This year at Dreamforce, the spotlight was on AgentForce, Salesforce’s attempt to get ahead of the AI hype. With companies like OpenAI and Microsoft leading the AI buzz, many corporate leaders feel pressure to adopt AI — but are unsure how to do so without risking data security. While AI shines in specific tasks like content creation and customer queries, integrating it into a consistent workflow remains unclear for many.

Nevertheless, Salesforce continues to push the message that they have the most trusted AI platform, with Marc Benioff introducing AgentForce (autonomous agents) as the next step beyond generative AI. His key message: "You’ve invested all this time and money into Salesforce—why would you look elsewhere?" and “Don’t DIY your AI,” appealing to companies that have already sunk costs in the platform.

Benioff also hinted at AI replacing the need to hire new employees in scenarios like Black Friday or Diwali sales, where companies might need to scale their call center staff quickly. AI agents could handle these surges without depending on human staff.

That’s a relatively gentle way of broaching the topic that these AI agents can take the place of human workers. But if these “artificial employees” prove themselves through the holidays, why not keep them on through the year? Especially since, unlike with “natural employees”, scaling artificial employees up and down does not require an HR team, recruiting, hiring, training, and dismissing people. Those costs are overhead costs that do not directly bring value to the company. And even if you could magically bring in and out fully trained human employees on demand, the chances are that you will find autonomous agents who could do the work for a fraction of the cost.

Salesforce, like other tech giants, reduced headcount last year. Part of the argument was that generative AI now enabled companies to do the same work with fewer employees. Layoffs reduce corporate expenses without any immediate impact on revenue, and so immediately improves companies’ apparent financial performance.

“Doing more with less” is a perennial goal for individuals and companies alike. But the message at Dreamforce was clearly meant to stoke anxiety: “You need to move to AI now. Because if your competitors get there first, they will outcompete you.” High-level decision-makers in companies have typically reached those positions through being very competitive, fearing falling behind. But becoming CEO does not automatically bestow a sense of contentment and satisfaction: It broadens your scope of concern so you begin competing not against other employees, but against other companies. So messaging about not falling behind is particularly effective for the highly-driven leaders of organizations.

There are two key ideas that stand out from this:

First, AI will reduce the level of skill required for humans to accomplish tasks, and increase their productivity. This is similar to the shift to moving workforces offshore (from the US) that happened in manufacturing in the 1980s and in software development throughout the 1990s-2000s.

One of the challenges with invoking less-skilled and less expensive teams, is that although they can create a vast amount of materials/software/content quickly, this can lead to messy and vulnerable output. As AI increases the rate of worker productivity, the challenge will shift to testing the work of both human and AI agents to ensure security, compliance, and quality.

The second key idea is the growing need to augment humans’ natural intelligence, not just artificial intelligence. AI agents can quickly create work at a level of complexity that’s difficult for humans to understand. Our focus needs to shift to simplifying, visualizing, and creating tools that help humans make sense of this complexity. The term “natural intelligence,” coined by AI researchers, refers to human, plant, and animal intelligence. Much of this intelligence is visual, which is why visualizations are incredibly powerful in helping people understand complex systems.

The hype around generative AI and autonomous agents is pointing to a future that seems almost inevitable: New forms of intelligence require less mental energy from humans. Generative AI tools like ChatGPT convert a small amount of human work (generating a prompt) into a large amount of output. In the process they consume significant electricity and water (to cool the servers) (https://www.washingtonpost.com/technology/2024/09/18/energy-ai-use-electricity-water-data-centers/) , but this is in keeping with a long-term trend for humans to harness electricity and other natural resources to support our activities.

Many types of work will inevitably disappear due to these technologies. But companies like Salesforce are betting that they can convince companies to invest early in unproven technology, in hopes of reducing headcount costs, and gaining or maintaining competitive advantage. Companies who invest in this technology from Salesforce are signaling commitment and helping finance what will probably be a chaotic scramble for Salesforce to deliver something that resembles these promises.

But expect significant chaos and human disruption as jobs are sacrificed prematurely, and as the output of these AI engines creates systems that are even harder and more expensive to manage at scale.

If the change is too rapid it will be extremely wasteful. Companies money will be wasted by paying upfront for the promises that won’t be delivered soon. Time will be wasted through disorganized changes that will require rework. And we will waste our investment in building trust and community within a company: layoffs damage or destroy the social fabric on which companies are built.

INSIGHTS TO CONSIDER / REFLECT ON:
* AI Integration and Market Pressure: Companies are under increasing pressure to adopt AI-driven solutions to stay competitive, but many are unsure how to implement them effectively without compromising data security.
* Salesforce’s AI Strategy with AgentForce: Salesforce’s introduction of AgentForce positions it as a leader in autonomous AI agents, encouraging companies to rely on its platform rather than experimenting with custom AI integrations.

Warm regards,

Andrew Davis