The Ravit Show

Last week Dreamforce was a blast!!!! 2024 was the year of AI pilots. 2025 was the year agents went into production. 2026 is about scaling them. That framing came up in my Dreamforce conversation with Vala Afshar, Chief Digital Evangelist at Salesforce, and Gurvinder Singh Sahni, CMO at Altimetrik on The Ravit Show.

Some numbers from the discussion:

Dreamforce 2026 brought 50,000 people in person from 150 countries, with 10 million tuning in for the keynote. The event generated over $100 million for the San Francisco Bay Area economy and supported 60,000 local jobs.

Across 400 early adopting companies studied over 15 months, average agent usage grew from 5 to 15, deployment time dropped to under a week, and utilization went up 4x.

What stayed with me:

Vala called AI the most important technology of our lifetime, the electricity of the 21st century. The pressure on businesses now is to actually become agentic enterprises, not just talk about it.

The hard part is what comes next. Going from dozens of agents to hundreds or thousands turns governance, security, compliance, accountability, and trust into the real problem to solve.

On keeping up with the pace, the advice was to stay a beginner. Ask questions, be radically transparent, and build a culture where people can experiment without fear of being blamed when something does not work.

And none of this happens alone. Strong engineering partnerships and ecosystem collaboration, like the one between Salesforce and Altimetrik, are what make these transformations land.

Full interview below.

Thank you Vala and Gurvinder for a great conversation.

#dreamforce #salesforce #agentforce #altimetrik #ai #aiagents #enterpriseai #theravitshow

Most enterprises have already spent heavily on Salesforce. The real question now is how to get more value out of it. That was the heart of my conversation at Dreamforce with Bryan Schappell from Emerson and Nripendra Tamang from Altimetrik.

We got into what it actually takes to move Salesforce and Agentforce from pilots into production, and where AI is starting to change how companies engage customers and run their operations.

A few things worth watching for:

- How Emerson's Salesforce journey has evolved, and the business priorities shaping what comes next

- Where Agentforce and generative AI can create real impact in their industry, beyond the hype

- The capabilities that matter most when taking AI initiatives out of experimentation and into day to day operations

- How Altimetrik brings together Salesforce expertise, AI native engineering, industry knowledge, and its partner ecosystem to help enterprises get there.

And Bryan's 1 recommendation for business leaders planning the next phase of their Salesforce and AI journey.

If you are leading a Salesforce or AI initiative right now, this one is worth your time.

Thank you Bryan and Nripendra for a great conversation.

#dreamforce #salesforce #agentforce #altimetrik #emerson #ai #enterprise #theravitshow

Most enterprises are not held back by AI models. They are held back by their data. That was the focus of my conversation with Savinay Berry, EVP and GM of Data Management, at @Salesforce’s Dreamforce. #AdSavinay also highlighted common data readiness mistakes. We explored what AI-ready data means for Salesforce customers and why enterprises need an AI infrastructure layer built to outlast today’s AI wave.

As Savinay put it, “I don't know the agents that our customers are going to create in the future, but I want a canvas, a platform, a harness that I can have which is truly compliant, secure and protected. And there's probably going to be only two or three of them as it happened in case of the cloud.”

We closed with an example of AI-ready data driving change at scale.

Worth a listen for teams getting started.

Thank you, Savinay.

Full interview is live. Watch it and tell me what stood out to you.

#df26 #salesforce #datacloud #agentforce #datamanagement #aireadydata #data #theravitshow #SalesforcePartner

AI is moving out of the back office and into the products people use every day. That was the thread running through my conversation with Velia Carboni, CIO of SharkNinja, here at Dreamforce on The Ravit Show. #sponsored

SharkNinja is taking AI beyond internal productivity and into its products and customer experiences. That shift changes what a CIO is responsible for. Technology leadership now sits much closer to the customer. We talked about where Salesforce and Agentforce fit into the way SharkNinja connects with its customers.

We also got into speed. SharkNinja launches new products across global markets, and technology has to keep pace with that.

And of course, what Velia is hearing on the ground at Dreamforce this year.

The energy here is unreal. Conversations like this one are exactly why I love being at Dreamforce.

Thank you, Velia, for a great conversation.

#dreamforce #salesforcepartner #salesforce #agentforce #sharkninja #ai #cio #customerexperience #theravitshow

Most companies don't have a data problem. They have a data readiness problem. They already have plenty of data. What they don't have is data an AI task can actually use. I sat down with Ho Bae, CEO of CUBIG on The Ravit Show, to talk about what it takes to make enterprise data ready for AI. Here are my takeaways:

1. Having data and having AI-ready data are 2 different things. The right information may already be in your systems, but it often needs to be prepared, changed, or validated before AI can use it.

2. "AI-ready" depends on the task. Data isn't ready in general. It's ready for a specific job.

3. Real enterprise data is messy in ways sample data isn't. Circuit IDs, project codes, and node names only make sense because of how they connect inside the business.

4. Removing too much data can break the AI. If you strip out the relationships, the model can't tell that 2 records belong to the same asset or event.

5. The answer should come back inside the workflow. LLM Capsule sits inside the tools teams already use and returns results with the real business values, ready to act on.

6. You can keep your current model. The model stays your choice. The data layer around it is what changes.

7. Start with 1 blocked AI task, not a product checklist. Know what you want to run, what's stopping it, and what success looks like.

Ho also walked through a live demo, showing how 2 network alerts get traced back to the same device without the model ever seeing the original device name.

If your AI works on sample data but struggles on real data, this one is worth your time.

CUBIG will also be at Oracle AI World 2026 in Las Vegas, October 25 to 28.

#data #ai #enterprisedata #dataengineering #genai #llm #cubig #theravitshow

What does it actually take to make AI work inside the enterprise? I recently sat down with Philip Miller from Progress to talk about the challenges companies are facing as they move AI from experiments into real production environments. We talked about Progress Data Platform, PDP Studio, context layers, AI guardrails, data governance, AI agents, cost optimization and what enterprises need to build AI they can actually trust.

We also discussed a real-world use case where report-writing time was reduced from around 90 minutes to just minutes.

One thing that stood out to me:

AI is not just about the model.

The data, context, governance and infrastructure around the model matter just as much.

#Progress #EnterpriseAI #AI #Data #AI #API #Agents #GenerativeAI #theravitshow

Last week, I had a blast at 1st Equinix Horizon. Sat down with Paul Dehnert, Vice President of Sales at Equinix on The Ravit Show. He runs the global sales team behind Equinix's interconnection portfolio, the internal wiring inside data centers that quietly makes up the network of networks we call the internet.

Sharing the takeaways while they are fresh.

The theme running through this entire event is open neutrality, and Paul connected that directly to what is happening in AI and inference right now. Together AI is bringing its inference platform onto Equinix to make open source models easier to run. Some hyperscalers are rolling out new interconnection capabilities on Fabric One that let developers connect directly using natural language instead of manual network configuration.

The line that stood out to me most was about enterprise spend. A lot of companies burned through their token budgets earlier this year and are now actively optimizing. Paul's read is that enterprises are getting more mature about this. They are realizing they do not need the frontier model for every use case. The real skill now is picking the right venue and the right model for the job, not defaulting to the biggest one available.

That is a quieter shift than the big keynote announcements, but it might matter just as much for how enterprise AI actually gets budgeted and run.

More conversations from the summit floor coming soon.

#equinix #equinixhorizon #ai #interconnection #enterpriseai #cloud #digitalinfrastructure #openneutrality #theravitshow

Great conversation from Equinix Horizon, this time with Kaladhar Voruganti, VP and Senior Technologist on the Global Value Advisory Team at Equinix on The Ravit Show. He runs field CTOs and 20 global solution validation centers, basically AI labs where customers and partners go to test what actually works.

Sharing the takeaways while they are fresh.

Most enterprises are not chasing exotic AI use cases. Coding, customer support, and employee productivity are where the real adoption is happening right now. And the architecture pattern Cal is seeing repeatedly is hybrid. Big cloud models for the hard problems, open models on private infrastructure at Equinix for everything else, mainly for cost and control.

The part that stuck with me was how enterprises are managing AI spend and governance. AI gateways act as traffic routers, sending simple queries to simpler models and complex ones to the cloud. Semantic caching alone is knocking out 20 to 60 percent of redundant queries before they ever hit a model. And AI firewalls are showing up at neutral interconnection points to scrub traffic before it reaches applications. This is the unglamorous plumbing that decides whether enterprise AI is actually affordable.

His advice for leaders was blunt in a good way. Centralized data lakes do not hold up for agentic AI, so shift to a federated approach that takes the query to the data instead of the other way around. Do not rush to prove ROI or cut headcount early. Let people get comfortable with AI first. And follow the sequence: experiment, then govern, then move to private AI clusters for cost control.

Kaladhar and his team publish weekly on networking, data strategy, governance, and industry verticals. Worth a follow if you are building any of this yourself.

More conversations from the summit floor coming soon.

#equinix #equinixhorizon #ai #enterpriseai #datastrategy #aigovernance #digitalinfrastructure #equinix #theravitshow

Always a blast chatting with Harmeen Mehta, Chief Digital and Innovation Officer at Equinix, live from Equinix Horizon on The Ravit Show. It is always fun meeting her and chatting with her, she is one of the best.

Sharing the biggest takeaways while they are still fresh.

Equinix Horizon itself is being positioned as an iconic customer event, built around co-creation, not just announcements. That framing matters and it showed in how Harmeen talked about the news.

On Inference Exchange, the 360 degree partnership with Together AI and NVIDIA, her framing stuck with me. AI is not staying in a lab or a dashboard. It is going to meet us and come to work for us every single day, through agents exchanging intelligence in the background. That is a very different picture of AI adoption than most enterprises are planning for right now.

On Fabric One, she broke it down simply. Anything to anything, anywhere to anywhere, connected in hours or minutes instead of weeks. When your infrastructure partner talks about connectivity in minutes, that changes what enterprise IT teams can actually promise the business.

The line that stayed with me most was about velocity. Change across this industry is accelerating because of AI, and organizations have to evolve how fast they decide and execute, not just what they decide. Her advice to enterprise leaders was direct. Embrace the velocity. The decisions made today are what will separate companies tomorrow.

Great conversation, great energy on the ground here at Equinix Horizon. More from the summit floor soon.

#equinix #equinixhorizon #ai #leadership #innovation #digitalinfrastructure #cloud #aiagents #theravitshow

This one is so so important!!!! I spoke to with Madhu Narasimhan, CIO at DaVita Kidney Care, at Glean:GO 2026 in San Francisco on The Ravit Show. Most companies ask whether AI is good enough to scale. A healthcare CIO has to answer a harder question first. What happens when it is not.

That difference ran through the whole conversation.

DaVita operates at real scale in kidney care. When you are that close to patient care, the distance between a promising pilot and something you put in front of clinical and operational teams is not a technology gap. It is a trust gap, and healthcare defines that word more strictly than most industries do.

Here is what we got into.

- What has to be true before you scale AI across healthcare. Not the model benchmark. The conditions underneath it. This is the question I wish more AI conversations started with, because in a regulated environment the preconditions are the strategy.

- Why connected company knowledge decides whether AI is useful at all. An assistant that cannot see how your organisation actually works will confidently give you an answer that is wrong in a way nobody catches. Scattered knowledge is not an inconvenience. It is the failure mode.

- Where AI is creating real value at DaVita today. We stayed on what is actually running, not what is on a roadmap slide. That is a rarer conversation than it should be.

- What role Glean plays in the broader AI strategy, and where a context layer sits relative to everything else being built.

- What moves employees from trying AI to genuinely trusting it. This was my favourite part. Adoption is not a licensing problem or a training problem. People start trusting a system when it is right about something they already know the answer to, repeatedly, inside their own workflow. That is much harder to engineer than a demo.

- And what will separate the companies compounding real value from the ones sitting on 40 disconnected pilots. My own view going in was that the dividing line is knowledge, not models. I came out more convinced.

Madhu, thank you for the time at Glean:GO. Conversations with operators running this at real scale are the ones I learn the most from.

Worth your time if you are trying to move AI out of pilot phase in a regulated industry.

If you are inside healthcare, finance or another regulated environment, what had to be true before you scaled AI beyond a pilot?

#data #ai #enterpriseai #healthcareai #cio #aiadoption #digitalhealth #aigovernance #theravitshow