Systems | Development | Analytics | API | Testing

This lab was presented at WSO2Con Africa 2026. Presenters: Menaka Jayawardena, Technical Lead & Head of Engineering, Agent Platform BU, WSO2 Viggnah Selvaraj, Lead Sales Engineer, WSO2 This is the second half of the hands-on AI Agent Manager lab. It picks up where part one left off and keeps the focus on operating AI agents that already work, rather than writing new ones.

This lab was presented at WSO2Con Africa 2026. Presenters: Asanka Abeyweera, Associate Director - Head of Product Management & Architecture, Agent Platform BU, WSO2 Viggnah Selvaraj, Lead Sales Engineer, WSO2 Getting an AI agent to work is only the start. Once it runs, you need to show exactly what it said and why, keep compliance rules in force across agents built by teams that have never met, tell whether quality is improving or slipping, and make sure each agent reaches only the systems it needs.

New in Katalon True Platform this month: Katalon AI Knowledge Base + Conflict Check: Knowledge base for project docs and glossary Linked Jira issues as extra context Conflict check across sources, with no edits made What feature should we ship next? Drop it in the comments.

AI costs are unpredictable and behave nothing like traditional cloud spend. FinOps teams face new challenges: opaque model pricing, and hidden costs in places like prompt design and context length. In this session, our Head of AI, Prajakta, will show you how to bring FinOps discipline to AI costs: how to see them, attribute them, and optimize them. She’ll give you practical levers that can cut AI costs by 20 – 80%.

Connect Google AI Studio to an on-prem SQL Server database without opening a single firewall port or handing over a database password. In this video I put DreamFactory in front of a SQL Server database on my local network, generate a REST API from it, lock that API down with a read-only role and a scoped API key, then I have AI Studio build a live delivery operations dashboard on top of it. AI Studio calls the API from its own Node server, so the key never reaches the browser.

Marketing teams spend more time waiting on ops tickets to segment their database than they do actually running campaigns. By the time the segmentation comes back, the brief has changed, the moment has moved on, and someone is already asking for a revision. In this clip from the Thought Tank: Marketing in the Age of Agents, ThoughtSpot SVP Marketing EMEA Katie Marcham introduces the agent her team built to change that. They call it Slicy McSliceface, and it does exactly what the name suggests.

Wondering which of your AI-driven tests need attention right now? Watch this video to learn how SmartBear BearQ spots flaky tests and failure trends the moment they happen, so you always know exactly where to focus. Each report summarizes what happened and surfaces the issues BearQ detected, whether while exploring your app or running a scheduled test. Drill into failed tests to see exactly which runs broke and why, track trending results over time, and review open issues found across both passing and failing runs.