YouTube Video
Google Antigravity → https://g.dev/ai/google-antigravity
Vibecode and Secure an AI Agent Lifecycle with Antigravity and TDD → https://g.dev/cloud/vibecode-codelab
The new SDLC whitepaper → https://g.dev/cloud/new-sdlc-docs
While coding agents have made writing code virtually effortless, generating lines of code was never the hardest bottleneck in the software development lifecycle (SDLC). The real challenge begins post-merge: reviewing changes, navigating production context, deciphering why systems page you at 3:00 AM, and verifying that new updates do not trigger silent regressions. In this episode of The Agent Factory, we explore what it takes to extend AI beyond the IDE and automate the entire operational loop in real-world production environments.
Shahram Anver, CEO of Cleric AI and former ML platform lead at Gojek, joins to break down why production-grade agents require organizational memory and live system context rather than raw code-generation capability. Shahram walks through live, end-to-end demonstrations using Antigravity paired with Cleric AI. You'll see how an agent inspects live Kubernetes logs, queries an auto-updating production knowledge graph to identify a missing product issue, writes the fix, generates regression tests, and opens a PR.
Chapters:
0:00 - Why Coding Is Solved, but Production Is Hard
1:05 - Why Production-Grade Agents Took Longer Than Coding Agents
6:56 - Demo Overview: Reactive vs. Proactive Problem Solving
8:30 - Live Demo: Antigravity Diagnosing Kubernetes Logs
9:39 - Inside Cleric: Real-Time Production Context & Knowledge Bases
15:03 - Autonomous Post-Merge Monitoring with Cleric & Slack
18:39 - Organizational Best Practices: How Clean Context Empowers AI
20:18 - The "Dark Factory" Vision for Software Engineering
23:47 - How to Onboard AI Agents Like Junior Engineers
25:16 - The Cost of Verification Across the SDLC
29:01 - Rapid Fire: Dark Factories, Runbooks & Staging Environments
31:26 - Shahram's Core Framework: First Principles Thinking
More resources:
Cleric.ai → https://g.dev/ai/fix-agents
Antigravity Remote Control → https://g.dev/ai/remote-control
Antigravity CLI → https://g.dev/ai/antigravity-cli
Speaker: Shahram Anver (CEO, Cleric AI)
LinkedIn → https://goo.gle/Shahram-on-LinkedIn
Host: Smitha Kolan
YouTube → https://goo.gle/Smitha-on-YouTube
LinkedIn → https://goo.gle/Smitha-on-LinkedIn
X → https://goo.gle/Smitha-on-X
Watch more of The Agent Factory → https://www.youtube.com/watch?v=qBOvM7SiDa4&list=PLIivdWyY5sqLXR1eSkiM5bE6pFlXC-OSs
🔔 Subscribe to Google Cloud Tech → https://goo.gle/GoogleCloudTech
Speakers: Smitha Kolan, Shahram Anver
Products Mentioned: Antigravity, Model Context Protocol, Google Kubernetes Engine, Gemini 3.7 Flash
Vibecode and Secure an AI Agent Lifecycle with Antigravity and TDD → https://g.dev/cloud/vibecode-codelab
The new SDLC whitepaper → https://g.dev/cloud/new-sdlc-docs
While coding agents have made writing code virtually effortless, generating lines of code was never the hardest bottleneck in the software development lifecycle (SDLC). The real challenge begins post-merge: reviewing changes, navigating production context, deciphering why systems page you at 3:00 AM, and verifying that new updates do not trigger silent regressions. In this episode of The Agent Factory, we explore what it takes to extend AI beyond the IDE and automate the entire operational loop in real-world production environments.
Shahram Anver, CEO of Cleric AI and former ML platform lead at Gojek, joins to break down why production-grade agents require organizational memory and live system context rather than raw code-generation capability. Shahram walks through live, end-to-end demonstrations using Antigravity paired with Cleric AI. You'll see how an agent inspects live Kubernetes logs, queries an auto-updating production knowledge graph to identify a missing product issue, writes the fix, generates regression tests, and opens a PR.
Chapters:
0:00 - Why Coding Is Solved, but Production Is Hard
1:05 - Why Production-Grade Agents Took Longer Than Coding Agents
6:56 - Demo Overview: Reactive vs. Proactive Problem Solving
8:30 - Live Demo: Antigravity Diagnosing Kubernetes Logs
9:39 - Inside Cleric: Real-Time Production Context & Knowledge Bases
15:03 - Autonomous Post-Merge Monitoring with Cleric & Slack
18:39 - Organizational Best Practices: How Clean Context Empowers AI
20:18 - The "Dark Factory" Vision for Software Engineering
23:47 - How to Onboard AI Agents Like Junior Engineers
25:16 - The Cost of Verification Across the SDLC
29:01 - Rapid Fire: Dark Factories, Runbooks & Staging Environments
31:26 - Shahram's Core Framework: First Principles Thinking
More resources:
Cleric.ai → https://g.dev/ai/fix-agents
Antigravity Remote Control → https://g.dev/ai/remote-control
Antigravity CLI → https://g.dev/ai/antigravity-cli
Speaker: Shahram Anver (CEO, Cleric AI)
LinkedIn → https://goo.gle/Shahram-on-LinkedIn
Host: Smitha Kolan
YouTube → https://goo.gle/Smitha-on-YouTube
LinkedIn → https://goo.gle/Smitha-on-LinkedIn
X → https://goo.gle/Smitha-on-X
Watch more of The Agent Factory → https://www.youtube.com/watch?v=qBOvM7SiDa4&list=PLIivdWyY5sqLXR1eSkiM5bE6pFlXC-OSs
🔔 Subscribe to Google Cloud Tech → https://goo.gle/GoogleCloudTech
Speakers: Smitha Kolan, Shahram Anver
Products Mentioned: Antigravity, Model Context Protocol, Google Kubernetes Engine, Gemini 3.7 Flash
SOCIAL SHARE CARD GENERATOR