What if an agent could anticipate the future before acting? This is the world of deliberative agents, the true strategic minds of artificial intelligence.
What is a deliberative agent?
A deliberative agent not only perceives and acts:
- Think before you act.
- Simulates different possible paths.
- Choose the plan that best brings you closer to your goal.
It’s like playing chess or StarCraft: you don’t win if you improvise every move. You need to think several moves ahead, calculate risks, anticipate the opponent… and then yes, act.
How do deliberative agents plan?
To achieve this, these agents use planning algorithms that allow them to model the world, predict future states and choose the best sequence of actions.
Some of the best known are:
- A* (A star):
- A classical optimal path finding algorithm.
- It is used in GPS navigation, video games or mobile robots.
- It allows finding the shortest path from point A to point B, avoiding obstacles and considering the cost of each step.
- Decision trees:
- They are diagrams that represent all possible actions and their consequences.
- Each decision is followed by a ramification of possible futures, until the best route is chosen according to the objective.
- They are key in AI for games or recommender systems.
- STRIPS:
- An approach used in automatic planning, especially in robotics and classical AI.
- It defines the current state, the desired state and a set of actions that transform the former into the latter. STRIPS helps the agent put together a step-by-step plan to reach his goal.
A deliberative agent in StarCraft:
- Perceive the map, the position of enemies, available resources.
- Simulate possible strategies: attack fast, defend and collect resources, or divide forces. Plan and adjust in real time according to what you discover.
- It is not just reacting or learning from experience: it is planning, foreseeing and deciding with a cool head.
And why does this matter?
Because in the real world, many problems cannot be solved by quick reactions or pure learning.
When facing complex and long-term environments, we need agents who know how to project, plan and adjust. From the logistics of a supply chain, to the coordination of autonomous fleets or the energy management of a city, deliberative agents are the key to efficiency and sustainability.
How do we apply this in SMS Sudamérica?
At SMS Sudamérica, deliberative thinking is not just for strategy games: we apply it to projects where optimizing each step is fundamental for the business. In industry, we help plan production processes where every decision affects time and costs.
In the public sector, we design solutions that project future scenarios to make strategic decisions.
And in complex systems, such as healthcare or energy, we apply planning models that simulate before acting, to reduce risks and inefficiencies.
For us, intelligence is not only about acting fast, but also about choosing the next step well, with a vision of the future. That is why we develop deliberative solutions that enable organizations to anticipate challenges before they occur.
In the next episode of “Stories of the Future, Today.”
We will delve into reinforcement learning: What happens when an agent does not have a map? but learns the way by making mistakes over and over again?
Note by: María Dovale Pérez
