At Harbinger, we accelerate innovation across software engineering and digital learning by integrating Agentic AI throughout development pipelines and content ecosystems.
Whether you're building enterprise-grade applications or rapidly producing instructional content, Agentic AI enables your teams to do more—faster and smarter. Our AI agents support real-time code reviews, cross-platform development, test automation, content generation, and accessibility checks.
In this episode, discover real-world stories of how Harbinger teams are applying Agentic AI to streamline workflows, improve quality, and reduce time-to-market in both product engineering and eLearning domains.
Watch Shrikant Pattathil, Chief Technology Officer at Harbinger Group, in conversation with Shweta Kulkarni, Vice President – Projects, and Poonam Jaypuriya, Vice President – eLearning. Together, they unpack real-world examples of how Harbinger is using agentic AI in product engineering and content creation, right from automating project management and testing to scaling multimedia content production. This episode provides actionable insights for teams ready to adopt agentic AI meaningfully.
1. How are businesses using agentic AI in product development today? Can you share real-world examples?
Agentic AI is being used to automate tasks and support decision-making. For example, Harbinger built an agent named Robo that integrates with JIRA to classify tickets, detect duplicates, and post updates automatically. Another agent we built, Watchdog, monitors system metrics like CPU spikes or error rates and simplifies root cause analysis. A third example is Mabl, which enables testers to create self-healing test cases that adapt to UI changes—reducing QA effort and accelerating release cycles.
2. What are some real examples of how Harbinger teams have applied agentic AI internally?
At Harbinger, we piloted agentic AI by embedding GitHub Copilot into multiple pod teams. While Copilot improved development speed, code quality varied, and review cycles grew longer. To solve this, we developed an AI agent that enforced coding standards around security, performance, and best practices—directly within the Copilot environment. This led to improved code quality and a sharp reduction in review comments. In another project, we built an agent to convert iOS code to Android, reducing duplicate effort and boosting developer productivity across platforms.
3. What best practices should teams follow when starting their AI journey in software development?
Start small but focus on a high-impact use case to get early buy-in. Always keep a human in the loop to supervise or override agent-driven actions. Design modular agents that can perform specific, independent tasks and be reused across workflows. Most importantly, treat agents like products—assign ownership, gather user feedback, and continuously refine their behavior.
4. In what ways has agentic AI transformed your approach to content development at scale?
We recently developed a full series of leadership training videos for a large digital publisher using agentic AI workflow. Starting with just course outlines, agents handled everything—from content extraction and scriptwriting to visual asset creation and video assembly. With human oversight at each step, we reduced production time from 6–8 days to under 2 days per video—while fully meeting the client’s quality expectations.
5. What are some best practices for digital learning providers beginning their agentic AI journey?
Use AI agents to augment, not replace, your instructional design teams. Define a clear role for each agent within your workflow and ensure human oversight for consistency and nuance. Start with one agent performing a focused task, iterate as needed, and scale gradually. This approach ensures faster development while maintaining learning effectiveness and quality.
Scale development speed, enhance quality, and unlock real results with agentic AI