Artificial intelligence (AI) is becoming ubiquitous – from generative AI applications like ChatGPT and Midjourney, to smart home assistants, self-driving cars and internet chatbots, the technology has permeated every aspect of our work and lives. AI is also a major component of Within3’s Insights Management Platform. But how is our AI technology uniquely suited to support pharma teams in the insights reporting process?
In this article, we caught up with Jason Smith – Within3’s Chief Technology Officer, AI and Analytics – to explore Within3’s history with AI, examine how AI-powered insights reporting works, and discuss how Within3 uses AI in tandem with clients’ strategies to surface powerful insights.
A history of AI at Within3
In 2021, Within3 acquired rMark Bio – an AI company specializing in key opinion leader (KOL) and stakeholder insights, where Jason Smith was CEO and co-founder. The technologies Jason and his team developed at rMark were to become the foundation of Within3’s AI-powered insights reporting.
Initially, our AI solutions focussed on gathering field insights and amalgamating academic research papers to provide medical affairs teams with insights into the current state of science. This developed into the use of transformer technology (a means of applying context to text analysis) capable of applying sentiment analysis to field observations – providing insights into how experts actually felt about particular topics and concepts, positive or negative.
This process has since evolved even further in the three-and-a-half years rMark was acquired:
“We can now look at large amounts of data without knowing what’s in it, and using unsupervised learning, cluster around key topics and concepts and then apply our transformer models to really go deep and suss out those concepts to summarize them back for the reader.”
– Jason Smith, CTO, Within3
How AI-powered insights reporting works
We’ve now progressed into the ‘third generation’ of AI-powered insights reporting. The AI-enabled insights dashboards many life science teams will be familiar with have paved the way for more sophisticated insights solutions. These solutions go beyond simply reporting data, and now provide context and recommendations that refer directly to a users’ strategy.
In fact, this third generation of AI-powered insights reporting can be said to ‘understand’ a user’s strategy, as well as the nuances between the different data streams it accesses – whether from social media, from the client’s CRM, from field medical interactions or congress reporting. The system then outputs this information as digestible insights that can be directly actioned to support or inform a user’s strategy.
“Life science has been inundated with data – we couldn’t separate ourselves from it. But now that we can do this very fast and very efficiently, we have to take that next leap. And that next leap is really contextualized AI – to contextualize results for our customers around their strategies.”
– Jason Smith, CTO, Within3
To take the next great leap, AI-powered insights reporting technologies like Within3 had to build on open-source tools like OpenAI. These ‘off-the-shelf’ large language models are great at what they do, but they’re trained using a lot of broad data, and lack the specificity to offer real value to life science. Within3 has created technologies that utilize unsupervised learning to ‘remove the noise’ in the data, and find the key topics that really resonate with the end user. Those topics are then summarized and analyzed for sentiment, using transformer models that have been specifically tuned for the needs of medical affairs and commercial teams.
How Within3’s AI supports our customers
In a highly-regulated, highly-sensitive industry like life science, we have to take great care about how we work with our customers where artificial intelligence is concerned. At Within3, we never use customer data to train our AI models, we don’t keep hold of our users’ sensitive data, and we never pass proprietary information through open-source tools. Nor do we handle patient data. We undergo annual SOC2 Type-2 audits to ensure we meet and exceed industry standards for data security, and maintain a robust backup and disaster recovery process with daily incremental and weekly full backups and 24-hour recovery time and recovery point objectives (RTOs and RPOs).
At the start of the insights reporting process, we collaborate with our clients to build ‘KITs’ (Key Insight Topics) that help to determine what they want to achieve, and what they’re measuring success against. This allows us to align the findings of our AI with our clients’ strategies. Our AI pulls data from a variety of different sources on the client side: CRM data, enterprise data lakes, congress data, and other forms of external data including clinical trial and social media data. Before it’s processed and analyzed, this data is just noise. So, our AI goes through and categorizes large chunks of text – either within each data stream, or more broadly to recognize macro themes independent of resources.
“What’s interesting is you don’t always know what you’re going to get. We allow the computer to tell us what’s there – that’s why it’s called unsupervised learning. We’re not telling the computer what to go look for. We’re saying to the AI: tell us what you found.”
– Jason Smith, CTO, Within3
Within3’s AI integrates with our clients’ CRMs to provide full implementation in just two to four weeks. We also work with clients’ IT teams to implement a single sign-on for the Within3 system – further cementing our commitment to data security.
What next for AI?
The field of artificial intelligence is developing rapidly, and dramatic changes should be expected over the next few years. Here at Within3, Jason and the AI team are working on how they can make our recommendations even better – training AI that can fully understand the business of medical affairs. And beyond that, we’re working to develop AI that learns each client’s organization, offering even richer personalization so that each client gets their own, unique AI footprint.
In the wider AI space, Jason expects artificial technology to become increasingly ubiquitous in our everyday lives. And the applications that win out will be those that solve real, tangible problems for consumers.
“It’s a common phrase: ‘AI is not going to take your job; the person who knows how to use AI is going to take your job’. And I think that’s pretty accurate for where the world’s going. So we need to ensure we make our tool highly accessible, easy to use, and always relevant in the lives of our customers.”
– Jason Smith, CTO, Within3
The Within3 Insights Management Platform uses artificial intelligence to solve real-world problems for our customers. We provide actionable insights you didn’t know you needed, enabling you to assess the success of your strategy, or pivot it in an entirely new direction. Schedule a demo today to find out what AI-powered insights reporting could do for your organization.
