Rewind 2024: Top enterprise AI trends in vogue

In 2024, the realm of artificial intelligence (AI) witnessed a significant surge with global technology players like Nvidia, Google, Microsoft, and AWS, among several others consolidating their position in the AI space. Throughout the year, AI and generative AI (GenAI) dominated the technological landscape, with virtually every IT conference, product unveiling, and news event being associated with AI, with experts calling 2024 “a critical year for AI” with organisations exploring how this technological leap can be integrated into daily life and work.

Here are some of the key AI trends observed this year that are likely to get bigger in the coming months:

Agentic AI transforming industries 

As businesses strive to innovate in a complex digital landscape, Agentic AI is emerging as a transformative force. This technology enhances intelligent decision-making and real-time adaptability across sectors like healthcare, banking, financial services and insurance (BFSI), energy, and retail, among others.

Kiran Raj, Practice Head of Disruptive Tech at GlobalData, says, “Agentic AI represents a big shift from traditional automation. By integrating perception, reasoning, and autonomous action, it empowers enterprises to navigate complexity with agility and precision, resulting in higher efficiency, reduced costs, and superior customer experiences.”

For example, an AI agent in environmental monitoring could detect early signs of forest fires, while a financial AI agent might manage investment portfolios using adaptive strategies, he said.

From streamlining product strategies to boosting customer engagement and automating tedious tasks, Agentic AI has enabled proactive and adaptable user interactions and is fostering dynamic exchanges that is benefiting both businesses and consumers, believes Raja Lakshmipathy, Vice President and Managing Director at Genesys India & SAARC.

The Rise of Multimodal AI 

Multimodal AI is transforming the AI landscape by unifying data from various sources—text, images, audio, video, and sensors—into cohesive models. Dhruv Pathak, co-founder and CTO of INDmoney sees this approach as enhancing AI decision-making, accuracy, and generalisation, enabling AI to process sensory information more like humans. For instance, OpenAI's GPT-4 can handle both images and audio, facilitating more natural interactions. Other examples include Gemini 2.0 Flash, Mistral, Cohere, and Llama.

Multimodal AI started finding diverse real-world applications. For example, in healthcare, it can analyse medical images alongside patient history and genetic data to improve diagnostic accuracy. Insurance firms can assess potential fraud in insurance claims. By incorporating unstructured data, like handwritten notes from adjusters, alongside structured data, the accuracy of the models significantly improved.

Shift to Open-Source AI 

Developing powerful AI systems, particularly large language models (LLMs), is resource-intensive and costly, historically limiting such advancements to tech giants. However, the rise of open-source AI is democratising access, enabling startups and independent researchers to contribute to cutting-edge technologies.

“The rise of open-source AI is democratising access and fostering innovation,” Murali Brahmadesam, fintech firm Razorpay’s Chief Technology Officer, said in an interview with TechCircle. The company open-sourced more than 140 projects, with a strong eye on tightening security. Not just Razorpay, a research from IBM shows that a majority of the companies are investing in AI for the long term, with a growing interest in open-source tools for driving return of investment (RoI) and innovation.

Customised Enterprise GenAI models 

While generative AI becomes more sophisticated in 2024, and general-purpose tools like ChatGPT continue to gain momentum, businesses found more value in smaller, niche AI models tailored to specific needs.
“As demand for customised generative AI grows, organisations prefer fine-tuning existing models on domain-specific datasets to make them more cost-effective,” says Sathesh Murthy, Managing Director & Engineering Head for India at RingCentral.

He highlights the high costs and latency of using large public models like GPT-4 via API, noting that tailored models can address specific business requirements in areas such as customer support, supply chain management, and document review, especially in sectors with specialised terminology like healthcare and finance.

Emergence of Shadow AI 

Shadow AI—often referred to as "bring your own AI"— has been in vogue this year. It occurs when employees use unapproved AI tools in the workplace without IT or security approval. Salesforce reports that over half of generative AI users rely on unapproved tools, and 70% of workers globally lack training on safe and ethical usage.

Sreekanth Menon, global head of AI/ML at IT firm Genpact sees it as a potential disaster that can cause havoc within an organisation and believes though it is early days of shadow AI, the trend shows no signs of slowing down.

Organisations should implement unified AI policies across departments and raise awareness of AI risks beyond security teams. CIOs must clarify these risks and encourage reporting of unauthorised large language model (LLM)-based application use. Promoting transparency is vital for identifying threats and mitigating risks in this evolving landscape.

Ethical AI in the Spotlight

As AI expands globally, the realisation dawned that managing its risks is increasingly vital. The rise of tools like ChatGPT also led to copyright disputes, including lawsuits against OpenAI for infringement. Governments and organisations globally started advocating for responsible AI development to combat issues like data manipulation, misinformation, bias, and privacy violations.

In March, the EU advanced comprehensive AI regulations to address consumer concerns, with implementation expected later this year. In the US, California led the state-level AI regulation, proposing laws to enhance accountability and combat discrimination, while Colorado enacted the first comprehensive AI legislation in May, focusing on algorithmic discrimination in critical sectors. In the same month, Buenos Aires hosted UNESCO’s first Regional Summit on AI, presenting nine oversight approaches for individual nations.

India’s Ministry of Electronics and Information Technology also released a blueprint for a new Digital India Act addressing high-risk AI systems.

“Ensuring AI operates within ethical and legal boundaries is essential to mitigate risks and foster trust in the technology,” says Muthumari S, Senior Director at Brillio, stressing the importance of incorporating design thinking—a human-centred problem-solving approach—to navigate some of these complex challenges effectively.