Enterprise AI startups don’t fail because the tech is weak. They fail because they sell the wrong way.

In this episode of the Alchemist Influencer Series, Ravi Belani sits down with Gnani Palanikumar, former Head of Product at Apigee (acquired by Google for $725M) and chair of Alchemist’s GenAI track.

Gnani shares a tactical, operator-level playbook for turning AI pilots into real enterprise adoption.

They discuss:

• Why traditional enterprise org structures break AI deployments
  

• The mismatch between 2-year AI cycles and 12-month buying cycles
  

• Why founders should sell at the department level, not enterprise-wide
  

• How to find pre-approved budgets and avoid multi-stakeholder gridlock
  

• The “4-week proof” tactic for voice AI in financial services
  

• Why domain expertise and fractional operators matter
  

• How to identify whether a company is AI-native or stuck in SaaS thinking
  

• Why AI projects must deliver value within a quarter
  

Gnani’s 5-word takeaway says it all:

Lead with trust. Deliver proof.

If you’re building enterprise AI, this episode is pure tactical insight.

Agentic AI is everywhere right now, but most teams are still struggling to price it, deploy it, and make it work in production.

In this Fireside Chat, Ravi Belani sits down with Vibhor Rastogi, who leads AI investing at Citi Ventures, to unpack what’s actually happening inside enterprise AI adoption and what founders often get wrong.

They dive into why pricing agentic software is still an unsolved problem, what’s genuinely overhyped in the “agent” wave, and how startups can build defensible AI businesses in a world of reskinned workflows and runaway costs.

Topics covered include:

• Why seat-based, consumption-based, and outcome-based pricing all break in different ways
  

• How CIOs think about predictability, guardrails, and runaway AI costs
  

• What investors really want to see from agentic AI startups
  

• Why many “agents” today are just rebranded APIs and workflows
  

• The importance of memory, learning, and real autonomy in agentic systems
  

• How founders can solve the data cold-start problem through design partnerships
  

• Why an “Agentic Ops” layer may be the missing piece for enterprise adoption
  

If you’re building, investing in, or deploying agentic AI, this episode will help you separate signal from noise.

Enterprise AI isn’t failing because of the models.It’s failing because of data.

In this episode of the Alchemist Accelerator Influencer Series, Ravi Belani sits down with Sidney Rabsatt, Chief Product Officer at MindsDB and former AI and infrastructure leader at Google Cloud, AnyScale, and F5 Networks.

Sidney breaks down why connecting AI to enterprise data is far harder than most teams expect and where the real unlocks are hiding. From messy, fragmented data systems to the false promise of endless ETL projects, this conversation explores what it actually takes to get real value from AI inside organizations.

You’ll learn:

• Why AI struggles when enterprise data is fragmented
  

• The hidden cost of traditional ETL and data clean-up
  

• How AI can work directly on messy data without years of prep
  

• Where startups can find real opportunities at the AI + data layer
  

• Why focusing on the use case matters more than the technology
  

If you’re building, deploying, or investing in enterprise AI, this episode will change how you think about data, workflows, and where real value is created.

Manjeet Singh lives where AI hype meets real enterprise reality. He is the Senior Director of Product Management at Salesforce AI and Agentforce, and previously led product teams at Iron Mountain and ServiceNow. He spends his days shipping AI products that millions of people actually use.

In this episode, Manjeet sits down with Ravi Belani to talk about what is really working in enterprise AI adoption, and where a lot of teams are still stuck in POC land.

They explore:

• The real state of enterprise AI and agentic adoption in 2025

• Why so many AI projects stall at proof of concept

• Where agents are already driving real KPIs, from software development to customer support

• How to tell if an AI opportunity is likely to deliver real ROI for an enterprise

• Why constrained, well defined workflows win first

• Vertical vs horizontal AI, and how to build more than “just an LLM wrapper”

• Why regulated and high cost of failure domains can be great markets for startups

• The missing “toolchain” for agents, from testing and evals to observability and tracing

• How to think about moats, domain expertise, and speed in the AI platform era

• Why Manjeet is excited about the intersection of AI, healthcare, and preventative fitness

If you are building, buying, or investing in enterprise AI, this conversation will give you a grounded view of what works, what does not, and where the next opportunities are likely to appear.

Marcelo Calbucci has been building tech for over 30 years. He’s a serial entrepreneur, multi patent holder, former engineering leader at Microsoft and Amazon, and someone known for having bold, contrarian takes on where AI is actually heading.

In this episode, Marcelo sits down with Ravi Belani to challenge some of the biggest assumptions people have about AI right now.

They dig into:

• Why layoffs and job freezes might have nothing to do with AI

• Why AI isn’t replacing software engineers anytime soon

• Where AI is being massively overused

• The areas where AI is actually underused and full of opportunity

• Why founders need to stop outsourcing strategy to LLMs

• How to use AI to disconfirm your ideas instead of chasing confirmation bias

• Why the next big wave of opportunities may come from “multiplayer AI”

• What roles AI is likely to replace, and which roles will remain essential

• Why adaptability will matter more than anything else in the next decade

Marcelo also shares which startup opportunities he’d pursue today, how he thinks teams will collaborate with AI agents in the future, and the one piece of advice he’d give founders in 6 words: We’re entering the world of adaptability.

If you want fresh, honest, and thought provoking perspectives on AI and the future of work, this conversation is a must watch.

Amr Awadallah has seen multiple waves of hype in tech and helped build real companies through it. He co founded Cloudera, took it public, watched it get acquired for 5.3 billion dollars, and is now building Vectara, a platform that helps enterprises ship AI apps that are accurate, safe and actually usable in production.

In this episode, Amr joins Alchemist Accelerator founder and CEO Ravi Belani to unpack a striking finding from an MIT report. An estimated 95 percent of enterprise AI projects are failing.

They dig into why that is happening and what to do differently:

• The real reasons enterprise AI fails in production
  

• Why “go do AI” from the board is not a strategy
  

• How fake exploration budgets create fake success stories
  

• The 3 whys every founder and buyer should test
  

• • Why do anything

  • Why now

  • Why this vendor
    

• How to tell a real business need from AI theater
  

• Why passion for the problem, not the tech, drives grit
  

• How cofounders, early hires and investors make or break your company
  

• What Vectara is building as an “operating system” for AI agents
  

• Why context, security, accuracy and cost control matter more than demos
  

Amr also shares raw, founder level advice. From firing fast in the first 50 hires, to choosing investors who have actually been in the trenches, to his simple test for whether you really love the problem you are working on.

If you care about building AI products that survive past the pilot stage, this conversation is a must listen.

Dr. Shelby Heinecke, Senior AI Researcher at Salesforce, joins Ravi Belani to explain why the future of AI will not belong only to giant models with hundreds of billions of parameters.

Shelby makes the case for small language models: compact systems with only a few billion parameters that can run faster, cost less, protect privacy, and still perform at a very high level when they are trained well on focused tasks.

In this episode, they dig into:

• Why small models are a different tool, not a weaker version of large models
  

• How fine tuned small models can beat much larger models on specific agentic tasks
  

• Where small models shine most: privacy, speed, cost to serve and on device use cases
  

• How Salesforce built “Tiny Giant,” a 1B parameter model that outperforms much larger models on selected tasks
  

• What really matters in training: data quality, workflows and trajectory style datasets
  

• How synthetic data, noise and guardrails help make models more robust in the real world
  

• Why founders should look closely at on device AI and domain specific small models
  

Shelby also shares practical advice for founders who want to build in the small model space, and closes with a simple takeaway: do not underestimate small models.

If you care about AI agents, privacy, edge computing or future startup opportunities, this conversation will give you a lot to think about.

Arjun Prakash, co founder and CEO of DistylAI, joins the Influencer Series to talk about one of the biggest shifts happening in tech right now. We all talk about generative AI, but Arjun breaks down the part that actually matters. Intelligence. Reasoning. And the rise of true AI assistants that can make decisions, take actions, and run workflows across a business.

Arjun walks us through the three chapters of this story:

1. What AI agents really are and why reasoning changes everything.

2. How AI native companies will operate with far lower costs and far more speed.

3. Why this shift will lead to a world where both massive companies and tiny startups can scale faster than ever.

He also gives clear examples of what happens when companies embrace new technology and what happens when they ignore it. From Carnegie Steel to Borders, the winners are always the ones who lean in early.

If you care about innovation, productivity, or the future of work, this episode will change how you see the next decade.