Demo: AI Agents for Pharma & Life Sciences

Summary

In this product demo, Vinod Iyengar, Head of Product at Tellius, showcases how AI agents are transforming data analytics—especially for commercial pharma teams.

With agentic AI workflows, you can automate complex, multi-step analyses like brand performance, HCP targeting, and territory optimization—without writing a single line of code. See how sales managers can go from dashboards and manual queries to proactive insights delivered every Monday with natural language-powered AI agents.

✅ Instantly analyze NRx and TRx performance

✅ Identify underperforming territories and specialties

✅ Automate weekly field insights reports

✅ Trigger multi-step workflows using natural language

✅ Dive deeper with root cause and statistical analysis

Whether you're in pharma analytics, field sales leadership, or a data-driven role looking to scale insights—this is a game changer.

Transcription

 Hi, my name is Vinod Iyengar. I'm the head of product at Tellius. Today I'm really excited to talk to you about agents. AI agents have been the buzz all around the market for the last few weeks. Even in the space of analytics, we've seen that AI agents can be extremely powerful in doing complex analysis, finding out insights about your business, and doing these multi-step workflows, which could take days, if not weeks, for analysts.

And now we can do it in a matter of minutes, not hours, all in the platform. This is the power of AI agents.  We announced this a few weeks ago, and I want to show some examples of what's possible with these new AI agents.

And just to describe a very high level, AI agents are defined as workflows where you have access to an LLM, access to the semantic layer with Tellius, which understands your data, which understands your knowledge, your enterprise key terms, synonyms and learnings. Use all of those to solve complex  workflows, which are defined by you, all in natural language.

I'm gonna show you all of those things in a minute. Okay? So let's jump into the demo really quick. 

So, here you see the Tellius interface. When you come to Tellius, click on Kaiya on the left. You'll be logged into our Kaiya interface, which basically allows you to ask questions of your data and get natural language answers. There's a whole bunch of demos around that. Feel free to watch them.

But I want to jump to agents, which is our new offering. With agents, what we've done is we've taken the power of Kaiya and added multi-step analysis and workflows. So now, imagine you are a sales manager for a large pharmaceuticals company. You are often looking at, how is my drug or my brand performing as compared to others in the market?

Which territories are doing the best? Which doctors or HCPs do we need to call? And who's prescribing our drug versus some of our competitors? And what's the overall trend?

That's the analysis you do. Day in, day out, if not every week. Typically what happens is you will create that dashboard, or you might go and ask multiple queries. But imagine you can have a system, an AI agent, that can go do all this work for you and tell you what you need to do at the beginning of the week or beginning of the day. That'd be so powerful. That's exactly what we're gonna show today. So here are a bunch of agents that are available in the platform, and I'll show you how easy it is to create them. But imagine you could be asking questions about market access or national performance, right? Or you could be doing analysis around NRx, just new to prescription, new to brand prescriptions. You could look at state wise analysis and so on. So let's pick one. For the purpose of the demo, so you're gonna look at NRx Deep Dive, which is new prescriptions, and you want to do brand performance at the subnational level. This is the goal of the agent.

And here we specify what data it has access to, because that's the data it's gonna be using in the model for its analysis. Once that is done, we can look at the steps and the steps is where you actually define what kind of analysis you want to do. This is the Composer interface where you can actually define your Agentic Workflows. You build your agents.

And here, you can describe basically, in natural language, what you want to do. So you're asking the end user, hey, which brand do you want to do the analysis for? It could be your own brand or it could be competitor brand. You can then ask for which metric. In this case, so that's a TRx, total prescriptions, NRx, which is new prescriptions, and then calls, which is how many calls do you make to those physicians or doctors. You can then look at change in TRx, or the metric that you picked, over the last four weeks and the selected brand. You can look at the change over the last 13 weeks. You can look at the change in brand and then look at the change in NRx by the last four weeks and the district name. You can define the number of steps you want the system to do and you can update it. It's super easy to do this, all in natural language. And more importantly, you can also specify which triggers you want this agent to run out. So in this case, we specify these three triggers.

So when you ask these questions, the agent knows that, okay, I'm gonna do this analysis for these particular triggers. As accessible as that. Once it's done, you save, hit and publish. You can also test this in the playground mode. Which is great because before you publish an agent, you want to actually validate, make sure that it's actually doing what it's supposed to do.

Okay? And once it's done, you can save and publish. Now let me jump ahead and show you how we can trigger this agent from the Kaiya interface. So if you're an end user, you come in and let me pick this view over here, right? And you want to run this query. Analyze brand performance.

And what this will do is actually trigger the agent here and say that, hey, here's an agent that is built for answering this question, I can hit run and it'll go with the analysis but in the interest of time, I'm gonna show you, I ran this earlier to show you what happens.

So we run a similar performance analysis for like brand performance. In this case we want to pick a particular brand and see how it performs over time. So I add this, it says, okay, let me do NRx analysis for you. Deep dive. The agent asks for the question, which brand are we looking to analyze? Here we pick Mycobutin. This is one of the brands we have in our demo data set. And then it's asked for which metric you care about. And we pick the metric over here, NRx, and then it does the analysis. It says, okay, last 13 weeks. Comparison for a Mycobutin particular drug. We look at the analysis we see in each of those steps, you can always go and look at the query inspector, which will tell you what is this query and what does it mean.

So if you have ever any doubts on, okay, did the model or the agent do what it was supposed to do, you can always go in and verify, so that's the nice thing about everything Tellius. You will see the analysis over here. You can see the SQL as well, right? So then it looks at NRx by district name over the last four weeks.

So you can see which districts are actually doing the worst, which territories are actually the worst performing regions, right? If you are a sales manager, you really need to know this information so that you can go back and target those, right? And after that you can look at that switch, change in territory.

For those last four weeks as well, right? So for those territories, this is by NRx name, and then you can look by territory name as well. So there are slight differences between territory and districts. You can also trigger what we call an insight. So insight is a tool in Tellius that allows you to do a statistical analysis that drills down and finds the reasons for why a change is happening.

There is a trend, there's a drop, in the NRx for this particular brand. And there's, it's a small drops 5%, but still meaningful, right? So then it goes with the analysis runs in, what you call an insight, and says what are the key contributors for that?

And it found that in this case, certain specialties, in this case, dermatology, rheumatology, these particular specialties seem to have an overall impact. They have reduced the volume, and that has contributed to the overall decline in NRx, right? Similarly, certain regions are underperforming, certain PSOs.

It'll also look for things that change. Sales is all always not like a simple story. There are declines in some parts and increases in some parts, and the net of it is what you need to know. So okay, some cases actually increased, so you can see certain deciles or certain territories have actually done better.

But overall performance has gone down. And it gives you a summary. And the nice thing is it then puts all the analysis we did so far, like the charts, we did the NRx over time, by brand, by territory, and the insights, and put them into a nice report. This kind of report, if I'm a sales manager, I want to see this every Monday, right, before I even get started on the week. What happened last week and what do we need to do this week, right?

So it tells me what are the key contributors, what are the positive contributors, the negative ones, and what is the total focus? So it tells me focus on the positive areas. Obviously make sure that the key areas where we are declining, we go after them, find out why we are not doing well in certain specialties.

That's an area of concern, let's go and analyze it. Gives me a very quick report. This kind of like a report that I might ask my analysts to do. It'd be nice to have, but if we can do this every Monday, that's super powerful. So that's just a very high level glimpse of what we can do with AI agents. This is available in Tellius 5.4. Go ahead, go to Tellius.com and sign up for a free trial.  You can play with the product. We'd love your feedback. Or if you are an existing Tellius customer, or you're doing a POC, talk to your CSM, and you can get access to this as well.

Would love your feedback. Thank you.

‍