Healthcare professionals have many questions when facing AI-driven change. Some of the voices we hear include: "In which scenarios should I use AI?", "Which AI tools are suitable for me?", and "How can AI connect with my business needs?" To help everyone take the first step in AI transformation quickly, we tried many things. Eventually, we decided to use AI to help ourselves first.
Today's story starts with a competition. In May this year, we held an AIGC competition involving the entire company. The final gold award went to a colleague from the web crawler team, who used AIGC plus crawler technology to solve a problem that had troubled us for a long time.
(*This article is excerpted from remarks by MeDomino founder and CEO Lu Wenqing at the 2024 DPIS Conference and 2024 CphMRA.)
01 AIGC + Crawlers: Solving a Recruitment Challenge
MeDomino is a digital intelligence company and, in today's environment, we can say we are standing on a strong wave of opportunity. We need to attract a broad range of talent and face a large number of recruitment needs. The first story I want to share is the gold award project from our internal AIGC competition:
To recruit, we bought memberships on recruitment websites. HR reviewed, screened, downloaded, and scheduled interviews one resume at a time, which took a great deal of time and energy. In the end, I even found that among nearly 10 open roles, perhaps only the top two priority roles were actually being recruited for. Roles ranked third and below had no time for screening, and candidates were not being invited for interviews.
In April this year, we made changes:
First, we clarified recruitment needs.
Second, we automatically collected all resumes available within the account quota every day, about 300 resumes per day.
Third, we used large language models to structure the data.
Now recruitment has become much easier. We directly call an LLM to quickly structure resumes into an Excel table, breaking them into fields such as work skills, work experience, job responsibilities, and university. Every field is clear. No matter what format the original resume uses, the key information can be mapped into the table one by one, ultimately forming a large resume table with 300 rows and dozens of columns. At present, we update it every afternoon, and department leaders can review resumes directly from the table and complete initial screening in under half an hour. In less than one month, two new colleagues have already joined us.
02 From Ourselves to the Industry: Exploring Consensus on AI Applications in Life Sciences
Starting from our own experience and then looking at the life sciences industry, we see both shared needs and individualized differences.
To gain broader consensus, we interviewed many senior practitioners in the pharma and medtech industry. During this process, I saw a large number of real scenarios and needs, such as information acquisition and organization, customer insight, content generation, process management, and more.
Based on these scenario needs, we evaluated multiple AI tools and summarized a shared consensus for AI + healthcare applications: the first priority is to consider which scenarios have not yet been supported, turning the impossible into the possible, rather than simply accelerating scenarios that were already working fairly well.
Read more: AI + Pharmaceutical Scenarios "4 Dos and 1 Don't"
Figure: The "4 Dos" for AI + pharmaceutical scenarios
When implemented in concrete scenarios, the first application that fits the "4 Dos" principle is intelligent Q&A. Why?
Because intelligent Q&A has a very low usage threshold and everyone can get started quickly. It also has a relatively high tolerance for error, because even if an AI answer contains some deviation, humans can further optimize and adjust it. More importantly, knowledge-base Q&A is something many pharma and medtech companies have not yet achieved. Employees' knowledge access needs have not been fully met, and this is visibly capable of bringing business value quickly.
The second is content generation.
I believe enterprise-level content production should mainly come from reusing previous content. Based on new scenario needs, suitable historical content can be combined with a small amount of new information and reorganized into content that meets current needs. However, enterprises often have thousands or even tens of thousands of content assets. If people have to extract and summarize them manually, it consumes enormous time, energy, and resources. This requires tools that help people quickly understand content, and AI is the most suitable option.
These scenarios are already being implemented. In the fourth quarter of last year, we helped a pharmaceutical company launch an AIGC-enabled content management platform. Based on actual usage data, one-click summaries accounted for nearly 45% of usage frequency. In addition, within four months of launch, a single user had used intelligent Q&A as many as 147 times. At present, these AI capabilities have already improved new content generation efficiency by almost 60% or more.
03 NICE: A Customer-Journey-Centered Digital Intelligence Tool for Business Growth
As we often say, AIGC is not a standalone thing. It must connect with various types of data across the entire marketing system to solve business problems. This is a digital intelligence marketing solution. We call this complete solution NICE, or Next Intelligent Customer Experience: a customer-journey-centered tool that drives business growth through digital intelligence by ensuring consistent strategy execution.
Module 1: MeDomino Content Hub
The scenarios mentioned above are all easy to perceive directly, but at their core, the foundation is enterprise-level content management. Once a secure and compliant enterprise knowledge base is in place, it can connect with other enterprise platforms and data, enabling multimodal content management and allowing upper-layer application scenarios to expand continuously.
Figure: Enterprise-level content management is the foundation of AIGC scenarios
Within the platform, we can automatically identify and tag content. These content tags will then correspond effectively with the customer journey. MeDomino Content Hub has also connected with enterprise-level approval platforms, so employees no longer need to run the same process across different systems. Both business workflow efficiency and internal user experience have improved significantly.
Module 2: HCP360
HCP360 is true integration of internal and external data, including ONE ID, all profiles, customer information accumulation, and subsequent analysis. By analyzing all data, it ultimately forms deep customer insight and locates where each customer sits in the customer journey.
Visit the HCP360 product page to learn more.
Module 3: Strategy Intelligence
After identifying where customers are in the journey, we can understand how many types of customers there are, which stages they are in, how many people are in each stage, what content is suitable, and how our content should match these customers.
Through strategy intelligence, targeted personalized strategies can be formed intelligently, then implemented and visualized, so business teams can clearly understand what they should do.
Module 4: ACE
Successful strategy execution depends on collaboration across channels. One very important link is the frontline team. ACE, Assistant to Customer Engagement, is an assistant tool built specifically for frontline colleagues. It includes a large amount of customer data from both internal and external sources, helping frontline colleagues build touchpoints and trusted relationships with each customer.
For example, based on the platform's prior deep customer insight and inventory of enterprise content resources, we can know that a certain customer is suitable for a certain piece of content or a certain meeting. After instructions are sent to frontline colleagues, they can view and execute directly through ACE.
Module 5: Digital Channels
In addition to ACE used by frontline colleagues, automated distribution can also be carried out through digital channels. The logic behind this is also based on customer segmentation and corresponding content tags.
With AI capabilities, repeated cycles of content tags x customer insight, personalized distribution, data collection, and analysis optimization make marketing actions increasingly refined. During this process, enterprise data assets can quickly accumulate.
04 Conclusion
Finally, returning to the opening question of how to take the first step in AI transformation, I want to say this: for digital teams in pharma and medtech companies, this is a historic and strategic opportunity, and a chance to become a core team within the enterprise.
So action must be fast. The key at this stage is not to wait: do not wait until everything is fully understood before acting, do not wait until the work is polished to 99% perfection before starting, and absolutely do not over-focus on decorative details.
