SW

Sadeepa Wijesekara

Things Sadeepa Says on Podcasts

Where to Find Them

Sadeepa Wijesekara has been a guest on The TWIML AI Podcast .

Recently: “Buying Tokens vs Building an AI Factory with Sadeepa Wijesekara - #785” on The TWIML AI Podcast (November 2026).

What They Said

“Agents are running autonomously. Somebody could add some poisoned information into a ticket or an email or a calendar invite. And then the agent nicely reads that, takes it as a system prompt or instruction and tries to execute it. And that creates a big issue for the organization.” — Sadeepa Wijesekara, The TWIML AI Podcast

Sam asked what managing fleets of agents means for security. Sadeepa lays out the prompt-injection risk for agents that read tickets, email and calendar invites, before describing the layered controls her platform uses.

The TWIML AI Podcast · 2026-11-17 Permalink → Listen →
The TWIML AI Podcast Around 26:25 into the episode
Sadeepa Wijesekara

…I think that's that's if I'm a CTO, I wouldn't spend my time on doing that because my business may be very different from just building hardware. And so that's a call that each CTO needs to make. Like how much should they invest in building the stack that they need to produce the business value?

Sam Charrington

…of agents or swarms of agents, some people call them, like what does it mean to manage all of that? It's certainly something that enterprises are not used as a new set of skills, a new set of expertise that needs to be developed. Like what are you seeing there with regards to that management layer?

Sadeepa Wijesekara

Yeah, so the management layer is definitely needed, especially when you have agents that you don't even know what they're doing, right? Like if you don't have that visibility as a CTO or a CISO, they would be very worried. Like, I don't know what my agents are doing, what kind of systems they are connecting to, what kind of data that they are retrieving. And more importantly, what kind of poisoning that could happen in these retrievals, right? Like agents are running autonomously. Somebody could add some poison information into a ticket or email or a calendar invite. And then agents nicely read that, take that as a system prompt or instruction and trying to execute it. And then create a big issue for the organization. So as a CISO or organization leader should be looking at what kind of a security mechanism you have to control your agents from, it's a depth and defense play. You got to have multiple layers of controls in place. And I mentioned about prompt injection detection. And it used to be, we say, hey, please don't read this or don't write in the prompt…

Sam Charrington

And at least some of what you described sounded a little bit like OpenShell and Secure Shell, a couple of open source projects from NVIDIA. Can you talk a little bit about those and what other components come together to enable you to provide all of this?

Sadeepa Wijesekara

Yeah, so OpenShell is the NUDS sponsored. It's a CNCF certified project. So it's an open project. And that one has a great way to keep the agents in a sandbox. And it allows the network policies to say where it can go, where it cannot go. And that is monitored and controlled by a policy. So that's…

Sam Charrington

Where are organizations in terms of their readiness to adopt all of these various technologies and new ways of thinking in order to do this smoothly without exposing them to new risks that they don't want to take on?

Speaker names come from our own diarization. The position is estimated from where the line sits in the episode.
“In the agreements, the frontier model providers say they are not training with that data. That may be true. But before data gets to training, there are multiple layers it goes through. All the prompts are getting logged, and they may be erasing that too. And then there are some evaluations they run. So they are not training the model, but evaluation might capture that scenario, and then the evaluation is what kicks off and says, this should be the answer. That's the level of detail that's not disclosed.” — Sadeepa Wijesekara, The TWIML AI Podcast

Sam raised recent accusations that a frontier lab trained on a researcher's data and published first. Sadeepa says contract promises not to train on customer data leave open what happens in logging and evaluation. It is her speculation about what providers don't disclose, not a claim about a specific case.

The TWIML AI Podcast · 2026-11-17 Permalink → Listen →
The TWIML AI Podcast Around 41:58 into the episode
Sadeepa Wijesekara

…a data gravity issue? Do you want to send your sensitive data? And these days, data and your internal processes is your IP. And once it goes to the cloud, how sure you are that it is protected? So I think those are the risk questions that you should be asking, not necessarily the cost questions.

Sam Charrington

…side. But then there's this whole other set of factors around privacy and IP leakage and control and like seeding your competitive advantage to a, for many organizations, to a potential future competitor. A lot of things that we heard in the days of cloud, right? That's right. They're analogous.

Sadeepa Wijesekara

And I look at them, they are kind of real in a way, because when you send a prompt out to a GPU, that prompt is visible in clear text. Administrator can just do some commands and dump the memory and see what's going on. So that is a risk if you're willing to take that because the prompts has real, could be a real business scenario that you might want to protect. So if you're ready for that, then you can go to Frontier and it's not a concern for you.

Sam Charrington

Yeah, it kind of strikes me that the closest analogy to this is like the, I don't know what we want to call it, the math gate stuff with OpenAI. Like, you know, there's a point in time where, you know, the, you know, the frontier model provider is going to steal my data as like of this big…

Sadeepa Wijesekara

Yeah. And I think, and they may, I mean, in the agreements, right, with the frontier model providers, they said they are not training with that data. That may be true. But before a model gets to that data or for training, there's multiple layers it goes through, right? So there's definitely going…

Sam Charrington

you're not training on my data, but are you making synthetic data based on my data that you are training on?

Speaker names come from our own diarization. The position is estimated from where the line sits in the episode.
“If you have a $20 subscription to ChatGPT, I'm not going to pick a lower model. I'm just going to use the best one I can. I think that's what's happening in the enterprise. It's like the early days of P-cards, purchasing cards: you can just buy anything and not worry about it. Then the companies put controls in place. Equally, I think that's what's happening with AI tokens, when companies are realizing there's maybe some waste in there.” — Sadeepa Wijesekara, The TWIML AI Podcast

Sam asked whether the recent stories about companies capping internal token budgets are just about spending less, or about using models more smartly. Sadeepa says people default to the biggest model when they don't see the cost, and compares it to how companies first got control of corporate purchasing cards.

The TWIML AI Podcast · 2026-11-17 Permalink → Listen →
The TWIML AI Podcast Around 07:00 into the episode
Sadeepa Wijesekara

…now we have agents in there. Then you equate that to what is the cost of agents, then you have a true financial equation on saying what is the value you're getting or the benefit you get by introducing AI. I think that's not super mature, but that's where the industry is heading to calculate those.

Sam Charrington

…read that is more about be smarter about how you're using the technology and like don't choose the biggest model for all of your work if it doesn't really require it. Does that and is that in line with what you're seeing? And like, how do you see folks start to grapple with that technically?

Sadeepa Wijesekara

I see that all the time. And like when you have when you have access to all these frontier models, you know, as an end user, if I don't know what the cost to the enterprise, I would just use the best one, right? Or just don't bother even, like, why would I change it? Like, if you have a $20 subscription to open OpenAI, ChatGPT, I'm not going to pick a lower model. I'm just going to use the best one I can, right? So, I think that's what's happening in the enterprise. And not only that, like I know I do this and a lot of folks do that. It's like they experiment, right? We just create stuff because it's like we can. So, I think that's probably one of the reasons that it's just like, it's like I'm thinking like early days of P cards, right? Like, you can just buy anything and then it's you don't have to worry about it. P-cards like purchasing cards. And then the companies put controls in place, like, hey, you cannot buy this, you can have to follow these processes to control the cost. So, equally, I think that that's what's happening in the AI tokens when companies are realizing there's a lot, maybe some waste in there. How do we control that? And if you put the onus onto the individual, at least now we have a control…

Sam Charrington

And even before we get to hosted externally versus on-prem, there's this big question about routing and like, how do I make it easier for folks to get the right work to the right model? Like, how are you seeing folks tackle that? Yeah,

Sadeepa Wijesekara

so there's multiple routing technologies or routing vendors out there today that sits in front of an LLM or as a selector when you have a chatbot or something, you can just pick the router as the endpoint and then router selects where to go. So, and the router's intelligence comes in. Okay, it…

Sam Charrington

Like, how do you know what to do when? Like a CIO or AI leader at a large enterprise, you see like your token spend forecast going this way, your CFO saying it needs to go that way or at least flatten out a little bit, or like, where's the asymptote? And you have, you know, a handful of technical…

Speaker names come from our own diarization. The position is estimated from where the line sits in the episode.
“The number of tokens consumed by a chatbot versus an agent is quite different. All the numbers are like 20x to 50x more for agents on token consumption. So even if the per-token price goes down, when you have a 50x increase in the usage, the total bill is going to be up.” — Sadeepa Wijesekara, The TWIML AI Podcast

Sam Charrington asked why falling per-token prices on providers' pricing pages don't show up as lower bills for enterprises. Sadeepa explains that agents burn far more tokens than chatbots, which is why spending keeps climbing.

The TWIML AI Podcast · 2026-11-17 Permalink → Listen →
The TWIML AI Podcast Around 03:53 into the episode
Sadeepa Wijesekara

…for that is running on-prem, could actually save about 60%, up to 60% of the token cost. So that's something that we are looking internally, and our customers are very well aware of that as well. And they're looking for how they can move the workloads appropriately to on-prem versus on-the-cloud.

Sam Charrington

…oh, these are going down. Like the models are getting better, the costs are going down. Is it obvious to the folks that you talk to why that doesn't necessarily reflect in their bill? Like, did they inherently get that? Or do you have to explain like the role of agents and how agents change that?

Sadeepa Wijesekara

I think we a lot of folks are getting it now. You know, just because the number of tokens consumed by a chatbot versus agent is quite different. I think we. All the numbers from like 20x to 50x more for agents on the token consumption. So, even the per token price goes down when you have a 50x increase in the usage, the total bill is going to be up. So, I think that that is the reality that our customers are facing today.

Sam Charrington

What about on the value side? Do you see folks starting to kind of wrap their arms around the value that these agents are bringing to the organization? Like, how do folks grapple with that?

Sadeepa Wijesekara

It is a fantastic question. And I think I'm part of the Tokenomics Foundation board. So, this is the question that we are trying to answer as the board. Just like, how do we give a good framework and ways to measure the value of these tokens? But as you know, the token, each token is not the same…

Sam Charrington

Thinking a little bit about the Apple story and some of the other stories about, again, like companies clamping down on token budgets, I think there's like a surface-level read on those stories that's primarily about spend less. But there's also like this more nuanced read that is more about be…

Speaker names come from our own diarization. The position is estimated from where the line sits in the episode.

Collections They Appear In