D-ID — Real-World AI Case Studies

Before vs after

Safety & evidence: This is a training simulation, not a clinical diagnostic or treatment system. The case describes anonymized feedback and planned academic validation. No patient-health outcome claim is made.

Sources: D-ID case study; SIU School of Medicine.

D-ID says PepsiCo deployed an AI visual agent for the Gatorade Sports Science Institute. “Anna the Hydration Coach” gives visitors a personalized, interactive way to explore hydration science. D-ID describes evidence-based information, resource recommendations and routes toward forms for meetings and purchases, with strict safety, accuracy and responsible-AI guidelines.

Causality: There is no public quantitative evidence here showing that Anna caused additional sales or leads. The documented fact is the deployment and its intended workflow.

Source: D-ID Visual Agent customer examples.

HARTMANN created Eva, a website-embedded visual agent for healthcare professionals and administrators. D-ID says Eva answers questions about HARTMANN products, recommends resources and directs users to forms for meetings and purchases.

Safety: This is healthcare-product information, not evidence that Eva diagnoses patients or provides clinical treatment advice.

Source: D-ID Visual Agents.

D-ID currently describes Rafael as a defense company using a series of visual agents to explain highly technical military products. Public material reviewed for this database does not disclose agent names, detailed architecture, sensitive capabilities, actions, deployment metrics or conversion results.

Source: D-ID customer examples.

CreatorUp developed an AI visual agent representing CEO Mike Tringe. D-ID describes the agent as a prospect-facing experience that explains CreatorUp's services combining AI capabilities with human creative talent.

Source: D-ID customer examples.

Hakuhodo Technologies, the technology arm of Hakuhodo DY Group, partnered with D-ID for customized streaming-avatar experiences. D-ID documents a Japan-specific pipeline using D-ID Premium+ avatars and streaming API integration with a specialized Japanese TTS model. Initial latency was reduced to under three seconds.

Sources: D-ID case study; Hakuhodo project page; IPRA awards.

My Last Will built “My,” an AI-powered conversational video assistant for digital legacy planning. D-ID says the system adds a human, conversational layer to a process previously based on static text, forms and limited video.

Source: D-ID case study.

Envy, a B2B tech marketing and RevOps agency, created a D-ID digital twin of managing partner Billy Cina for event follow-up, newsletter communication and sales outreach.

Causality: The case records results after avatar use and attributes them to the campaign experience, but it does not establish a controlled causal experiment isolating D-ID from message, audience or campaign effects.

Source: D-ID case study.

Pitango used D-ID's Creative Reality Studio to animate an AI-generated avatar named Sam for its “The Juice” newsletter, aimed at an audience of more than 10,000. The workflow included avatar design, D-ID generation and human video editing with transitions, subtitles and background imagery.

D-ID reports a notable spike in newsletter open rates and website traffic, tracked through Google Analytics, but exact figures were not published. Pitango's Head of Marketing said the team planned to reuse Sam and other avatars.

Source: D-ID case study.

Gameloft's Creative Agency Team in Sofia used D-ID Studio to turn static game-character images into speaking-character short-form user-acquisition ads. The workflow was image → D-ID facial animation/text-to-speech → conventional video editing → distribution through Facebook, Google Ads and sometimes YouTube.

Source: D-ID case study.

The Alzheimer's Foundation of America introduced Allison, an AI-powered virtual helpline assistant developed by Bold Crow AI using D-ID's avatar API. AFA's own public announcement says D-ID powers the streaming animation while Bold Crow AI programmed the assistant.

Sources: D-ID case study; Bold Crow AI project page; D-ID announcement; AFA public announcement.

Diplomat Group's global Learning & Development team used D-ID Creative Reality Studio and its PowerPoint plug-in to bring video production in-house. The company had relied on external video vendors and wanted faster, lower-friction content creation for training and communications.

Source: D-ID case study.

German social-impact organization SPIN Sport Innovation integrated D-ID digital people into online capacity-building courses in sports, health and education. The avatars introduced or accompanied course content and reduced the need for staff travel to record content.

Source: D-ID case study.

SingIt, an EdTech company with a reported 150,000 users in Israel, integrated D-ID's API to create face-to-face AI English teachers. The avatars supplied audio and visual feedback in real time.

Source: D-ID case study.

Mexican broadcaster Radio Fórmula created three AI-generated newscasters — Nat, Sofi and Max — using D-ID Creative Reality Studio. Journalists and producers wrote the scripts; B-roll and graphics were added; the segments were distributed through the video-streaming channel and repurposed for social media. D-ID also describes a secondary HR onboarding/training use.

Outcome: D-ID reports a month-on-month increase in viewership, particularly among younger audiences that had been reducing news consumption. No exact percentage is published.

Human + AI: Humans remained responsible for journalism, scripts, production and editorial content; D-ID handled avatar generation.

Source: D-ID case study.

Vantage Innovation Lab and SOMOS, a Mexican food brand, partnered with D-ID and ElevenLabs to create a real-time streaming avatar of SOMOS co-founder Miguel Leal's mother, Veronica Villarreal. The experience allowed users to access her culinary guidance through the website and WhatsApp.

Results: The reviewed primary announcement documents the experience and access channels but does not publish robust conversion, sales or ROI metrics.

Source: D-ID announcement.

Defined.ai, Microsoft and Daredata Engineering worked with Portugal's Agency for Modernization in a government-sponsored pilot to improve digital interactions. D-ID technology supplied lifelike avatar interaction. The project sat within the Accelerat.ai initiative under Portugal's Recovery and Resilience Plan.

Architecture: the broader project included speech-to-text, natural-language understanding, dialogue management, natural-language generation and text-to-speech. The D-ID contribution should not be treated as responsible for the entire stack.

Result: Pilot/deployment documented; quantitative user or service outcomes not disclosed in the reviewed D-ID case.

Source: D-ID case study.

Convo.ai integrated D-ID's streaming API to make user-created AI characters visually and vocally conversational. D-ID's case describes a mobile app with customizable characters, facial expressions and more than 150 AI voices; ElevenLabs supplied voices and language support.

Reported scale at case publication: approximately 1.6K App Store downloads. D-ID says a significant portion of subscribers paid for D-ID-generated characters. This is a company case-study claim, not an independently audited subscription metric.

Source: D-ID case study.

chat.D-ID was a web app combining D-ID's real-time text-to-video streaming technology with OpenAI's ChatGPT so users could converse face-to-face with a photorealistic AI avatar. It was launched in beta and presented as a demonstration of D-ID's streaming technology.

Current status: The 2023 launch should not be treated as evidence that chat.D-ID remains a current standalone product. Current D-ID materials instead emphasize Visual Agents and V4 Expressive Visual Agents.

Source: D-ID launch announcement.

D-ID launched a real-time streaming/chat API for face-to-face conversations with AI digital humans. The announcement described integration with LLMs such as GPT-3 and LaMDA and potential uses in customer support, sales, training and personal finance.

Evidence boundary: These were product capabilities and intended use cases. They are not customer outcome claims. Current implementation details should be checked against D-ID's current API documentation rather than assuming the 2023 architecture remains unchanged.

Source: D-ID launch announcement.

D-ID ran controlled-style tests of personalized video email campaigns against non-personalized formats. The tests targeted less-active contacts and measured open/engagement, CTR and coupon redemption.

Classification: This is not a customer case. It is a D-ID pilot using its own marketing technology, so the findings should be treated as company-reported internal evidence.

Source: D-ID pilot case.

D-ID compared a personalized Christmas video email with a non-personalized version. The personalized group received a first-name greeting and 20 Studio credits; the comparison group received a generic greeting and the same incentive.

Reported result: D-ID states the personalized email achieved a 300% higher conversion rate. The reviewed case does not provide enough experimental detail to generalize that result to all campaigns or isolate every causal variable.

Source: D-ID case study.

D-ID announced Agentic Videos as a capability that turns traditional one-way video into interactive experiences in which a visual AI agent can continue a conversation after the video. D-ID describes knowledge, memory, a digital-human interface and agentic capabilities, plus analytics such as conversation volume, engagement time, question topics and sentiment.

Evidence boundary: The launch demonstrates a product capability. It should not be converted into claims about customer performance until individual deployments disclose results.

Source: D-ID launch announcement.

D-ID launched V4 Expressive Visual Agents for real-time, LLM-connected conversations. D-ID reports sub-0.5-second conversational turns, up to 4K output, sentiment-aware facial expression, optional camera/vision, inline media/UI and MCP-app integrations. D-ID's technical page reports a June 2026 SyncNet benchmark and other performance figures; those are vendor-reported product benchmarks, not customer outcomes.

Workflow: avatar + voice + behavior/personality + knowledge/RAG + LLM + optional webhooks/API actions + WebRTC session + conversation logs/analytics.

Source: D-ID V4 announcement; D-ID V4 technical specifications.