Ep #8: Enterprise AI Field Notes: Agents at Work, Live Demo, Guardrails + Responsible Innovation

Описание: Hosts: Srini Annamaraju:   / sriniuk   &
David Royle:   / davidroyle  .

Guest: Ravi Ramchandran:   / raviramchandran  

Welcome to episode 8.
AI agents are getting easier to build. That’s the exciting bit. The risky bit is that organisations can now create weak, badly governed automations before leadership has worked out what “good” actually looks like.

In this episode, Ravi joins Srini and David to pull the conversation out of buzzword-land and into real work. He walks through a practical example of building an agent that turns meeting transcripts into status reports, then digs into what matters underneath: prompt discipline, guardrails, safe experimentation, risk metrics, and why handing people tools without changing operating practice is asking for trouble.

The conversation moves from macro AI noise to enterprise reality. How should leaders think about the 70-20-10 split of routine, experimental, and visionary work? Where does human friction still belong? And how do you encourage innovation without creating a quiet flood of low-quality AI output across the firm?

What we cover

Macro AI reality check - Why the sensible middle matters more than the hype-or-panic cycle.
Productivity is starting to show up - Early signs of measurable uplift are emerging, even if the landing is still messy.
The 70-20-10 work model - How to cut routine work and create more room for experimentation and higher-order thinking.
Innovation becomes everybody’s job - The barrier to building has dropped so far that innovation can’t stay in a corporate side room.
A live agent example - Ravi demonstrates how meeting transcripts can be turned into weekly status reporting.
Why prompts are not enough - One decent output is not the same as a repeatable capability.
Risk metrics for the AI era - Traditional productivity measures are no longer enough.
A seven-day build plan - Ravi shares a practical way to identify, scope, and build useful agents.

Chapters

1. AI noise vs real enterprise adoption
2. Why productivity gains are starting to matter
3. The 70-20-10 model for redesigning work
4. Innovation becomes everybody’s business
5. Live demo: agent for weekly status reports
6. Prompting, grounding, and hallucination risk
7. Guardrails, policy, and engineering practice
8. Risk metrics and trust in production
9. A seven-day framework for useful agents

Top-5 Takeaways

Tools alone do not transform organisations
Agents need boundaries, not vibes
AI risk is now operational risk
Safe experimentation needs leadership air cover
Frameworks beat random enthusiasm

Who it’s for

Enterprise Leaders in all functions inerested in AI adoption.

Help Spread the Word

Enjoyed the episode? Follow us!

Template Takeaways
Ravi has kindly shared these two templates he walked us through for general open access. Please feel free to download them from this Google Drive folder.

https://drive.google.com/drive/folder...