Building a Use Case from Scratch for Agentic AI in ServiceNow: Insights from the Frontlines

Earlier this week, I had the privilege of hosting a pivotal webinar in our run-up to the highly anticipated AI Hackathon, “Germany vs Netherlands: Mastering Agentic AI”. The session, titled Building a Use Case from Scratch for Agentic AI in ServiceNow, served as a deep-dive exploration into the next frontier of enterprise automation.

With Mel from Strat Five and Luuk from Sequal Consultancy at my side, we dissected not only the “what” and “how” of Agentic AI, but also the “why now.”

We published the webinar on our Youtube Channel, so you can watch it anytime.

What emerged was a thoughtful, grounded conversation on practical architectures, pitfalls, and transformative opportunities.

In the following article, I would love to summarise my discussion with these fantastic guests and provide you the key takeaways and relevant background information.

What is Agentic AI, Why Now?

Only 20% of enterprise AI use cases deliver on their promise, according to a recent McKinsey report. That statistic should be a wake-up call for every IT decision-maker. The clear takeaway? Success hinges on clearly defined, outcome-driven use cases.

This is where Agentic AI makes its entrance – not as a hyped-up chatbot, but as a strategic leap forward.

Mel defined it well: Unlike deterministic workflows (which follow an “if this, then that” logic), Agentic AI takes on goals and autonomously navigates multiple paths to achieve outcomes. Picture a digital colleague you simply brief with a task – “do my homework” – and it determines the needed resources, executes the task, and delivers the output. .

Understanding Agentic AI in Context

At its core, Agentic AI extends traditional automation by introducing autonomy and context awareness. It gathers data, connects across systems, makes decisions, and delivers outcomes without constant human steering.

Luuk explained that within ServiceNow, much of the foundational architecture for Agentic AI is already in place. The Now Platform not only contains native process data and logic but is also capable of connecting to external systems without physically moving data, thereby upholding security and compliance boundaries.

This leads to a significant insight: Agentic AI is not an add-on. It’s an evolution of platform-native intelligence.

Real Use Cases: What Works (and What Doesn’t)

Mel and Luuk both emphasized that not every process is ripe for Agentic AI. Here’s what typically makes a good candidate:

• High information retrieval: Cases requiring repeated lookups across multiple systems, like triaging a customer support ticket before assigning it.
• Conditional routing: Scenarios where logic varies based on parameters like geography, regulation, or product type.
• Cross-system orchestration: E.g., registering results back into ERP, CRM, and compliance systems.

Poor candidates? Deterministic, linear workflows such as simple order processing or rigid task sequences.

Here’s a powerful takeaway: Agentic AI thrives in ambiguity where human-like decisioning adds value.

Architecting from Scratch: Lessons from the Trenches

Luuk shared an implementation view that demystifies the process. “It’s like building with LEGO blocks,” he said. ServiceNow’s Now Assist comes with built-in GenAI and Agentic AI capabilities. Developers can assemble workflows using pre-defined or custom-built skills, without heavy coding. The real skill lies in knowing your environment – your data, your processes, and your business logic.

Mel added a critical layer: Before building anything, platform owners must set up guardrails.

This includes:

• Control towers to monitor AI activities
• Clear data exposure policies
• Cost oversight for internal and external LLM usage
• Compliance readiness, especially with incoming regulations like the EU AI Act

This foundational step, often skipped, is vital to scale safely and effectively.

Starting Small: The Agile Path to Success

Perhaps the most resonant piece of advice from Luuk was: “Start small.” Many companies fail by trying to implement full-blown autonomous systems from the get-go. Instead, pick one part of a workflow (like data gathering) and automate that. Over time, layer additional agents.

This principle echoes a broader market trend: According to IDC, over 60% of enterprises plan to adopt modular AI in 2025. Micro-use cases, rather than monolithic solutions.

In this report, the analysts emphasize that GenAI success hinges on more than just technology: enterprises must enable citizen developers, embed explainability, and build AI agents with clear goals and measurable outcomes. With retrieval-augmented generation (RAG), hybrid edge-cloud deployments, and strong governance becoming standard, companies must now balance innovation with control.

Rather than aiming for fully autonomous systems from day one, IDC advises a modular approach. It’s starting with focused, value-driven use cases that scale. The future of AI is not general-purpose. Deeply integrated, business-aligned, and responsibly governed. Those who act now will shape the competitive edge of tomorrow.

Market Context: What’s Driving This?

The market dynamics surrounding Agentic AI are both urgent and transformative. During the webinar, this context subtly underpinned much of our discussion—particularly in how Agentic AI sits at the crossroads of three converging forces: AI commoditization, platform consolidation, and enterprise readiness. That means, that businesses are looking to deploy Agentic AI in platforms like ServiceNow.

AI commoditization:

LLMs are now APIs. Capabilities like retrieval-augmented generation (RAG) or vector search are off-the-shelf.

Platform convergence:

With systems like ServiceNow incorporating AI natively, the friction to adoption is vastly reduced.

Governance maturity:

Businesses now treat AI less like a toy and more like a strategic asset. It complete with compliance, ethics, and ROI metrics.

Over the past two years, we’ve witnessed a dramatic shift: once the reserve of elite R&D teams, foundational AI capabilities are now available as APIs. Thanks to providers like OpenAI, Google, Meta, and others, large language models (LLMs) and retrieval-augmented generation (RAG) systems are plug-and-play. This democratization reduces time-to-value for organizations and aligns with what Luuk noted during the webinar: “You don’t need much technical skill. You need to know what you want”

One of the strongest advantages ServiceNow offers (emphasized by both Mel and Luuk) is its tightly integrated ecosystem. The platform already houses workflows, process logic, permissions, and historical data. Agentic AI plugs directly into this context, delivering vertical intelligence instead of generic AI assistance.

This is crucial in an enterprise market moving toward contextual autonomy, where AI needs to understand workflows, compliance requirements, and user roles in order to act reliably and securely.

Business impact is driven by faster deployment cycles (reuse of native assets e.g., workflows, roles, records), higher trust (data does not leave the ecosystem – critical in regulated sectors like banking, healthcare, and government), and lower friction for scale. As Luuk put it, “The data is already there, and the platform knows what to do with it”.

To summarize, Agentic AI is no longer a technology trend. It’s a strategic shift in how work gets done. As the market accelerates toward autonomous operations, now is the time for CIOs, platform owners, and architects to redefine their roadmap. Or, risk being defined by someone else’s.

Looking Ahead: The Hackathon as a Testing Ground

This all brings us to the upcoming Hackathon on June 25th. Across Munich and The Hague, teams will compete in a cross-border showdown to design and prototype Agentic AI solutions on the ServiceNow platform. The event is more than a coding sprint. It’s a sandbox to test real use cases, guided by the principles we discussed.

Mel put it best: “It’s fun, but it’s also functional. You get a real solution at the end of the day.”

Whether you’re a platform owner, consultant, or innovator, this is your chance to get hands-on and challenge your assumptions.

The event is designed to translate theory into practice by enabling teams to:

• Identify real business pain points.
• Design and build Agentic AI use cases using ServiceNow’s Now Assist and GenAI modules.
• Gain hands-on experience in automation, data orchestration, and outcome-driven design.

Format and Flow

• Date: June 25, 2025
• Locations: Live and in parallel at Munich (Germany) and Den Haag (Netherlands)
• Participants: ServiceNow customers and partners—architects, consultants, IT leaders, and decision-makers
• Structure:
  • Short briefings and warm-ups
  • Teams of 3–5 participants
  • Choose from real-world or bring-your-own use cases
  • Build Agentic AI agents under guidance from experts
  • Presentations and live demos at the end of the day
  • A competitive jury-led scoring and selection of the best team

What Makes This Unique

• Cross-border challenge: It’s Germany vs. Netherlands. Like football, but for enterprise tech.
• Live environment: Build and demo directly in the ServiceNow platform.
• Multidisciplinary collaboration: Business, IT, and operations working together to simulate real-world dynamics.
• Mentorship: On-site support by ServiceNow experts and implementation consultants (from Teiva Systems, Strat Five, Sequal Consultancy).

At the end of the hackathon, each team walks away with a working prototype of an Agentic AI use case, a better understanding of the real deployment cycle, feedback from industry leaders and peers, and a lot of fun.

And of course, for one team. The prize and bragging rights of winning the international showdown.

This makes it a valuable opportunity not just to explore technology, but to stress-test AI readiness across people, process, and platform.

Final Thoughts

As the host, I came away from this webinar with one major learning: Agentic AI isn’t the future – it’s already here. But how you implement it determines whether it becomes a business differentiator or just another shiny tool.

We hope to see many of you in person at the Hackathon, where theory meets practice. Until then, stay curious – and keep building.

---

Teiva Systems is a ServiceNow Partner and provides advisory, architecture, implementation, training, and support services. Teiva Systems specialises in building custom applications and integrations with ServiceNow App Engine and GenAI leveraging platform capabilities like common data model, security, integration layer, now assist etc.

Kostya Bazanov is ServiceNow Leading Architect and Technology Advisor since 2011. He acts as technical consultant, implementer, solution architect, team lead in national and global projects and ServiceNow platform rollouts.

Kostya Bazanov, Managing Director, Jun 17, 2025

Eager to take the next step? Contact us today!

Privacy policy (GDPR)

* Required fields

Submit