Pluto AI
Magyar Telekom, the Hungarian division of Deutsche Telekom, has introduced Pluto AI, a cutting-edge internal AI platform, in a daring attempt to capitalize on the revolutionary potential of artificial intelligence (AI). This project is a big step towards the company’s eventual goal of integrating AI into all aspects of its business operations and empowering all employees to take advantage of the enormous potential of AI.
Magyar Telekom confronted similar problems, such as personnel with differing degrees of AI comprehension and a lack of easily available tools for experimentation and practical application, after realizing that AI expertise is no longer a luxury but rather a must for future success. In order to address this, the business developed a scalable system that could serve a wide range of use cases and adjust to changing AI demands, democratizing AI expertise and encouraging experimentation.
This requirement led to the creation of Pluto AI, which at first concentrated on providing business teams with a straightforward prompting tool to enable the safe and legal use of generative AI. The business teams were also given training and educational materials on generative AI and its potential applications. This tactic effectively led the company’s adoption of generative AI, enabling the platform to swiftly add support for more use cases without necessitating a thorough understanding of every new application by the core platform team.
Developing Pluto AI
The Google Cloud Consulting team and the AI Team of Magyar Telekom worked together to build Pluto AI. This partnership was crucial to ensuring the platform maintained industry best practices, telecom industry security and compliance requirements, and industry standards.
Teams can quickly integrate, alter, and update AI models, tools, and architectural patterns in Pluto AI due to its modular design. Because of its architecture, the platform can support a broad range of use cases and develop quickly in tandem with Magyar Telekom’s AI goal. Retrieval Augmented Generation (RAG), which combines LLMs with internal knowledge sources, including multimodal content, to provide grounded responses with evidence, API access that enables other parts of the organization to integrate AI into their solutions, Large Language Models (LLMs) for natural language understanding and generation, and code generation and assistance capabilities to increase developer productivity are some of the key modules integrated into Pluto AI.
Additionally, the platform enables customers to design AI assistants that are customized to meet certain business requirements.
In terms of scalability and dependability, Pluto AI is technically based on Compute Engine and uses virtual machines. It makes use of foundation models from the Model Garden on Vertex AI, including as the Claude 3.5 Sonnet from Anthropic and the Gemini, Imagen and Veo models from Google. The knowledge bases required for RAG processes are stored in ElasticSearch on Google Cloud. Production-ready apps are developed using other Google Cloud services, such as Cloud Logging, Pub/Sub, Cloud Storage, Firestore, and Looker.
The user interface and experience were given a lot of attention throughout development. Pluto AI’s user-friendly interface ensures that employees can take use of AI capabilities without facing a high learning curve, regardless of their level of technical skill.
With hundreds of daily active users from a variety of departments, the platform has had excellent adoption rates. Employees have given the platform excellent comments due to its extensive use and practicality. Numerous AI assistants, such as those for knowledge management, software development, legal and compliance duties, and customer care chatbots, have been made possible by Pluto AI.
Quantifiable outcomes demonstrate the influence of Pluto AI. Tens of thousands of API calls and hundreds of thousands of unique users are recorded by the platform every day. Benefits include a 15% reduction in coding errors and an anticipated 20% reduction in the amount of time spent examining legal papers.
Vision and future roadmap for Pluto AI
In the long run, Magyar Telekom views Pluto AI as a key component of its AI strategy. There are plans to extend the platform to more departments, business divisions, and markets. Additionally, the business is investigating the possibility of providing Pluto AI as a product or service to additional Deutsche Telekom markets.
Additional improvements are planned, including the development of an improved interface for managing RAG solutions and tracking usage, the addition of a language selection for multilingual support, and an investigation into agent-based AI systems for automating complicated activities. Another priority is keeping an eye on and optimizing cloud resource consumption and expenses.
Pluto AI has essentially transformed Magyar Telekom’s AI environment, making AI useful, approachable, and influential throughout the company. Pluto AI establishes a new benchmark for internal AI adoption by encouraging experimentation and providing measurable business benefits.
