and Workspace responses.

That same week, Perplexity shipped presentation generation for its Comet AI browser. When a user runs a Deep Research session and asks Perplexity to generate slides, the citations from that research get embedded into the deck. Those decks get shared in meetings, emailed to decision-makers, uploaded to Google Drive. Every deck carries the brand citations Perplexity pulled. Every brand absent from Perplexity's citation pool is absent from every one of those shared documents.

These are not edge cases. They are part of a pattern: April 2026 put AI into vehicles, native desktop apps, and shareable documents at the same time that standard GEO measurement tools are still calibrated primarily for browser-based AI search. The gap between what most audits track and where citations actually happen is growing.

What a standard GEO audit covers today

Most AI visibility audits track a version of the same set: ChatGPT web, Perplexity web, Google AI Mode, Google AI Overviews, and sometimes Gemini's browser app. A well-run audit tests a defined set of prompts across these surfaces, records which brands appear, and measures share of voice.

That scope was close to complete in 2025. It is no longer close to complete in April 2026.

The citation surfaces that launched or expanded significantly in the past six weeks alone include two in-vehicle AI assistants, four native desktop AI apps, and a presentation-generation capability that embeds citations into files people distribute inside their organizations. None of these surfaces appear in the default audit scope of any monitoring tool currently on the market.

The vehicle surface

The GM deployment is the largest single generative AI rollout in the automotive industry to date. Four million vehicles over several months, starting with owners who have active OnStar connections and Google accounts. The queries drivers submit are conversational, local-inflected, and often category-level: what's the best restaurant nearby, tell me about that brand I heard on the radio, what's the leading tool for X.

Those queries route through Gemini's citation pool. A brand well-represented in Gemini's citation index will appear. A brand absent from it will not. The vehicle deployment doesn't require any change to how Gemini sources information, which is what makes it worth attention. GEO investment that builds Gemini citation presence today extends automatically to every in-vehicle Gemini query that follows.

ChatGPT arrived in Apple CarPlay in April 2026, available hands-free for iOS 26.4 users in supported vehicles. The query type is similar: voice-driven, often local or informational. CarPlay queries draw from ChatGPT's standard citation pool. Brands in that pool appear. Brands outside it don't.

Neither surface is primarily B2B. But for brands with physical presence, retail distribution, regional events, or any touchpoint that intersects with consumer contexts, both surfaces are live citation surfaces that no current monitoring tool is measuring.

Four desktop AI apps that weren't there a month ago

The browser is no longer the primary interface for AI in professional workflows.

Google launched a native Gemini Mac app on April 24, 2026 as part of the Gemini Drop April 2026 release. It runs as a persistent desktop application, always available without opening a browser. That release also integrated NotebookLM into the Gemini app directly, meaning buyers can now organize research projects in Gemini Notebooks, with cited brands becoming persistent reference points across sessions rather than single-query outputs.

Perplexity launched its Personal Computer for Mac in mid-April. Claude for Enterprise has been available as an enterprise desktop deployment for some time. Microsoft Copilot runs across the entire M365 stack for organizations that have activated it.

Position Digital's April 2026 data puts Copilot referral traffic growth at 25.2x for the year, and Claude at 12.8x. Both are faster-growing AI referral sources than ChatGPT relative to their user base sizes. The growth mechanism is straightforward: both are embedded in the tools professionals already use all day. Copilot in Outlook, Teams, Excel, and Word. Claude in enterprise workflows that developers and analysts have built around it.

The implication for GEO measurement: the April 25 intel from Position Digital suggests that for B2B SaaS brands, the citation optimization priority order may now be Gemini (via Workspace) above Copilot above ChatGPT above Claude. That ordering is the inverse of what most 2025 GEO playbooks recommended. And the tools measuring this shift as four separate surfaces, not as "ChatGPT and some others," don't exist yet at scale.

The shareable document problem

Perplexity's presentation generation is the surface that gets the least attention, and it may be the one with the most compounding effect.

When a user runs a Perplexity Deep Research session and asks Perplexity to generate a presentation from the results, the output is a full slide deck with citations embedded as sources. That deck gets emailed to a VP, shared in a board prep channel, attached to a sales brief. The citations in the deck are treated as authoritative because a research process generated them.

This is a different artifact from a search result. When an AI answer appears in a browser, the person reading it decides whether to accept it, click a link, or run a follow-up query. When citations are embedded in a finished document that gets distributed without further review, the citation set becomes fixed. The brands included in Perplexity's research output appear as the authoritative sources in every copy of that document across every inbox it reaches.

Most GEO thinking frames citations as something AI surfaces to a human who then exercises judgment. The presentation-generation model removes that judgment step from the distribution chain.

Why optimization for one surface doesn't transfer to another

The reason this matters is that AI citation pools don't overlap. Sill's analysis of 139 brands found that 91.6% of cited URLs appear on only one AI platform. Near-zero overlap between the citation sets of different platforms. Moz's 2026 AI Mode research confirms the same pattern from a different angle: 88% of AI Mode citations don't appear in Google's organic top 10 for the same query.

The practical consequence: a brand that has invested heavily in ChatGPT citation optimization is not automatically present in Gemini's citation pool, which feeds both in-vehicle queries and Workspace Studio agents. A brand well-positioned in Perplexity's research citation pool is not automatically embedded in Copilot's Bing-grounded index. Each surface requires its own optimization work.

Sill's study also found that 55% of brands have a 10-point or greater share-of-voice spread between their best and worst AI platform. And 23% of brands score zero across all platforms, with no presence on any measured AI surface. For B2B SaaS companies that have been running SEO programs for years, the absence rate is higher than most marketing teams expect.

The EMGI Group study of 150 SaaS companies found that 44% of brands in Google's top 10 get zero ChatGPT citations. The citation and search rank systems are running largely independent evaluations. The expansion of citation surfaces into vehicles, desktop apps, and shareable documents extends that independence into new contexts.

How to check your coverage across the new surfaces

The audit approach for expanded surfaces is less complicated than it sounds. Most of the new surfaces draw from existing platform citation pools, which means the audit question is: are you present in the right platform pools, not whether you have separate vehicle-specific content.

For Gemini vehicle queries: Check your Gemini citation presence on local and category-level queries. In-vehicle queries are conversational versions of the same questions your buyers ask in browser interfaces. Brands well-represented in Gemini's citation index for category queries appear across all Gemini surfaces, including in-vehicle. The Google AI Overviews and AI Mode optimization work that builds Gemini citation presence applies here.

For Copilot (M365): Copilot draws from Bing's web index. Check your Bing indexation quality, not just Google indexation. Test Copilot-specific brand queries in a Business or Enterprise environment by asking Copilot category-level research questions without naming your brand. Compare the output to ChatGPT's knowledge-only response for the same query. The gap tells you whether your content reaches Bing's index effectively.

For Perplexity presentations: Run a Perplexity Deep Research on a category question in your space and ask it to generate a presentation. See whether your brand appears in the research citations. If it doesn't appear in the base research, it won't appear in the generated deck. The Perplexity citation optimization work (Reddit, LinkedIn, structured content) applies here, because presentation generation draws from the same citation pool as standard Perplexity queries.

For Gemini Notebooks: The Notebooks feature within Gemini app means citation inclusion in research sessions becomes persistent across a buyer's project timeline, not just a one-time query response. Brands that appear in a Gemini research session can become standing references in a project Notebook the buyer maintains for weeks. The audit question is the same: are you in Gemini's citation pool for the queries your buyers run when building a research project in your category.

What to prioritize

The surface expansion doesn't require a parallel expansion of strategy from scratch. The optimization work that builds citation pool membership in each major platform applies across all surfaces that platform powers.

Gemini citation presence: applies to Google AI Overviews, AI Mode, Gemini app, Gemini Mac app, Gemini Notebooks, Google Workspace Studio agents, and now in-vehicle Gemini. One citation pool, multiple surfaces.

Perplexity citation presence: applies to standard Perplexity queries, Perplexity Deep Research, and Perplexity-generated presentations. One citation pool, expanding artifact types.

Copilot citation presence: applies to M365 Copilot across Outlook, Teams, Word, Excel, and PowerPoint. Build it through Bing indexation and the same LinkedIn + structured content approach that drives Bing-grounded responses.

The multi-platform GEO strategy that 91.6% platform specificity demands is unchanged. What's changed is the count of surfaces those pools feed. Every citation pool feeds more surfaces in May 2026 than it did in January 2026. That increases the return on investing in platform-specific citation presence, because each citation investment now extends across a longer surface list automatically.

What changes in practice: the argument for building Gemini citation presence specifically just got substantially stronger. Copilot's 25.2x growth makes the Bing indexation work more valuable than it was. And the presentation generation surface makes Perplexity citation work more compound in its effect: a Perplexity citation now doesn't just produce a web answer, it can become embedded in a document that lives in someone's inbox for months.

FAQ

Does Gemini in-vehicle search use a different citation pool than Google AI Mode?

No. In-vehicle Gemini queries draw from the same Gemini citation pool as Google AI Mode, the Gemini browser app, and Gemini Mac app. The query phrasing differs (voice-driven, often local and conversational), but the underlying source selection is Gemini's standard citation index. Brands already present in Gemini's citation pool for category queries will appear in vehicle responses for those same categories. There is no separate in-vehicle optimization track.

What kinds of queries does ChatGPT handle in CarPlay, and do they affect B2B brands?

CarPlay ChatGPT queries are voice-initiated and skew local or conversational: restaurant recommendations, product category questions, brand follow-ups after a radio mention. Direct B2B SaaS impact is limited. But for B2B brands with consumer touchpoints, physical presence, regional events, or brand awareness campaigns, this is a live citation surface. The bigger point is structural: ChatGPT now runs in more contexts than browser tabs, and that count will continue growing.

How does Perplexity presentation generation change citation strategy?

When Perplexity Deep Research generates a slide deck, the citations embedded in that deck become fixed artifacts distributed inside organizations. A brand absent from Perplexity's citation pool is absent from every presentation those users generate in your category. The optimization approach is the same as standard Perplexity strategy (Reddit coverage, LinkedIn articles, structured content), but the distribution effect is higher because citations get preserved in documents rather than refreshed on each query.

Why is Copilot growing faster than other platforms if its user base is smaller?

Copilot's 25.2x referral traffic growth (Position Digital, April 2026) reflects adoption within an installed base that was already using M365. Enterprise organizations that activate Copilot see it integrated into every email, document, and meeting tool their teams use daily. The growth isn't coming from new user acquisition the way ChatGPT grew. It is coming from existing users shifting research and discovery behavior into a tool already open on their screen. For B2B brands, Copilot's embedded position in professional workflows puts it ahead of its user count in terms of citation priority.

How do I know if my brand appears in Gemini Notebooks?

There is no direct audit tool for Notebook citation persistence yet. The proxy is your Gemini AI Mode citation presence: if your brand appears consistently in Gemini AI Mode responses for category queries, it is likely entering the research sessions that buyers build into Gemini Notebooks. The gap between one-time citation and persistent Notebook reference is primarily a function of citation frequency and content freshness. Brands that appear repeatedly across related queries are more likely to become standing Notebook references than brands that surface on a single query type.

The citation map keeps expanding

The pattern from April 2026 is not a one-time event. It is the direction.

Google Cloud Next 2026 put Gemini into Gmail, Docs, Sheets, Drive, Meet, and Chat via Workspace Studio. ChatGPT Workspace Agents are now running automated research workflows across enterprise sales teams. GM is the first large automotive deployment; other manufacturers follow the same platform partnerships.

Each new surface draws from an existing citation pool. The citation optimization work you do for Gemini today feeds every new surface Gemini expands into. The same is true for Perplexity, ChatGPT, and Copilot. The investment compounds across surfaces, not just across queries.

Only 7.4% of Fortune 500 companies currently implement AI search optimization, according to Searchfy.ai's April 2026 study. That figure understates the gap for B2B SaaS companies, most of which have neither the Fortune 500 brand recognition to carry training data coverage passively nor the enterprise marketing budgets to build it through traditional channels.

The surface map is the argument for starting now. Today's Gemini citation presence isn't just about today's AI Mode query. It is the citation presence that will feed 4 million vehicles, desktop research sessions, and Workspace Studio agents simultaneously, with no additional optimization work required for each new surface. That is a compounding return on a single investment. Most competitors haven't made it yet.