AI and a proactive path to (projected) prosperity
AI is having a Wild West moment. One biotech is taking the technology (and its future) firmly in hand.
THE PROMISE OF AI
WAS CLEAR. HOW
COULD THEY APPLY
IT AS SOON
AS POSSIBLE?
The Situation
Here’s an old saw: Failing to plan is planning to fail. Of course nobody plans to fail, but plenty of business leaders, heads down trying to make next quarter’s numbers, may feel they don’t have the luxury of looking up and out at their horizons.
Not so with leadership at one global biotech company—specialists in genetic medicine. Maybe it’s because biotech products take so long to develop; maybe because company culture included a baked-in comfort with advanced technologies. Whatever the reason, the company’s leaders understood the transformative promise of artificial intelligence (AI) and wanted to start applying the technology as soon as possible.
How? They didn’t know yet. But they did know that AI, strategically deployed, could deliver real impact and value. And they knew this because they’d done their homework. They’d learned how, in life sciences, AI could accelerate drug discovery and development. How trials are conducted. How patients are engaged. Then there was the prospect of AI improving general business efficiencies—efficiencies that could increase revenues, decrease costs, and reduce time to market.
Put together, these things could create a competitive advantage.
The company had even done some small-scale prototyping—enough to confirm that they’d reached a go-big-or-go-home moment. Because unlocking the value they were learning about would take a sustained, organized effort with the funding to support it. It would take both business engagement (working with functional leaders in R&D, legal, HR, finance, and others to create use cases) and technology delivery (the AI skills and the IT architecture needed to deliver on those use cases).
So, the company had an overall plan; they just needed an assist in implementing it. They called Deloitte Consulting’s specialists in AI services.
THE SOLVE
Here’s another old saw: Solid foundations lead to strong structures.
Which is why, once engaged, the Deloitte team approached the company’s foundational AI program methodically and systematically, starting by helping stand up an AI Center of Excellence (CoE). This CoE would manage the business engagement side of the effort, responsible for both AI strategy (including program vision, structure, and governance) and yearly roadmaps (use case identification, assessment, prioritization, and activation).
Among the CoE’s first orders of business was to establish a set of guiding principles (i.e., how to responsibly use AI to increase revenue, reduce time to market, lower costs, and improve customer experience).
Overlaying this foundation: a shared understanding of the core AI capabilities on which the as-yet determined solutions would be built (i.e., to maintain focus, let’s agree on the stuff that AI does really well and limit solutions to those areas). To that end, the working group established that whatever they built would either summarize (extract and condense valuable information from lots of data), generate (create original content in different media formats), or search (find things faster).
Finally, to manage potential risks in deploying these new solutions, the CoE instituted an AI Governance Council, composed of representatives from legal, compliance, and IT departments. Their job: to serve as a backstop to the development process, reviewing proposed AI tools—their function, their design, their data requirements—through the lens of a risk framework, then assess how to proceed. No go-ahead from the Governance Council, no solution.
Meanwhile, the Deloitte Tech Foundry team was ramping up the delivery side of the house to manage building and deploying the solutions greenlit by the CoE and Governance Council, and to provide post-production support to boost the tool’s use and impact.
With these new organizational structures (and guardrails) in place, it was time to actually build some things. But what? To find out, the CoE reached out to more than 100 business stakeholders across more than half a dozen functions to learn where AI summarization, generation, or search (the AI capabilities) could best align with the AI program objectives (increase revenue, reduce time to market, lower costs, or improve customer experience). The result: 140 potential use cases, from which the CoE prioritized 10—with a particular focus on research and clinical operations—for deployment within the fiscal year.
The approved solutions would be built on Amazon Bedrock, a fully managed service from Amazon Web Services (AWS) that makes pre-trained Foundation Models (large-scale, pre-trained neural networks used to develop specialized AI applications) from leading AI companies easily accessible. (Claude, specifically, would be the Foundation Model used).
(Side note: Given the novelty of these technologies, the Deloitte Tech Foundry team built proofs of concept for several proposed solutions while the CoE did its work. This helped validate feasibility and avoid the gotchas that can come with innovation. And to be sure these solutions, once built, enjoyed the widest possible adoption and were put to their best possible uses, a different set of Deloitte professionals put together a comprehensive, seven-week change management plan for the company.)
140 POTENTIAL USE
CASES, 10 AT THE
FRONT OF THE LINE
(AND ALL OF
THEM BUSINESS
BOOSTERS)
The Impact
Here’s a final old saw: Measure what matters.
And what matters here is that the company’s journey into an AI-assisted future is not only underway with élan, but that the AI program overall is well positioned—with a charter, guiding principles, strategic and governance structures, policies, procedures, and controls—to stay agile as market and technology landscapes inevitably evolve.
Of the 10 use cases identified for the first fiscal year, seven have been delivered and are in active use:
- An organization-wide digital assistant (chatbot) for both internal and external questions that has revolutionized information flow and real-time response (thus improving organizational efficiency).
- A product complaints agent (that runs in the background) for summarizing incoming complaints and product adverse events (thus significantly reducing processing time).
- A Health Authority Query (HAQ) intelligence agent (chatbot) that retrieves a list of similar, relevant past HAQs (thus speeding turnaround time and saving significant time and costs in regulatory processes).
- An invoice flagging agent (that runs in the background) for scanning invoices to identify claw-back opportunities (generating substantial incremental savings).
- A plain language summary agent (via a custom user interface) that simplifies complex medical documents (generating significant cost and time savings).
- A marketing material localization agent (via custom user interface) that translates and applies local guidelines to treatment guides in other countries (producing a drastic reduction in process time and agency spend).
- An investor relations (IR) assistant (chatbot) that generates potential analyst questions and responses for investor events (thus boosting C-level preparedness and enhancing the firm’s reputation during these events).
…with more to come.
Explore more stories
Faster iteration or tighter governance? An innovative AI/ML platform supports both
Deloitte helped Thomson Reuters enable a stream of AI/ML capabilities within a single user interface
Bringing AI to the forefront of a tech-driven food revolution
Compass Group Australia and Deloitte Australia launch an AI-enabled system for improving nutrition among older adults.
AI targeting. Human assisting.
How predictive modeling helped one state serve its most vulnerable population