{"id":35886,"date":"2026-08-23T05:03:36","date_gmt":"2026-08-23T05:03:36","guid":{"rendered":"https:\/\/www.sozee.ai\/resources\/create-real-time-virtual-influencer\/"},"modified":"2026-09-02T12:12:55","modified_gmt":"2026-09-02T12:12:55","slug":"create-real-time-virtual-influencer","status":"publish","type":"post","link":"https:\/\/www.sozee.ai\/resources\/create-real-time-virtual-influencer\/","title":{"rendered":"How to Create Real-Time Virtual Influencer Videos in 2026"},"content":{"rendered":"<h2 id=\"key-takeaways\">Key Takeaways<\/h2>\n<ul>\n<li>Real-time virtual influencer streaming in 2026 depends on locked likeness, reusable environments, and an LLM memory layer that removes daily re-prompting and face drift.<\/li>\n<li>Sozee Live Mode lets creators go live in under two hours using a webcam or phone, three reference photos, and an OBS account.<\/li>\n<li>Reusable assets such as environments, outfits, and objects gain value over time, speeding up each shoot and keeping visuals consistent across sessions.<\/li>\n<li>Built-in OBS integration, human-in-loop moderation, and FTC-compliant disclosure tools support multi-platform streaming with full regulatory compliance.<\/li>\n<li>Ready to scale your live virtual influencer? <a href=\"https:\/\/app.sozee.ai\/sign-up\" target=\"_blank\"><strong>Sign up for Sozee now<\/strong><\/a> and launch your first stream today.<\/li>\n<\/ul>\n<h2>Step 1: Cast a Character with a Locked, Trustworthy Likeness<\/h2>\n<p>The foundation of any real-time virtual influencer is a stable, locked identity. Sozee offers two paths. Upload three reference photos and Sozee reconstructs the likeness with hyper-realistic accuracy, keeping the same face and body in every frame. Alternatively, use the AI Character Builder to define origin, ethnicity, skin, eyes, hair, physique, and any distinctive detail that must appear in every generation. This process produces a face that has never existed yet remains consistent from the first frame forward.<\/p>\n<p>Locked likeness is not a cosmetic feature. It forms the basis of audience trust and engagement. <a href=\"https:\/\/vinfluencer.ai\/virtual-influencer-trends-2026\" target=\"_blank\" rel=\"noindex nofollow\">Reported human-likeness scores from blind audience tests have improved in recent years<\/a>, and <a href=\"https:\/\/vinfluencer.ai\/virtual-influencer-trends-2026\" target=\"_blank\" rel=\"noindex nofollow\">average engagement on virtual influencer posts is often higher than for human creators at similar follower counts<\/a>. That performance advantage exists because audiences recognize the same character every time and build familiarity that drives repeat engagement.<\/p>\n<blockquote>\n<p><strong>Pro Tip:<\/strong> Maintain a character bible that documents face, hair, wardrobe palette, camera style, speaking style, values, and topics to avoid. <a href=\"https:\/\/kling.ai\/blog\/virtual-influencers-examples-creation-guide\" target=\"_blank\" rel=\"noindex nofollow\">Review each post against that system before publishing to prevent drift.<\/a> A useful starting reference set includes a front-facing portrait, three-quarter facial view, side profile, neutral full-body image, one or two typical outfits, and a few natural expressions.<\/p>\n<\/blockquote>\n<h2>Step 2: Build Reusable Rooms, Wardrobes, and Props<\/h2>\n<p>Every element of a Sozee shoot becomes a permanent asset and removes the daily setup tax that makes many tools hard to use at scale. Settings are built from up to four reference shots so the room stays the same from stream to stream. You can build a bedroom once and shoot in it for a year. Outfits assemble from one piece per category such as tops, bottoms, shoes, and accessories. Objects like a handbag, a latte, or a phone are saved to a library and attached to any future shoot. The @ operator drops any saved element into the prompt inline, keeping the workflow fast and predictable.<\/p>\n<p>The compounding effect is the point. <a href=\"https:\/\/synthetic.actor\/blog\/ai-ugc-workflow-brand-ready-assets\" target=\"_blank\" rel=\"noindex nofollow\">A consistent AI creator workflow uses recurring rooms, objects, wardrobe, camera style, and context saved as reusable presets so the creator remains recognizable across outputs<\/a>, and every shoot makes the next one faster.<\/p>\n<blockquote>\n<p><strong>Pro Tip:<\/strong> Export 1080p 15-second clips from your saved environments as teaser content. Sozee supports video output up to 1080p and up to 15 seconds in every major aspect ratio. These clips work immediately as platform teasers without extra editing.<\/p>\n<\/blockquote>\n<h2>Step 3: Configure Real-Time Live Mode for Low-Latency Performance<\/h2>\n<p>Live Mode renders the character onto a webcam or phone feed in real time. The creator acts and the character performs. <a href=\"https:\/\/khaby.ai\/blog\/real-time-ai-avatars-technology\" target=\"_blank\" rel=\"noindex nofollow\">Real-time AI avatar systems target low end-to-end latency<\/a>, which is allocated across speech recognition, LLM response, TTS synthesis, and neural rendering. Sozee Live Mode stays within that latency budget and avoids the lag that makes competing tools feel robotic.<\/p>\n<p>Sozee supports a full SFW-to-NSFW ramp within platform guidelines. The creator sets the pacing and ceiling, not the platform. Agencies gain precise control over content tiers across different streaming destinations.<\/p>\n<blockquote>\n<p><strong>Common Pitfalls:<\/strong> Prompt-only workflows, where a new text prompt is submitted for each session, are the primary cause of face drift. <a href=\"https:\/\/tldr.takara.ai\/p\/2606.22905\" target=\"_blank\" rel=\"noindex nofollow\">Existing diffusion-based models struggle to maintain visual temporal consistency during long streaming sessions<\/a>. Lock the likeness once in Sozee and avoid re-prompting for identity.<\/p>\n<\/blockquote>\n<h2>Step 4: Add an LLM Brain and a Session Memory Layer<\/h2>\n<p>A live virtual influencer that cannot respond to chat behaves like a video loop, not a stream. Sozee integrates a conversational LLM layer that drives automatic reactions such as responses to viewer comments, nods, laughter, and topic pivots. The creator does not need to script every exchange manually.<\/p>\n<p>The recommended architecture for this layer is a cascaded streaming pipeline. <a href=\"https:\/\/spatius.ai\/blog\/best-ai-avatar-apis-byo-llm-tts-2026\" target=\"_blank\" rel=\"noindex nofollow\">User audio is transcribed by ASR, the transcript is sent to an LLM that streams tokens via SSE, the LLM output is chunked into TTS for streaming audio, and the resulting audio drives the avatar renderer<\/a>. Sozee handles this orchestration natively. <a href=\"https:\/\/forasoft.com\/blog\/article\/interactive-ai-avatar-development\" target=\"_blank\" rel=\"noindex nofollow\">Streaming every pipeline stage, so partial ASR output triggers LLM token generation, which triggers partial TTS and immediate avatar frame rendering, is required to stay under the 1-second first-frame budget<\/a>. Sequential batch processing adds several seconds of latency and breaks conversational flow.<\/p>\n<p>The memory layer keeps track of context across a session, including viewer names, topics discussed, and products mentioned. The stream feels coherent rather than stateless and viewers feel recognized.<\/p>\n<h2>Step 5: Connect OBS for Multi-Platform Streaming and Compliance<\/h2>\n<p>Sozee Live Mode outputs a virtual camera feed that OBS picks up natively. No plugins and no encoding workarounds are required. From OBS, the stream routes to any RTMP destination such as Twitch, YouTube Live, TikTok Live, Kick, or a private platform.<\/p>\n<p>Human-in-loop moderation runs in parallel. A moderator monitors the chat queue and can flag, delay, or redirect viewer inputs before they reach the LLM layer. <a href=\"https:\/\/glaad.org\/smsi\/2026\/key-findings-and-recommendations\" target=\"_blank\" rel=\"noindex nofollow\">AI systems should be used to flag content for human review rather than for automated removals<\/a>. Sozee\u2019s moderation architecture follows this principle and keeps a human decision point in the loop for sensitive inputs.<\/p>\n<p>Compliance obligations apply to every live stream. <a href=\"https:\/\/blog.promise.legal\/ftc-endorsement-disclosure-rules-streamers\" target=\"_blank\" rel=\"noindex nofollow\">The FTC recommends clear and conspicuous disclosures for sponsored content including persistent visual overlays and periodic verbal callouts<\/a> to satisfy the \u201cclear and conspicuous\u201d standard. <a href=\"https:\/\/influencers-time.com\/ai-likeness-rules-2026-disclosure-guide-for-marketers\" target=\"_blank\" rel=\"noindex nofollow\">For livestreams using a synthetic host, disclosure must be visible or audible so that viewers joining midstream still encounter it<\/a>. A label such as \u201cAI-generated likeness. Paid partnership.\u201d should appear on-screen throughout. <a href=\"https:\/\/accordshield.com\/blog-influencer-marketing-ftc-creator-contracts-2026\" target=\"_blank\" rel=\"noindex nofollow\">New York\u2019s synthetic-performer advertising disclosure law, effective June 9, 2026, requires any person with actual knowledge that a synthetic performer appears in an advertisement to make a conspicuous disclosure<\/a>, with first violations carrying a $1,000 civil penalty.<\/p>\n<h2>Step 6: Schedule Content, Clip Streams, and Monetize<\/h2>\n<p>Sozee Scheduler connects Instagram, TikTok, X, Facebook, Reddit, and Fanvue on a per-character basis. Photos, carousels, reels, and stories publish with a caption per platform and a live preview of the real post before it goes out.<\/p>\n<p>Clips from Live Mode sessions land in the Vault automatically. From there, they feed the Scheduler directly. A 90-second highlight from a two-hour stream becomes a scheduled reel within minutes. Voice Notes let the character respond to fan messages in her own cloned voice without the creator recording anything.<\/p>\n<p>Monetization paths include subscription platforms such as Fanvue and OnlyFans, brand sponsorships with FTC-compliant disclosure overlays, and paid live events. <a href=\"https:\/\/vinfluencer.ai\/virtual-influencer-trends-2026\" target=\"_blank\" rel=\"noindex nofollow\">Annual brand spending on virtual personas has crossed $1.3 billion<\/a>, and <a href=\"https:\/\/vinfluencer.ai\/virtual-influencer-trends-2026\" target=\"_blank\" rel=\"noindex nofollow\">roughly a third of Gen Z report having bought something promoted by a digital persona<\/a>.<\/p>\n<p><a href=\"https:\/\/app.sozee.ai\/sign-up\" target=\"_blank\"><strong>Start creating now \u2014 build your first live virtual influencer on Sozee.<\/strong><\/a><\/p>\n<h2>Step 7: Measure Success with Analytics and Benchmarks<\/h2>\n<p>Once revenue channels are live, the next priority is understanding which ones perform. Sozee Analytics reports impressions, reach, likes, comments, shares, and engagement, and splits what Sozee posted from what the creator posted manually. That split shows exactly what the platform contributes to the business.<\/p>\n<p>These target benchmarks define a mature Sozee Live Mode setup and signal that the system can support daily operations without constant manual fixes:<\/p>\n<ul>\n<li>Streams running daily with under 200 ms avatar render latency<\/li>\n<li>Consistent character identity across 30 or more days without re-prompting<\/li>\n<li>Engagement rates tracking toward the <a href=\"https:\/\/vinfluencer.ai\/virtual-influencer-trends-2026\" target=\"_blank\" rel=\"noindex nofollow\">virtual influencer benchmark<\/a>, versus human-creator averages<\/li>\n<li>Measurable revenue lift from scheduled clip repurposing and sponsored stream segments<\/li>\n<\/ul>\n<p>Once you hit these benchmarks consistently for 30 days, advanced capabilities unlock additional revenue and scale. Advanced next steps for agencies and scaling creators include the following moves:<\/p>\n<ul>\n<li>Scale to agency workspaces with one login and every client fully isolated with its own characters, vault, connected accounts, and credits<\/li>\n<li>Add voice cloning for multilingual streams, using Sozee\u2019s voice cloning layer to reach audiences in additional languages without re-casting the character<\/li>\n<li>Test paid live events on Fanvue or private RTMP destinations to establish direct revenue from the live format<\/li>\n<\/ul>\n<h2>Comparison: HeyGen LiveAvatar vs. D-ID Agents vs. Sozee Live Mode<\/h2>\n<p>If you are evaluating real-time avatar platforms, the table below focuses on four capabilities that determine whether a system can sustain daily live streaming without face drift or daily re-prompting.<\/p>\n<table>\n<thead>\n<tr>\n<th>Feature<\/th>\n<th>HeyGen LiveAvatar<\/th>\n<th>D-ID Agents<\/th>\n<th>Sozee Live Mode<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Locked likeness (no drift)<\/td>\n<td>No, <a href=\"https:\/\/forasoft.com\/learn\/ai-for-video-engineering\/articles-ai\/real-time-avatar-lipsync-in-call\" target=\"_blank\" rel=\"noindex nofollow\">prompt-driven sessions produce face drift across days<\/a><\/td>\n<td>No, <a href=\"https:\/\/spatius.ai\/blog\/best-ai-avatar-apis-mobile-apps-2026\" target=\"_blank\" rel=\"noindex nofollow\">photo-to-video animation prone to lip-sync degradation on unstable networks<\/a><\/td>\n<td>Yes, likeness locked at character creation and reused across every session<\/td>\n<\/tr>\n<tr>\n<td>Reusable environments and outfits<\/td>\n<td>No, assets must be re-uploaded or re-prompted per session<\/td>\n<td>No, no native asset library for environments or wardrobe<\/td>\n<td>Yes, settings, outfits, and objects saved once and reused indefinitely<\/td>\n<\/tr>\n<tr>\n<td>Native scheduling and analytics<\/td>\n<td>No, requires third-party scheduling tools<\/td>\n<td>No, no native scheduling or cross-platform analytics<\/td>\n<td>Yes, Sozee Scheduler and Analytics built in, with Sozee-vs-manual post split<\/td>\n<\/tr>\n<tr>\n<td>Typical avatar render latency<\/td>\n<td><a href=\"https:\/\/forasoft.com\/learn\/ai-for-video-engineering\/articles-ai\/real-time-avatar-lipsync-in-call\" target=\"_blank\" rel=\"noindex nofollow\">First-frame latency over WebRTC<\/a><\/td>\n<td><a href=\"https:\/\/www.computerweekly.com\/blog\/CW-Developer-Network\/Inside-D-IDs-real-time-AI-avatar-technology\" target=\"_blank\" rel=\"noindex nofollow\">Works at sub-200 ms latency while generating video at 100 FPS<\/a><\/td>\n<td>Under 200 ms avatar render target within the full pipeline<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>Note: HeyGen and D-ID latency figures reflect published benchmarks for their respective streaming APIs. Sozee Live Mode latency reflects the platform\u2019s internal render target. Full end-to-end latency, including ASR, LLM, and TTS stages, follows the <a href=\"https:\/\/forasoft.com\/blog\/article\/interactive-ai-avatar-development\" target=\"_blank\" rel=\"noindex nofollow\">standard five-stage pipeline budget of approximately 900 ms time-to-first-frame<\/a> when all stages run in streaming mode.<\/p>\n<h2>Success Metrics and Advanced Next Steps<\/h2>\n<p>A Sozee Live Mode deployment running at full capacity hits all four benchmarks listed above while tracking toward the virtual influencer category engagement standard. <a href=\"https:\/\/vinfluencer.ai\/virtual-influencer-trends-2026\" target=\"_blank\" rel=\"noindex nofollow\">The global virtual influencer market is projected to grow substantially in coming years with long-range forecasts pointing to major increases<\/a>. Infrastructure built today compounds in value as the market grows.<\/p>\n<p>For teams ready to move beyond a single character and single stream, the advanced path runs through three areas. Agency workspaces in Sozee isolate each client\u2019s characters, vault, connected accounts, and credits under one login. Voice cloning extends the character into multilingual streams without re-casting. Paid live events, such as ticketed streams on Fanvue or private RTMP destinations, establish a direct revenue line from the live format that does not depend on platform algorithm distribution.<\/p>\n<h2>Frequently Asked Questions<\/h2>\n<h3>How can I create a virtual influencer?<\/h3>\n<p>Creating a virtual influencer requires three components: a stable visual identity, a content production system, and a publishing workflow. On Sozee, the process starts by uploading three reference photos or using the AI Character Builder to generate an original face. The platform locks the likeness at creation so it never drifts between sessions. From there, environments, outfits, and objects are built once and saved as reusable assets. Content such as photos, video, live streams, and voice notes is produced through Sozee\u2019s direction controls and published via the built-in Scheduler to Instagram, TikTok, X, Facebook, Reddit, and Fanvue. The full setup from character creation to first published post can be completed in a single afternoon. Try Sozee to build your first virtual influencer today.<\/p>\n<h3>Will AI replace influencers?<\/h3>\n<p>AI is not replacing human influencers. It removes the production ceiling that limits them. Human creators face a structural constraint because demand for content outpaces what any individual can physically produce. Virtual influencers and AI-assisted content systems extend a creator\u2019s output without requiring additional time or physical presence. The most effective model in 2026 is a hybrid. A human creator directs the character, sets the brand voice, and manages the business, while AI handles the production volume. Sozee is built for this model. The creator directs and the platform produces. Sign up to see how far that extends your output.<\/p>\n<h3>How much do virtual influencers earn?<\/h3>\n<p>Virtual influencer earnings vary widely by audience size, niche, and monetization model. Brand sponsorships are the primary revenue channel for most virtual influencers, with annual brand spending on virtual personas having crossed $1.3 billion globally. Subscription platforms such as Fanvue and OnlyFans provide recurring revenue independent of brand deals. Paid live events, affiliate commissions, and digital product sales add additional layers. The engagement advantage noted earlier, where virtual influencers outperform human creators at similar follower counts, translates directly into higher sponsorship rates. Sozee\u2019s Analytics dashboard tracks impressions, reach, and engagement per character so revenue potential is measurable from day one. Start building your revenue-generating virtual influencer on Sozee.<\/p>\n<h3>What tools support real-time AI avatars in 2026?<\/h3>\n<p>Several platforms offer real-time avatar capabilities in 2026, but they differ significantly in what they lock and what they leave to daily re-prompting. HeyGen LiveAvatar streams avatars over WebRTC but produces face drift across sessions because identity is not locked at the asset level. D-ID Agents provide REST and WebRTC endpoints for photo-to-video animation but have no native environment or wardrobe library and are prone to lip-sync degradation on unstable networks. Tavus Phoenix-4 targets sub-600 ms first-frame latency for conversational use cases. Sozee Live Mode is the only platform that combines locked likeness, reusable environments and outfits, a native LLM memory layer, OBS integration, built-in scheduling, and cross-platform analytics in a single workflow. Try Sozee to compare the difference directly.<\/p>\n<h3>How do I moderate live AI content?<\/h3>\n<p>Effective moderation for live AI content operates at three levels: platform controls, human review, and legal compliance. At the platform level, Sozee supports SFW-to-NSFW content ramps with the ceiling set by the creator, keeping content within chosen guidelines automatically. At the human review level, a moderator monitors the chat queue and can flag or redirect viewer inputs before they reach the LLM layer. AI flags content for human decision rather than triggering automated removals. At the legal level, FTC rules require that sponsored content disclosures are clear and conspicuous, which for live streams can include persistent visual overlays and periodic verbal callouts so viewers joining mid-stream encounter them. New York\u2019s synthetic-performer disclosure law, effective June 2026, adds a separate obligation to disclose the AI-generated nature of the on-screen performer. Sozee\u2019s overlay tools support persistent on-screen disclosure labels that satisfy these requirements. Sign up to access Sozee\u2019s full moderation and compliance toolkit.<\/p>\n<h3>Is Sozee Live Mode uncensored?<\/h3>\n<p>Sozee supports a full SFW-to-NSFW content pipeline, with the pacing and ceiling controlled by the creator rather than the platform. This control means creators can set exactly where on the content spectrum each stream or asset sits, including adult content for appropriate platforms, without the platform overriding that decision. All content must comply with applicable laws and the terms of service of the destination platform. Sozee\u2019s compliance tools, including disclosure overlays and content-tier controls, are built to support that requirement. The character\u2019s likeness, voice, and world remain locked regardless of content tier, so consistency is maintained across the full range of output. Try Sozee to explore the full range of content controls available.<\/p>\n<h2>Conclusion<\/h2>\n<p>The seven-step workflow above, which covers casting with locked likeness, building reusable assets, configuring Live Mode, adding an LLM memory layer, integrating OBS, scheduling and monetizing, then measuring, forms a complete pipeline for consistent real-time virtual influencer streaming in 2026. Competing approaches often require daily re-prompting, produce face drift, or lack the asset infrastructure needed to sustain a daily stream. Sozee Live Mode closes all three gaps at once with locked likeness that never drifts, reusable environments and outfits that compound in value over time, and directable controls that keep the creator in charge of every frame. The global virtual influencer market is growing quickly. The creators who build consistent, scalable live stream infrastructure now will own the category.<\/p>\n<p><a href=\"https:\/\/app.sozee.ai\/sign-up\" target=\"_blank\"><strong>Build your locked-likeness character and launch your first live stream on Sozee today.<\/strong><\/a><\/p>\n<section data-read-next=\"true\">\n<h2>Read Next<\/h2>\n<ul>\n<li><a href=\"https:\/\/sozee.ai\/resources\/best-virtual-influencer-creator-software\" target=\"_blank\">Best Virtual Influencer Software for Content Creators 2026<\/a><\/li>\n<li><a href=\"https:\/\/sozee.ai\/resources\/ai-video-tools-virtual-influencers\" target=\"_blank\">How to Build Virtual Influencers Who Post Daily<\/a><\/li>\n<li><a href=\"https:\/\/sozee.ai\/resources\/virtual-influencer-systems-beyond-onlyfans\" target=\"_blank\">Virtual Influencer Content Creation Systems Beyond OnlyFans<\/a><\/li>\n<li><a href=\"https:\/\/sozee.ai\/resources\/hyperrealistic-virtual-content-creators\" target=\"_blank\">Create Hyper Realistic Virtual Influencers with AI<\/a><\/li>\n<li><a href=\"https:\/\/sozee.ai\/resources\/best-virtual-influencer-ai-tools\" target=\"_blank\">Best Virtual Influencer AI Tools for Creators in 2026<\/a><\/li>\n<\/ul>\n<\/section>\n","protected":false},"excerpt":{"rendered":"<p>Build and stream AI virtual influencer content live in 2026. Sozee&#8217;s Live Mode delivers low-latency, monetizable video. Start creating today.<\/p>\n","protected":false},"author":2,"featured_media":35885,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[3,7],"tags":[],"class_list":["post-35886","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-ai-influencers","category-ai-video"],"_links":{"self":[{"href":"https:\/\/www.sozee.ai\/resources\/wp-json\/wp\/v2\/posts\/35886","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.sozee.ai\/resources\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.sozee.ai\/resources\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.sozee.ai\/resources\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.sozee.ai\/resources\/wp-json\/wp\/v2\/comments?post=35886"}],"version-history":[{"count":1,"href":"https:\/\/www.sozee.ai\/resources\/wp-json\/wp\/v2\/posts\/35886\/revisions"}],"predecessor-version":[{"id":41799,"href":"https:\/\/www.sozee.ai\/resources\/wp-json\/wp\/v2\/posts\/35886\/revisions\/41799"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.sozee.ai\/resources\/wp-json\/wp\/v2\/media\/35885"}],"wp:attachment":[{"href":"https:\/\/www.sozee.ai\/resources\/wp-json\/wp\/v2\/media?parent=35886"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.sozee.ai\/resources\/wp-json\/wp\/v2\/categories?post=35886"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.sozee.ai\/resources\/wp-json\/wp\/v2\/tags?post=35886"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}