Frequently Asked Questions

Agentic AI & AI Agent History — Frequently Asked Questions

Last updated: April 29, 2026

Direct, cited answers to the most-asked questions about the history of AI agents and agentic AI. For the long-form chronological treatment, see the timeline.

Origins & pioneers

What is the history of AI agents?

The history of AI agents spans roughly seven decades. The conceptual seed is planted at the 1956 Dartmouth Summer Research Project on Artificial Intelligence. Multi-agent coordination is formalized by Reid G. Smith's Contract Net Protocol in December 1980. Marvin Minsky's Society of Mind (1986) argues intelligence emerges from the interaction of many small agents. Michael Bratman's BDI framework (1987) supplies the dominant agent architecture for the next two decades. Stuart Russell and Peter Norvig's textbook (1995) reorganizes the field around the rational-agent abstraction. The modern LLM-driven era begins with the ReAct paper (October 2022), accelerates with the OpenAI APIs (March 2023), and explodes with BabyAGI and AutoGPT later that month. Computer-use agents from Anthropic (October 2024) and OpenAI Operator (January 2025) mark the next major shift. The full chronology is on the timeline.

Who invented AI agents?

No single person invented AI agents. The credit is shared across decades and disciplines. The 1956 Dartmouth attendees — John McCarthy, Marvin Minsky, Claude Shannon, and Nathaniel Rochester — established the field that would later define itself in agent terms. Reid G. Smith first formalized multi-agent coordination in 1980. Marvin Minsky gave the multi-agent view of intelligence its theoretical foundation in 1986. Michael Bratman supplied the philosophical backbone of BDI agents in 1987. Stuart Russell and Peter Norvig made "agent" the standard unit of analysis in AI in 1995. For modern LLM-based agents, the inventors are different people again: Shunyu Yao and co-authors (ReAct, 2022), the Adept team (ACT-1, 2022), Yohei Nakajima (BabyAGI, 2023), and Toran Bruce Richards (AutoGPT, 2023).

Who pioneered AI agents?

The pioneers of AI agents fall into two clear groups.

Foundational pioneers (1980–1995): Reid G. Smith (Contract Net Protocol), Marvin Minsky (Society of Mind), Michael Bratman (BDI), Yoav Shoham (Agent-Oriented Programming), and Stuart Russell & Peter Norvig (textbook synthesis).

Modern LLM-agent pioneers (2022–2024): Shunyu Yao and the ReAct co-authors, David Luan and the Adept team (ACT-1), Yohei Nakajima (BabyAGI), Toran Bruce Richards (AutoGPT), and the founders of the first agent-first companies — Cognition's Scott Wu, Steven Hao, and Walden Yan; Sierra's Bret Taylor and Clay Bavor.

Who are the pioneers of multi-agent systems?

The canonical pioneers of multi-agent systems are:

• Reid G. Smith (1980) — the Contract Net Protocol gave the field its first formal coordination framework.
• Marvin Minsky (1986) — Society of Mind gave the field its theoretical foundation: intelligence as a collective.
• Yoav Shoham (1993) — Agent-Oriented Programming proposed agents as a primary unit of software design.
• Tim Finin and the KQML community (1994) — the first widely deployed agent communication language.
• FIPA (from 1997) — the standards body that produced FIPA-ACL and the most widely adopted agent communication standards.

See the 1980, 1986, and 1990s sections of the timeline for primary citations.

Who are the pioneers of modern multi-agent systems?

"Modern multi-agent systems" generally refers to LLM-driven multi-agent architectures. The pioneers in this narrower sense are:

• Shunyu Yao and ReAct co-authors (October 2022) — the reasoning-plus-action paradigm that almost every modern agent framework uses.
• The LangChain team, led by Harrison Chase (from late 2022) — the framework that productized the ReAct pattern and let thousands of developers build agents.
• Microsoft Research's AutoGen team (2023) — one of the first widely adopted multi-agent LLM frameworks.
• The CrewAI team (2024) — popularized role-based multi-agent architectures with structured handoffs.
• The LangGraph team at LangChain (2024) — graph-based multi-agent orchestration.

When did AI agents start?

Five plausible "start" dates, depending on what you mean:

• 1956 — the conceptual founding of AI at Dartmouth.
• 1980 — the first formal multi-agent coordination protocol (Smith's Contract Net).
• 1995 — "agent" becomes the standard definition of AI in mainstream education (Russell & Norvig).
• October 2022 — the ReAct paper defines the modern LLM-agent paradigm.
• March 2023 — autonomous LLM agents enter the mainstream with BabyAGI and AutoGPT.

Most contemporary use of "AI agents" — meaning autonomous, tool-using LLM systems — points at March 2023 as the practical starting point.

When did agentic AI begin?

"Agentic AI" as a term enters mainstream usage during 2024 and becomes ubiquitous in 2025. The capability it describes — autonomous, planning, tool-using LLM systems — begins with the ReAct paper (October 2022) and the AutoGPT/BabyAGI moment (March 2023). By the time the phrase is in common use, the underlying systems have been in development for two to three years.

Firsts & definitions

What was the first AI agent?

The phrase has many defensible answers depending on definition. Among them:

• First goal-directed symbolic AI program: the Logic Theorist by Newell, Simon, and Shaw (1956).
• First formal multi-agent system: the Contract Net distributed sensing system (Smith, 1980).
• First widely demonstrated transformer-based action agent: Adept's ACT-1 (September 2022).
• First publicly demonstrated end-to-end autonomous LLM agent: BabyAGI (Mar 28, 2023) and AutoGPT (Mar 30, 2023), within two days of each other.

What was the first autonomous AI agent?

The first publicly available autonomous LLM agent that demonstrated the now-canonical loop of objective → planning → execution → reprioritization is BabyAGI, posted by Yohei Nakajima on March 28, 2023. AutoGPT followed two days later (March 30, 2023) and overtook BabyAGI in mindshare almost immediately because it shipped with web browsing, file operations, and code execution out of the box. We treat them as effectively simultaneous origins of the modern autonomous-agent movement.

What is the history of agentic AI?

The history of agentic AI is the history of AI agents seen through the lens of contemporary marketing. The capability arc runs from Contract Net (1980) to Claude Opus 4.5 (November 2025). The terminology arc is much shorter: "agentic AI" was a niche academic phrase before 2023 and became the dominant industry framing during 2024 and 2025, as model labs and venture capitalists rebranded "AI assistants" into "AI agents." See the timeline for the underlying capabilities.

What is the difference between agentic AI and an AI agent?

An AI agent is a specific system: a piece of software that perceives, decides, and acts. Agentic AI is the broader category — the architectural and product approach in which AI systems are built to be agents rather than chatbots. In practice the terms are used almost interchangeably in 2025–2026, with "agentic AI" usually referring to the field or paradigm and "AI agent" usually referring to a specific instance.

How is an AI agent different from a chatbot?

A chatbot responds to messages. An AI agent pursues goals.

A chatbot is reactive: you send a prompt, it returns a reply, and the loop ends. An AI agent is proactive within an objective: given a goal, it plans, calls tools, observes results, and iterates — often over many steps and minutes — until the goal is achieved or the agent gives up. Modern AI agents are usually built on top of large language models, but they add three things on top: a planning loop (the ReAct pattern), tool use, and memory.

Specific people, papers, and products

What is the ReAct paper and why does it matter?

The ReAct paper — ReAct: Synergizing Reasoning and Acting in Language Models by Shunyu Yao, Jeffrey Zhao, Dian Yu, Nan Du, Izhak Shafran, Karthik Narasimhan, and Yuan Cao — was posted to arXiv on October 6, 2022 and presented at ICLR 2023. It proposes interleaving reasoning traces ("Thought") with actions ("Act") and observations from the environment, in a single LLM prompt-completion loop. This pattern is the architectural template for nearly every LLM agent framework that followed: AutoGPT, BabyAGI, LangChain agents, CrewAI, and modern Agents SDKs all derive from ReAct.

Was AutoGPT the first AI agent?

No, but it is the project that made "AI agent" mainstream. AutoGPT was released on March 30, 2023, two days after Yohei Nakajima's BabyAGI. AutoGPT became the top-trending repository on GitHub by April 3, 2023, and crossed 100,000 stars within weeks — the fastest growth of any open-source project at that scale at the time. The two are best understood as effectively simultaneous, and both were preceded by Adept's ACT-1 demo (September 2022) and the ReAct paper (October 2022) that defined the underlying pattern.

Who are the leading AI agent companies today?

As of April 2026, leading AI-agent companies include:

• Cognition (Devin, the AI software engineer; reported to be in talks at a $25B valuation in April 2026 after raising $400M at $10.2B in September 2025).
• Sierra (enterprise customer-experience agents; $350M at $10B valuation in September 2025).
• Anthropic (Claude, the leading frontier model for agents and computer use).
• OpenAI (Operator and the Agents SDK; integrated agent capabilities across the ChatGPT product line).
• Adept (ACT-1 and successor action transformer models).
• Manus / Monica (general-purpose autonomous agent; March 2025 launch).
• Microsoft (AutoGen, Microsoft 365 Copilot agents).
• The LangChain team (LangGraph and the LangChain agent framework).

See the funding table on the History page.

Why call this site a museum?

Because that is what it is: a curated, dated, sourced archive of how a field came to exist. Museums preserve objects and the context around them; we preserve product launches, papers, demos, and funding rounds, and the context that links them into a lineage. The choice is also a stance: we are not selling an agent. We are documenting the record so that researchers, journalists, and builders have a neutral place to start. See About the Museum for our methodology and source standards.

---

Continue: Read the full timeline · About the Museum · What is an AI agent?