Tubegraph

Trends and Predictions for Agents in 2025

4 min read

From: aidotengineer

The field of AI engineering is rapidly evolving, with a significant pivot towards agent-based technologies. The AI Engineer Summit has shifted its focus to become the Agent Engineering conference, indicating a strong belief in the importance and progress of AI agents [02:45:00]. This decision was made to concentrate efforts on agents, even if it meant saying no to other popular topics like RAG, open models, and GPUs [02:51:00].

The Evolution of AI Engineering

AI engineering is maturing and spreading across different disciplines [01:19:00]. While there are still debates about whether AI engineering is simply an extension of machine learning or software engineering, it is expected to emerge as its own distinct discipline [01:57:00]. Currently, it’s described as 90% software engineering and 10% AI, but the AI component is anticipated to grow significantly over time, particularly in 2025 [02:03:00].

The role of an AI engineer is evolving towards building agents, similar to how ML engineers build models and software engineers build software [11:00:00].

Defining an Agent

Despite widespread discussion, the definition of an “agent” can be fluid. Machine learning practitioners might view an agent in terms of reinforcement learning environments, actions, and goal achievement [05:41:00]. Software engineers, in a more reductive sense, might see it as a “for loop” [05:51:00].

Simon Willison has crowdsourced over 300 definitions, highlighting common themes:

OpenAI also recently released their own definition of agents [06:52:00].

Why Agents are Gaining Traction Now

Several factors contribute to the current surge in agent development and adoption:

  • Increased Capabilities: Agent capabilities have significantly grown, hitting human baselines around now [07:19:00]. This includes better reasoning, improved tool use, and more sophisticated tools like mCP [07:37:00].
  • Model Diversity: The market share of major AI model providers has diversified. OpenAI’s market share has dropped from 95% two years ago to 50%, with two new frontier model labs emerging as potential challengers [07:51:00]. This competitive landscape is exciting for 2025 [08:06:00].
  • Lower Cost of Intelligence: The cost of GPT-4 level intelligence has decreased 1,000 times in the last 18 months, with similar trends starting for 01-level intelligence [08:14:00]. This significant reduction in cost makes agent deployment more feasible.
  • Reinforcement Learning (RL) Fine-tuning Options: The availability of RL fine-tuning options further enhances agent capabilities [08:28:00].
  • Focus on Outcomes: Discussions with industry leaders like Brett Taylor highlight the shift towards charging for outcomes rather than just costs [08:43:00].
  • Multi-agents and Faster Inference: Developments in multi-agent systems and faster inference due to improved hardware are also contributing factors [08:48:00].

There is a strong push from major industry figures like Satya Nadella, Roman, Greg Brockman, and Sam Altman who want people to believe that 2025 is the “year of Agents” [04:32:00]. While skepticism exists, even those initially doubtful, like the speaker, now acknowledge the shift [04:48:00].

One simple formula for making money in 2025 is the concept of “everything plus agent works” (e.g., agent + RAG, agent + search) [04:00:00].

Agent Use Cases and Anti-Use Cases

Confirmed product-market fit (PMF) exists for:

However, there are “anti-use cases” that developers are urged to stop demonstrating, such as agents that book flights or Instacart orders, as users often prefer to handle these tasks themselves [09:24:00].

Explosive Growth in AI Product Usage

The growth of any AI product is predicted to be tightly linked to its reasoning capabilities and the number of agents it can ship to users [10:41:00]. This is evidenced by OpenAI’s reported user growth:

  • OpenAI reported 400 million users, a 33% growth in three months [09:47:00].
  • ChatGPT reached 400 million users in 2.5 years [09:59:00].
  • The speaker notes that ChatGPT’s usage stagnated for a year because they didn’t ship any “agentic models” [10:05:00].
  • The “01 models” have doubled ChatGPT’s usage [10:24:00].
  • ChatGPT is projected to hit a billion users by the end of 2025, quintupling its user base from September of last year [10:28:00]. This represents roughly one-eighth of the world’s population [10:49:00].

This rapid growth suggests a significant amount of money is available for everyone else in the AI agent market [10:53:00]. The future prospects in AI and agent-based technologies appear promising, with a focus on challenges and opportunities in AI and agent capabilities.

Graph View

Backlinks