Atomicwork brings AI agents to service management with Azure AI Foundry

Customer background

Atomicwork, an agentic service management software provider, was founded in 2022 and has quickly grown to over 50 employees in the United States, Singapore, and India. The company specializes in an AI-first approach to IT and enterprise service management, setting itself apart by developing its platform from the ground up with artificial intelligence at its core. The company's unique value proposition is its conversational service management platform, which fundamentally reimagines how businesses approach IT service management (ITSM) and enterprise service management (ESM).

As hybrid work reshapes the modern enterprise, efficient IT service management and digital workplace automation have become essential to driving productivity and enhancing employee experience. With Microsoft Teams and Slack at the heart of workplace collaboration, Atomicwork delivers service management solutions directly to end users, simplifying IT operations and empowering businesses to scale faster and operate more efficiently. With Atomicwork the portal experience comes to the users; the goal is to avoid going to a portal altogether.

Unlike traditional service management providers who have had to retrofit AI capabilities into legacy systems, Atomicwork's platform was designed in the age of advanced AI, allowing them to create an AI-native architecture from the get-go. This timing advantage has allowed them to develop a more integrated, efficient solution that directly addresses the changing needs of the modern hybrid workforce rather than bolting-on AI as an afterthought. Since emerging from stealth mode, the company has achieved significant milestones, including replacing established players in enterprise environments.

The challenge: from service desk to innovation leaders

Traditional service management systems struggle to provide the agility and efficiency required in today's digital workplace. Atomicwork identified a critical issue: legacy platforms are siloed, with disconnected tools and outdated technologies impeding real-time communication and adaptability. Implementing enterprise-ready service management solutions becomes a complex task, requiring significant effort to build and maintain integrations across various components of the employee tech stack.

The effects on productivity are significant. Service teams such as IT and HR are inundated with operational tasks and repetitive inquiries. Basic questions like "How do I get VPN support?" or "Where can I find my payslip?" consume their workday, diverting valuable resources away from strategic initiatives. This reliance on human involvement for routine service requests causes delays and inefficiencies, especially in Level 1 and Level 2 support scenarios.

The burden extends beyond service teams and affects the entire organization. Employees experience app fatigue and productivity drains as they deal with massive amounts of digital data across numerous software tools. Teams at all levels, from individual contributors to leadership, find themselves spending too much time looking for information and status updates on traditional support portals, rather than focusing on value-added tasks. The cumulative effect undermined organizations' ability to set and achieve strategic objectives.

According to Lenin Gali, Chief Digital and Business Officer at Atomicwork, "Over the last three decades—from BMC Remedy to ServiceNow—traditional service management has evolved but still requires high setup overhead and maintenance. However, they have all been focused on how IT can best serve employees. What we needed in the next generation was to empower employees and IT to best serve themselves using automation and an Agentic AI approach."

The transformation of service management from a cost center to a strategic enabler required a complete rethinking of how employees interact with IT services. The key is to embed operational knowledge and business automation directly into the flow of work—in the collaboration tools that employees use every day.

The solution: building an AI-native service platform 

Atomicwork chose Microsoft Azure AI as the foundation for their next-generation service management platform because of its deep integration into business ecosystems and widespread adoption by global enterprises. Recognizing that many of their target customers were already using Microsoft products like Teams, Atomicwork saw an opportunity to integrate their AI Agent, Atom, directly with the tools employees use daily. This includes everything from Teams to Microsoft Entra ID, Intune, SharePoint, Outlook and GitHub Enterprise; every knowledge and workflows app an enterprise is likely to be using.

"When it came to AI models, the Azure AI platform was miles ahead for enterprise-segment offerings compared to alternatives," says Lenin Gali.

“The combination of accuracy, performance, and built-in security aligned perfectly with our AI and information security principles.” Most enterprise environments already had Microsoft Teams, so embedding Atomicwork's conversational interface was a natural fit.

Atom is the platform's AI Agent, built on an ensemble architecture that uses multiple AI models for various aspects of service management. Unlike traditional systems that add AI capabilities to existing architectures, Atomicwork designed its platform with AI embedded at every layer, ensuring seamless integration and optimal performance.

Atom connects with an organization’s knowledge sources, including SharePoint, Confluence, Jira, among others. The data is segmented into manageable chunks and embedded using OpenAI's embedding model, after being hydrated with user and enterprise context. A troop of AI agents manage user interactions using a sophisticated set of steps. When a user submits a question, it goes through content moderation, ambiguity checking, and multiple guardrails in place, before retrieving and serving the right answers generated to the user. Cohere's Reranker model also plays a crucial role in prioritizing the best result to return. For service requests requiring structured data, Atomicwork’s AI streamlines form completion by automatically filling in known details saving users time and reducing manual input using Cohere’s Command model. Model evaluation and selection are critical steps in the implementation process. Atomicwork uses a curated golden dataset to evaluate model accuracy and efficiency across a variety of use cases, including re-ranking and slot filling. They use performance metrics, such as similarity and relevance scores, to ensure that each model meets precise accuracy and performance requirements.

Results and benefits: delivering impact from day one

Atomicwork's Agentic service management platform has produced impressive results for its organizations, unlocking enterprise IT and service teams’ productivity. One customer's experience of achieving a 65% deflection rate within six months of implementation with projections of an 80% rate by the end of the year demonstrates that efficiency improves quickly as the system is adopted across functions in the organization.

Technical performance metrics demonstrate the Azure AI platform's effectiveness. The implementation of Cohere Rerank resulted in a 20% increase in accuracy, while overall system performance showed a 75% reduction in response latency when compared to competing solutions.

“Our decision to build on Azure OpenAI Service has proven instrumental to Atomicwork’s success. The platform’s capacity and throughput enable us to handle enterprise-scale demands with exceptional response times, while multi-regional support ensures consistent performance for our global user base. Combined with GPT-4o’s enhanced accuracy, we’re delivering AI solutions that don’t just meet enterprise standards—they set new ones." says Suresh Ponnusamy, Head of Platform Engineering at Atomicwork.

The transformation achieved through Atomicwork's solution is particularly well illustrated by the experience of Karthik Chakkrapani, SVP, Corporate Operations & CIO at Zuora, who explains: “With Atomicwork’s modern service management and Agentic AI platform, we are transforming the digital workplace experience at Zuora. Our employees now enjoy instant support, seamless information access, and automated service workflows—all via Slack. This partnership is driving faster resolutions, enhanced productivity, and a smarter, more connected workplace.”

The impact on operational efficiency has been similarly impressive. The platform's ability to automatically handle routine Level 1 and Level 2 support requests has given IT teams more time to focus on strategic initiatives. This efficiency extends beyond the IT department, as Ryder Hampton, Head of Technology, Enterprise Services at Peppermoney points out: "Atomicwork is a valuable partner in our journey towards World-Class IT Operations. We were able to deploy this new platform and replace a number of incumbent solutions within 6 weeks. We have already seen significant improvements across ticket deflection, self service and most importantly employee experience. Moving across to the Atomicwork solution also reduced our TCO by consolidating 3 different solutions into one platform."

Atom's seamless integration with Microsoft Teams and other enterprise systems has greatly enhanced the employee experience. Employees can access information and services without having to switch between multiple applications or wait for human intervention, as the operational knowledge is embedded directly into the workflow. This has resulted in lower operational overhead for service management teams and faster response times for employees who require assistance.

Looking forward: blueprint for evolution

Atomicwork's platform roadmap aims to broaden its capabilities through deeper Microsoft service integrations. The company is actively developing connections to Azure DevOps for improved collaboration between IT teams DevOps teams, PowerApp and Power Automate for expanded automation capabilities, and Power BI for advanced analytics and reporting. They also intend to integrate with Microsoft's Copilot ecosystem, which will expand the capabilities of their AI Agent.

"At Atomicwork, we are committed to building and deploying AI solutions with TRUST - Transparency, Responsibility, User-centricity, Security, and Traceability at our foundation. Microsoft’s technology and partnership align with our mission. Their focus on security, compliance, and responsible AI practices plays a pivotal role in enabling us to meet these commitments for enterprises," says Aparna Chugh, Head of Product at Atomicwork.

The platform's ensemble AI architecture will evolve further, optimizing the use of various AI models specialized for different service management tasks. By leveraging these expanded integrations and enhanced AI capabilities, Atomicwork hopes to provide even more seamless automation and self-service capabilities within their customers' existing Microsoft Teams.