Persistent Systems (PERSISTENT) Status Update summary

Industry trends and AI evolution

• AI is rapidly advancing, with significant commercial implications and is seen as the next major technological revolution after the internet and mobile technologies.

• Generative AI (GenAI) is only a subset of AI; broader AI techniques are driving transformation across industries.

• AI is expected to impact all sectors, including healthcare, finance, and education, enhancing productivity and changing human interaction.

• Software engineers will need to adapt, as AI will surpass humans in code writing, shifting the focus to integrating and managing AI systems.

• Agentic AI, which enables autonomous, end-to-end processes, is poised to redefine industries with both significant opportunities and risks.

Technical foundations and model development

• Deep learning and neural networks, enabled by advances in hardware and data availability, have driven recent AI breakthroughs.

• Large language models (LLMs) are trained on vast datasets and can be fine-tuned for domain-specific applications, with both open and closed-source options available.

• Smaller, domain-specific models are gaining traction due to lower costs and easier customization, especially for enterprise use.

• Techniques like fine-tuning and retrieval-augmented generation (RAG) allow organizations to enhance model performance with their own data.

• Training LLMs is resource-intensive, but innovations like DeepSeek demonstrate that efficient, high-quality models can be built with fewer resources.

Agentic AI and enterprise deployment

• Agentic AI refers to autonomous systems capable of perceiving, reasoning, and acting independently across complex workflows.

• Building agentic AI solutions requires integrating multiple agents, verifiers, and API calls, demanding significant engineering and planning.

• Enterprises will increasingly use smaller, fine-tuned LLMs for mission-critical applications, balancing cost, control, and performance.

• Industries with high-quality, well-organized data (e.g., finance, healthcare, supply chain) are best positioned to adopt agentic AI early.

• The shift to agentic AI will be gradual, with initial adoption of off-the-shelf models, followed by customized, end-to-end solutions.