Agent AI can redefine how companies operate, compete and grow faster than many leaders expect. It gives you the most clear glimpse into what the future of work and business will look like. This is expected and exciting. New ecosystems are already emerging to support technology and redesign technology stacks and workflows. The opportunities are enormous and confusion too. So, what should leaders know now and what should they do?
What is an agent? Like AI, there is no definition of consensus. Trying to force it really doesn't help. In general, “agents” are AI models designed to complete unprecedented complex multi-stage tasks without human intervention. Most are models built on top of existing LLMs (all advantages and disadvantages), calling multiple datasets and other websites to knit tasks. For example, “Please book at Jaws at 7pm on Tuesdays and schedule your car to get there on time in light of traffic conditions.”
Are we there already? At the time of writing, it is not. Most commercially available agent-style AI still requires substantial human input, alignment, guidance and monitoring. Other models already cluster agents into “multi-agent” systems, and in fact do good (and improved) jobs that simulate autonomous functionality by linking the agents together to solve tasks. Today, these systems are more “agents” than agents. In the future, the landscape will become even more complicated, including agent models, human-guided agents, narrow AI, and regular interactions between existing human processes and fully autonomous agents. This hybrid environment will be a governance challenge for leaders of tomorrow.
Who has agents and uses them? More and more, anyone can access pre-built agents and easily build their own agents. Conceptually, there is no technical limitation on the number of agents that a person or an organization can have. Existing organizations need to integrate agents into current workflows and processes (often with significant redesign). Meanwhile, new companies will build on Agent Core from day one.
Agent vs Agent Ish, is that important? yes. Agent-isch models are more suitable for a specific type of task than agents' models, so both foreseeable future types work and require very different strategic and governance approaches.
What kind of work can these agents do? The list continues to grow, but it is best suited to repetitive tasks where your business already has a clear or proven sense of what it looks like (e.g. assemble regular financial reports, marketing campaigns, summarizing materials, creating a variance of key findings, and perhaps it may be performed).
Operational advice (for now)
1. These tools are available here, available, and only get better. Existing businesses need to learn to succeed in an environment with more automated competition and competitors.
2. Know your data. These tools run on data as all attendant dependencies depend on data quality, accessibility and governance.
3. Update your policies, guardrails and training. Determine how and when the agent tool is permitted to use in your business. Define, encode and train core values, standards, principles and ethics for your people and your agents. This training is essential to their success.
4. Develop a method for selecting an agent's use case. Not all use cases (e.g., medical diagnosis) or specific problems (e.g., high value and everyday conflict) are suitable for a true agent model. Therefore, organizations need a process to determine which types of tools and levels of autonomy are suitable for which problems.
5. Build an agent tracking and management system. Agents can easily multiply. Find out which one is working and what it's doing. Agents interact with each other. And the agent behavior cannot be modified after the fact.
6. Establish agent governance protocols. Agents need to align with the goals of the human set (for now). These objectives should be reviewed, aligned and contoured with quality standards and boundary conditions identified, established and implemented. This process also requires training the workforce.
7. Start redefine the workflow. Adding agents to existing processes may not unlock the most optimal or competitive advantages. But by rethinking and redesigning your workflow and incorporating competent agents and well-skilled managers, you start unlocking these benefits, and that redesign work could begin now.
