Data and Innovation

These are the bulletpointed notes from the keynote at LifeSciences Dreamin 2024.

1. Bad Data is Junk Food for AI

AI models rely on clean, well-structured data to perform effectively. Feeding an AI poor-quality data is like giving it junk food—it leads to weak or inaccurate outcomes. This analogy underscores the importance of data quality in AI-driven decision-making processes.

2. Why Are We Only Worried About Data Quality Now?

Data quality has always been crucial, but with AI’s rise, the awareness is higher. In the past, data quality was essential for:

• Decision Making: Ensuring dashboards reflect accurate information.
• Governance: Supporting reliable reporting systems.
• Operational Efficiency: Running businesses smoothly with reliable data inputs.

However, the AI revolution amplifies these concerns, requiring attention to every data point.

3. The Impact of Technical Debt and Poor Documentation on Data Quality

Technical debt accumulates from a lack of a rigorous implementation approach; a lack of business analysis before rushing into build. Based on our Change Intelligence Research Series we have the data and can see the consequences:

• 51% of custom objects are never used
• 41% of custom fields rising to 88% core objects are never populated
• 150 fields on opportunity page layouts, when only 7 are ideal, leading to wasted time, increased user frustration, and poor data.

When documentation is insufficient or outdated, duplication of fields occurs because teams aren’t aware of existing fields. This duplication muddles data quality further—creating confusion, especially when determining the “master” field among multiple entries.

4. Data Siloes and Volumes: A Growing Challenge

The sheer volume of structured and unstructured data is skyrocketing, with businesses needing to aggregate data held in multiple siloed apps to get a 360 degree view of a customer using Data Lakes and Salesforce Data Cloud.

5. AI’s Power to Analyze Massive Data Volumes

AI excels at handling and analyzing massive structured and unstructured data sets, offering speed and insights that humans could never match. For example: AI in just 1 hour replaced 19,000 hours of development work. With AI, businesses need to shift their mindset to “why can’t we” and “what if we did,” fostering a new era of innovation.

6. The AI-Human Balance

While AI can analyze large volumes of data, humans still play a critical role in adding gut intuition and filling in contextual gaps. To date, the best results come from a collaboration between AI and human decision-makers, marrying algorithmic precision with human insight.

7. What Is AI Really?

Is AI:

1. Machine learning algorithm
2. Next-word prediction
3. MAGIC!!!

Here is a podcast that AI generated from this blog post – with a single click; the script, the voices. Amazing, but not necessarily accurate.

8. Enter Agents: A New Era of AI

Bots follow pre-planned, structured paths. But now, AI agents represent a new level of sophistication. Agents are dynamic—they can plan and initiate actions, breaking from traditional rule-based systems. To create an effective agent, it is crucial to:

• Identify Roles: What actions does the agent take?
• Provide Data Sources: What data does the agent need access to?
• Grant Capabilities: What actions can the agent take on behalf of a human?
• Set Guardrails: When does the agent hand back control to humans?
• Decide Channels: In which environments will the agent operate?

Salesforce has branded its AI offerings, including its framework for building agents – Agentforce. They used the song Secret Agent Man as the backing track to their promo video. So the band created this short Secret Agent Man video.

9. The Implications of AI Agents

AI agents will transform processes across industries, but they come with new requirements and challenges, including:

• A deeper understanding of operational processes.
• Better governance and data quality.
• Position on legal and reputational risk management.
• More mature implementation, feedback, and monitoring.

10. Are We Ready for AI?

Many businesses are on the fence. While perfect data doesn’t exist, neither should it stop experimentation. Begin small, in low-risk, non-operational areas to prioritize fixes based on pilot results. Waiting for perfect data to begin AI initiatives is akin to waiting for perfect processes before hiring an intern. Start now and scale.

12. AI Won’t Replace You, But Someone Who Uses AI Will

The message is clear: AI will not take your job, but someone who leverages AI effectively might. Starting small, through pilot projects, will help you harness AI’s potential, ensuring you stay ahead in this data-driven world.