Executive Summary: Intelligent Automation Services - U.S. 2024
The individual quadrant reports are available at:
ISG Provider Lens™ Intelligent Automation Services - Intelligent Enterprise Automation - U.S. 2024
ISG Provider Lens™ Intelligent Automation Services - Next-Gen Automation - U.S. 2024
Agentic AI and autonomous Ops emerging as disruptive forces in intelligent automation services
Advancements in AI applications are redefining the landscape of business service automation in the U.S., disrupting use cases and codifying service knowledge using generative AI (GenAI) technologies. These innovations, powered by large language models (LLMs), vision-language models (VLMs), small language models (SLMs), retrieval-augmented generation (RAG) and other appropriate finetuning techniques, are revolutionizing how organizations automate customer service, IT support and business operations.
The evolution of AIOps: Integrating GenAI for predictive analytics, autonomous incident management and self-healing IT systems
AIOps, which uses AI and ML to enhance and automate IT operations through GenAIbased knowledge search and retrieval-based relevance and recency, ensures improved performance, enhanced incident resolution and efficient resource management. Since 2023, AIOps has been increasingly integrating GenAI technologies to bring new capabilities to business service automation.
A key focus area in AIOps is predictive analytics, where GenAI models and finetuned LLMs analyze large amounts of operational data to predict system failures or identify issues before they escalate. Using historical data and patterns, these AI-powered tools can detect anomalies in IT systems in near real time and generate accurate predictions of potential issues, enabling businesses to resolve them before they affect performance. For instance, AI-driven incident management platforms such as Moogsoft and Dynatrace, now augmented with GenAI capabilities, are using LLMs and
VLMs to autonomously analyze system logs and assess infrastructure performance and load curves to predict potential failure points, automate ticket generation and escalation and even propose potential solutions based on previous incidents. In complex IT environments, GenAI models and AI tools can identify root causes by analyzing system logs, event data correlations and recommend appropriate remediation steps.
LLMs are being integrated into AIOps platforms to automate system updates, new feature release cycles, patches and configurations. By analyzing past data, AI tools can predict the optimal time to execute updates and patch applications without disrupting ongoing operations. This ensures continuous improvement of IT infrastructure with minimal human intervention. Another growing trend in 2023 and 2024 is deploying self-healing systems that use AI models to detect, diagnose and resolve issues autonomously. These systems continuously monitor platform health, detect performance issues and trigger automated workflows to fix problems, thus enhancing uptime and operational resilience.
Transforming business service automation: Leveraging GenAI for smart RPA, personalized customer service and cross-functional efficiency
Overview:
GenAI is instrumental in driving innovation in business service automation, leaping into intelligent RPA services, particularly in customer service, finance, HR and other business operations. The ability of GenAI to produce content, assist in decision-making and automate workflows has allowed businesses to enhance service delivery while reducing
operational costs.
Customer service has been one of the primary beneficiaries of GenAI, as businesses automate support operations, improve personalization and deliver quick resolutions. LLMs such as OpenAI’s GPT models in customer service chatbots are a growing trend. These AI-powered virtual assistants can engage with customers using natural language, provide personalized responses and resolve queries without human intervention. Using RAG techniques, these chatbots can retrieve and present relevant information from databases or documentation to answer complex queries accurately. Unlike traditional chatbots, which follow predefined scripts, modern AI-driven bots use VLMs and LLMs to understand customer queries in context. For instance, VLMs enable businesses to integrate image recognition capabilities into customer service systems, allowing users to upload images for troubleshooting, such as reporting a defective product or verifying identification.
GenAI will enable businesses to offer personalized support at scale in omnichannel, context-integrated or platform-neutral approaches. For example, SLMs can be deployed to handle specific customer segments, generating responses tailored to a customer’s history, preferences and behaviors across calls, video chats, emails and text chats. This level of personalization improves CSAT and fosters long-term loyalty. In addition to customer service, GenAI drives automation across various business functions, including finance, HR and supply chain management.
AI models, particularly LLMs finetuned for specific robotic or hybrid business tasks, are being used to automate the processing of invoices, contracts and other documents. AIpowered tools such as UiPath and Automation Anywhere integrate with generative models to read, interpret and process unstructured data from business documents, reducing manual labor and improving accuracy.
GenAI is being applied to automate decisionmaking in finance and HR operations. In HR, AI models can analyze employee performance data to generate recommendations for promotions, training or restructuring. In finance, AI-driven platforms can generate financial forecasts, manage accounts payable and receivable, and identify fraud risks, streamlining operations and enhancing decision accuracy.
Revolutionizing business operations by leveraging finetuned AI models
Finetuning AI models has become a significant trend in the business service automation landscape, allowing businesses to customize LLMs, VLMs and SLMs with industry-specific knowledge and functional updates tailored to enterprise-specific business operational contexts. Large pretrained models such as GPT- 4 and BERT are competent as they are often finetuned to meet the specific requirements of a business domain, including the governance, risk and compliance (GRC)-related knowledge of transactional rules and security and data privacy guardrails. This finetuning process involves training AI models on proprietary data to improve their accuracy in generating contextually relevant responses. For instance, finetuned AI models are used in healthcare to create personalized treatment plans or review patient records, integrating billions of parameters and patterns of genetic models and complex protein chains in different verticals such as pharma and genetic medicine. Similarly, legal firms are finetuning LLMs to automate contract analysis, case law research and document generation, improving the speed and efficiency of legal services. In retail, AI models are finetuned to generate product recommendations, create marketing content and manage inventory. These models analyze consumer behavior and sales data to predict demand, optimize stock levels and personalize marketing campaigns.
RAG techniques and finetuning are also applied in customer service to deliver more accurate and context-aware responses. RAG combines GenAI with retrieval-based systems, where the AI model generates responses based on
information retrieved from a knowledge base. This is particularly useful in customer service, where AI models must access a company’s product documentation, FAQs and customer history to provide accurate and relevant support. Finetuning techniques are enabling AI models to offer multilingual support, addressing the needs of a diverse customer base. By training models on region-specific data, businesses can ensure that their AIdriven services resonate with local customers, improving communication and engagement.
How VLMs and SLMs are transforming visual and speech data processing in high-performance AI automation
While LLMs focus on text-based automation, in the successive waves of high-performance AI-powered automation, VLMs and SLMs transform how businesses handle visual and speech data, automating more complex service tasks. VLMs such as Contrastive Language-Image Pretraining (CLIP) enhance e-commerce, manufacturing and healthcare automation by allowing AI systems to interpret and act on visual information. CLIP efficiently learns visual concepts from natural language supervision. The tool can be applied to any visual classification benchmark by simply providing
the names of the categories for recognition, starting with the zero-shot capabilities of GPT-2 and GPT-3.
Although deep learning has revolutionized computer vision, earlier approaches had several significant challenges. Typical vision datasets were labor-intensive and costly to create, often focusing on a narrow set of visual concepts.
Standard vision models excelled at a single task but required considerable effort to adapt to new tasks. Models that perform well on benchmarks often exhibited disappointing performance during stress tests. In contrast, new approaches are trained on diverse image datasets with a wide variety of natural language supervision readily available on the internet. These networks are designed to be instructed in natural language to perform various classification
benchmarks without directly optimizing for the benchmark’s performance. This is similar to the zero-shot or few-shot capabilities of earlier GPT models. New systems enhance generalizability and robustness, closing the gap by up to
75 percent while achieving performance on par with the original ResNet-50 zero-shot model, all without using any of the millions of labeled examples. VLMs enable platforms to automate support for visually rich tasks, such as product troubleshooting. Customers can upload images of defective products, and AI systems equipped with VLMs can diagnose issues and provide instant solutions.
Similarly, voice assistants are powered by energy-efficient SLMs to provide more natural and efficient customer service experiences. These assistants can understand spoken language, process queries and offer spoken responses, reducing the need for text-based support. In the healthcare, finance and legal sectors, SLMs automatically transcribe
conversations and ensure compliance with regulatory requirements. These AI tools enable quick audits and accurate recordkeeping by integrating voice data with business systems.
While integrating AIOps and GenAI into business service automation brings numerous advantages, organizations must also address the challenges in enterprise application scenarios. As AI models access and analyze vast amounts of business and customer data, ensuring privacy and security remains a critical concern. Organizations must implement robust data governance frameworks to comply with regulations like GDPR and California Consumer Privacy Act (CCPA) while ensuring that AI models do not inadvertently expose sensitive information. AI models, especially LLMs and
VLMs, are also prone to biases if not properly finetuned and monitored.
The rapid integration of GenAI into business service automation is reshaping industries across the U.S., setting a new benchmark for efficiency, accuracy and personalized services. With advanced AI models like LLMs, VLMs and SLMs driving innovations in AIOps, smart RPA and cross-functional automation, organizations are enhancing service delivery and achieving significant cost reductions. Predictive analytics, autonomous incident management, personalized customer engagement and finetuned AI capabilities are vital to this transformation, providing scalable and contextaware solutions across various sectors such as IT, customer service, finance and HR. However, organizations must address data privacy, security and model bias challenges as they use these AI advancements. As GenAI continues to evolve, its ability to optimize complex workflows and enhance CX will make it indispensable in the future of enterprise automation.
Access to the full report requires a subscription to ISG Research. Please contact us for subscription inquiries.