EPISODE · Feb 16, 2026 · 28 MIN
FinOps Lessons for the AI Era | Best practices from a startup CEO
from Mala's Podcast · host Mala Ramakrishnan
Puneet Gupta, Founder & CEO of Amberflo, joins Mala Ramakrishnan to unpack one of the biggest blind spots in AI adoption: cost, governance, and tying LLM usage to real business value.A former General Manager at Amazon Web Services and founding member of Oracle Cloud Infrastructure, Puneet brings deep cloud infrastructure experience to the AI era. In this conversation, he explains why token usage is exploding, why experiments don’t translate cleanly into production, and why enterprises must “get ahead of the bill” with AI governance.The episode dives into AI gateways, attribution models, multi-cloud realities, usage-based pricing, and the leadership discipline required to scale startups without chasing false momentum.If you're a founder, CIO, or operator deploying AI at scale, this episode is a practical guide to managing cost, governance, and long-term strategy.________________________________________________________Timestamps00:00 Intro and Puneet’s background01:00 Early days at AWS and cloud lessons02:00 The origin story of Amberflo03:00 Inside AWS metering and billing systems04:30 Usage-based business models in the cloud era06:00 From billing to AI governance and FinOps07:30 Moving from AI experiments to production09:00 Why LLM token usage explodes with agents10:30 Tying AI spend to business outcomes11:45 Why parsing vendor bills isn’t enough12:30 The rise of AI gateways14:00 CIO use cases: cost control and developer tooling15:30 Product-facing AI attribution challenges16:30 Context windows and unpredictable token costs17:30 Founder advice: don’t confuse momentum with progress19:00 Choosing the right design partners20:30 Why founders must focus on a niche22:00 Lessons from building metering before billing24:00 Is AI overhyped? What will last25:30 Multi-cloud strategy and competing with hyperscalers27:00 Final thoughts on AI infrastructure opportunity________________________________________________________🔗 Connect with Puneet Gupta → https://www.linkedin.com/in/puneetguptausa/🔗 Connect with Mala Ramakrishnan → https://www.linkedin.com/in/malaramakrishnan🎧 Subscribe to the podcastYoutube: https://www.youtube.com/channel/UCnL3D6aI60R-cCvypFIkh4ASpotify: https://open.spotify.com/show/1NIDE4cT5fuVjC3HaGbCDK?si=eoSTNmhqQNKHQl9d1Jx0UQApple Podcast: https://podcastsconnect.apple.com/my-podcasts/show/talking-to-the-leaders-of-ai-the-ceo-series-with-mala/22ad2bae-502b-4eb1-aa74-53bc79754db7Visit our Website: www.malaramakrishnan.com | www.founderscreative.ai
What this episode covers
Puneet Gupta, Founder & CEO of Amberflo, joins Mala Ramakrishnan to unpack one of the biggest blind spots in AI adoption: cost, governance, and tying LLM usage to real business value.A former General Manager at Amazon Web Services and founding member of Oracle Cloud Infrastructure, Puneet brings deep cloud infrastructure experience to the AI era. In this conversation, he explains why token usage is exploding, why experiments don’t translate cleanly into production, and why enterprises must “get ahead of the bill” with AI governance.The episode dives into AI gateways, attribution models, multi-cloud realities, usage-based pricing, and the leadership discipline required to scale startups without chasing false momentum.If you're a founder, CIO, or operator deploying AI at scale, this episode is a practical guide to managing cost, governance, and long-term strategy.________________________________________________________Timestamps00:00 Intro and Puneet’s background01:00 Early days at AWS and cloud lessons02:00 The origin story of Amberflo03:00 Inside AWS metering and billing systems04:30 Usage-based business models in the cloud era06:00 From billing to AI governance and FinOps07:30 Moving from AI experiments to production09:00 Why LLM token usage explodes with agents10:30 Tying AI spend to business outcomes11:45 Why parsing vendor bills isn’t enough12:30 The rise of AI gateways14:00 CIO use cases: cost control and developer tooling15:30 Product-facing AI attribution challenges16:30 Context windows and unpredictable token costs17:30 Founder advice: don’t confuse momentum with progress19:00 Choosing the right design partners20:30 Why founders must focus on a niche22:00 Lessons from building metering before billing24:00 Is AI overhyped? What will last25:30 Multi-cloud strategy and competing with hyperscalers27:00 Final thoughts on AI infrastructure opportunity________________________________________________________🔗 Connect with Puneet Gupta → https://www.linkedin.com/in/puneetguptausa/🔗 Connect with Mala Ramakrishnan → https://www.linkedin.com/in/malaramakrishnan🎧 Subscribe to the podcastYoutube: https://www.youtube.com/channel/UCnL3D6aI60R-cCvypFIkh4ASpotify: https://open.spotify.com/show/1NIDE4cT5fuVjC3HaGbCDK?si=eoSTNmhqQNKHQl9d1Jx0UQApple Podcast: https://podcastsconnect.apple.com/my-podcasts/show/talking-to-the-leaders-of-ai-the-ceo-series-with-mala/22ad2bae-502b-4eb1-aa74-53bc79754db7Visit our Website: www.malaramakrishnan.com | www.founderscreative.ai
NOW PLAYING
FinOps Lessons for the AI Era | Best practices from a startup CEO
No transcript for this episode yet
Similar Episodes
Mar 26, 2026 ·1m
Jan 2, 2026 ·47m
Dec 21, 2025 ·46m