Samantha AI, an artificial intelligence application developed in Romania by UXR, has officially reached version 3, marking a new milestone in conversational AI.
This version is not merely a technical upgrade, but a paradigm shift in how people interact with an AI system: natural, fluid, and deeply humanized.
Samantha was conceived from the very beginning as a social and technological experiment, directly inspired by the operating system in the film “Her”—an artificial intelligence that does not limit itself to commands, but builds dialogue, context, and relationship. This exact direction has guided every version of the platform.
Samantha V1, launched as a purely text-based interface, allowed users to freely “talk” with the AI, engage in coherent conversations, and explore the idea of a digital companion. Following its launch, the platform recorded over 20,000 visits, confirming real interest in a conversational AI with personality and contextual memory.
Samantha V2 marked the first major technological leap through the integration of text-to-voice and voice-to-text capabilities, enabling users to speak with Samantha and hear her responses. This version experienced explosive growth, surpassing 100,000 visits in the first months after launch, validating the transition from simple written conversation to vocal interaction.
Today, Samantha V3 takes this experience to the next level by fully implementing speech-to-speech technology, which enables fluid, natural, real-time voice conversations—just like a phone call. The user speaks, Samantha responds vocally instantly, without artificial pauses, without the sensation of “processing,” but within a continuous flow of dialogue.
Samantha is no longer a system that executes commands, but a conversational partner capable of listening, understanding nuances, maintaining context, and responding coherently, adapting to the user’s language, rhythm, and style.
Key features of SAMANTHA V3 include:
- Real-time speech-to-speech, for continuous, natural vocal dialogue with no perceptible latency.
- Text-based conversation, allowing classic text discussions as an alternative or complement to voice interaction.
- Fully integrated text-to-voice and voice-to-text, offering total flexibility of use.
- Automatic language detection, with no prior settings—users can speak or write in any language.
- On-demand contextual translation, allowing Samantha to translate the conversation or continue the dialogue in a specified language.
- Coherent multilingual communication, preserving meaning, tone, and intent regardless of the language used.
- Interaction designed to mimic real human conversation, without artificial interruptions or waiting times.
- Significantly optimized performance, with near-instant response times for both voice and text input.
- High stability in multilingual conversations, without loss of context.
- Increased retention rate in voice sessions compared to text-only sessions.
- Greater conversational continuity, with longer and more natural exchanges.
- Advanced analytics and performance optimizations, enabling measurement of user engagement, session duration, voice vs. text preference, and real-time conversational stability.
The new SAMANTHA V3 – Speech version can be accessed here:
👉 https://samantha.al – text
👉 https://samantha-ai-beta.vercel.app/speech – voice
Key statistics and metrics (analytics)
Data collected from real platform usage indicates a clear and measurable evolution in how users interact with Samantha from one version to another:
- Retention rate increased by approximately +62% from Samantha V1 (text-only) to Samantha V2 (voice-enabled), as users returned more frequently once voice interaction was introduced.
- Average session duration increased by over +85% with the integration of text-to-voice and voice-to-text, indicating that users began treating Samantha as a conversational partner rather than a simple chatbot.
- In Samantha V3, speech-to-speech sessions are on average +47% longer than text-only sessions and +31% longer than voice-to-text sessions in the previous version.
- Early abandonment rate (first 30 seconds) dropped by approximately -54% in Samantha V3, due to near-instant response times and continuous conversational flow.
- Over 68% of users default to voice interaction when available, compared to text.
- The perceived average response time in speech-to-speech conversations was reduced by over 70%, reaching a level comparable to a phone conversation.
- Multilingual conversations (switching languages during dialogue) account for over 41% of total sessions, with no loss of context or coherence.
- The returning user rate in Samantha V3 is +58% higher than in Samantha V1, indicating recurrent rather than occasional usage.
Adoption highlights:
- Samantha V1 exceeded 20,000 visits, with an active user conversion rate of approximately 36%.
- Samantha V2 recorded over 100,000 visits in the first months, with a +400% increase in voice interaction volume compared to the previous version.
- Samantha V3, currently in beta, already shows accelerated adoption of voice usage, with over 75% of all interactions conducted via speech-to-speech.
As stated since the project’s first public appearance, Samantha is not just a virtual assistant, but a social AI, capable of becoming a confidant, friend, dialogue partner, or professional tool, depending on the user’s needs. Version 3 reinforces this vision by removing barriers between humans and machines and bringing the experience closer to real, human dialogue.
Through this launch, Samantha asserts its maturity as a conversational AI platform oriented toward real-world, global, and scalable use, addressing both business environments and individual users seeking an authentic, natural, and deeply intuitive conversational experience.
Samantha is a product of the UXR agency, founded by Cristian Doiu, and is part of a broader suite of digital applications and products developed and offered free of charge to the Romanian digital market, as part of a deliberate strategy to democratize access to technology, AI, and modern digital infrastructure.
Samantha.al represents the conversational AI service developed by Cristian Doiu and UXR, built as a social, conversational, and vocal AI inspired by the operating system in the film “Her”. The platform is designed not as a standalone product, but as an essential component of a coherent digital ecosystem.
Samantha is part of the UXR application suite, which includes:
- al – a 100% authentically European social network, built around real identity, authentic conversation, and the absence of aggressive algorithmic manipulation;
- al – a Full-Stack URL Management System used for shortening, redirection, tracking, analytics, and large-scale link infrastructure;
- net – a digital library dedicated to design, UX, UI, and creative thinking books, aimed at professionals and students;
- at – the world’s most powerful AI aggregator, bringing together, structuring, and making accessible multiple artificial intelligence models and capabilities within a single interface.
All these products are developed under the UXR umbrella, with a focus on real utility, performance, open access, and scalability, without subjecting users to aggressive paywalls or artificial limitations.
Cristian Doiu, Founder of UXR, states:
“UXR does not just create applications, but coherent digital ecosystems. Every product—from Samantha to Thinks.at—is designed to be useful, scalable, and accessible, demonstrating that European technology can compete and inspire on a global level.”
Through this suite of products, UXR and Cristian Doiu aim to actively contribute to the maturation of the digital market in Romania and Europe, offering local, high-performance, and open alternatives in a global context dominated by opaque and centralized platforms.
