This September, Evonence (a Cloudnyx.ai company) and Cloudnyx.ai, hosted an engineering meetup at DevX, HITEC City, Hyderabad. The event brought together developers, architects, and business leaders to explore the rise of the agentic model of work and how enterprises can practically adopt it with Google Agentspace and the Agent Development Kit (ADK).
Our speakers, Pramod Parmeshwaran (Delivery Head, Evonence), Tushar Patil (Google Cloud Data Engineer, Evonence), and Masiyuddin Khan (AI/ML Developer, Evonence), shared their experiences in building AI agents that deliver value across functions, while addressing enterprise concerns like integration, governance, and security.
As Pramod noted in his session, Agentspace is to enterprise what the browser is to the web. It provides a unified, secure platform to build, manage, and scale AI agents across organizations.
Instead of siloed tools, Agentspace brings together three critical capabilities:
Find: AI-powered search across text, image, video, and enterprise data.
Understand: Reasoning and contextual insights via Gemini and Google-quality search.
Act: Intelligent agents that take actions across connected applications.
With 100+ enterprise application integrations, Agentspace enables workflows that span HR, procurement, marketing, sales, IT, and beyond. The message was clear: agents are no longer isolated chatbots; they are enterprise-grade systems designed to deliver measurable outcomes.
Tushar walked the audience through how Agentspace is engineered to function at scale. The platform’s reliability comes from three pillars:
Connectors: Securely link enterprise data sources like Google Drive, SharePoint, Salesforce, and ServiceNow. This ensures agents have access to real, permissioned enterprise data.
Knowledge Graphs: Unify and contextualize structured and unstructured data into a graph ontology. This enables context-aware, multimodal search across people, documents, and media.
Actions: Go beyond answering queries. Agents can create calendar events, update Jira tickets, or process approvals directly inside connected applications.
Together, these features mean that pipelines matter as much as models. Just as in predictive systems, the reliability of the agentic ecosystem depends on how well ingestion, processing, reasoning, and action-taking are designed.
Masiyuddin highlighted a common concern for enterprises adopting agentic platforms: security and governance. Agents cannot operate in silos; they must comply with organizational security standards.
Agentspace addresses this with:
IAM roles and context-aware access to manage fine-grained permissions.
VPC Service Controls to prevent data exfiltration.
Audit logs and access transparency to ensure accountability.
Encryption and customer-managed keys for sensitive data.
The key takeaway: innovation with agents must move hand-in-hand with enterprise security practices. Without governance, agent adoption cannot scale in regulated industries.
One of the liveliest parts of the discussion was around business use cases. The speakers walked through how Agentspace and ADK are already driving measurable outcomes:
Talent Acquisition: AI-driven sourcing, pre-screening, and interview scheduling.
Marketing: Brainstorming and ranking campaign ideas based on past performance, then generating campaign assets.
Sales: Shortening sales cycles by automating reporting and insights.
IT Operations: Summarizing open tickets and auto-generating stability reports.
Procurement: Drafting performance summaries of vendors and processing invoices.
Early results show ROI in weeks, from reducing manual reporting time by 95% to lowering employee churn by 20%.
For developers, the Agent Development Kit (ADK) was a highlight. ADK provides a modular framework for creating, testing, and deploying custom agents tightly integrated with Google’s ecosystem and Gemini models.
Pramod and Tushar explained how ADK supports both:
Conversational agents (built on Dialogflow and similar tools).
Custom enterprise agents that integrate through Google’s Application Integration and Cloud Marketplace.
The promise is clear: developers can go from local testing to enterprise deployment, with governance, scalability, and observability built in.
The Hyderabad meetup underlined a shift in how enterprises are thinking about AI. It’s no longer about individual models or experiments. It’s about agents that can reason, act, and integrate deeply with enterprise workflows.
The fundamentals are clear:
Build pipelines that connect enterprise data securely.
Leverage knowledge graphs for context-rich insights.
Use ADK to design agents that scale confidently.
Balance innovation with security from day one.
We believe the future of work is agentic, and it is already unfolding.
If you joined us at the event, thank you for contributing to the conversations. If you couldn’t, we hope this recap gives you a practical sense of what was shared. We’d love to hear how your teams are exploring AI agents. Connect with us to continue the discussion.
Author
Nicole de Mocskonyi
Director of Marketing
Cloudnyx
