In June, Andrew Ng, a prominent AI researcher, introduced the term “Agentic” to describe a new class of AI that moves beyond merely responding to commands—it takes action independently. Unlike traditional AI tools that rely on user prompts, Agentic AI is envisioned to handle complex tasks autonomously, such as analyzing data, predicting outcomes, and even executing decisions. This evolution is rapidly gaining traction, with tech leaders and startups alike rallying behind the concept, promising a future where AI systems act as true digital agents, managing business processes and making decisions without constant human oversight.
The term has quickly become a buzzword, as companies aim to capitalize on the hype surrounding AI’s next phase. However, Ng points out that defining what qualifies as “agentic” is still not straightforward. While some systems exhibit agent-like qualities, they may not be fully autonomous. Ng admits there’s a “gray zone,” where technology blends human input with machine-led actions, leaving room for interpretation as AI continues to evolve into more advanced forms. A key figure in this movement is Karl Friston, the world’s leading neuroscientist. His brain-inspired approach forms the mathematical foundation for agentic AI, enabling it to be more autonomous, adaptive, and collaborative.
Let’s look at some startups that are already ahead in the race and are doing some tremendous work the space.
Agency
Founders: Adam Silverman, Shawn Qiu, Braelyn Boynton, Howard Gil
Funding: $2.6 million seed round in 2024
Work: Developed tools for creating reliable AI agents at scale and built AgentOps, an industry-leading agent observability platform. Curated a collection of over 400 AI agents in production, assisted clients in prototyping and deploying ambitious AI agent projects, and provided expertise in building and reviewing AI agents.
Recent Developments: Launched AgentOps, an agent observability platform, and created a vetted library of over 400 AI agents in production. Expanded services to support startups and enterprise clients in building agents at scale, while also developing open-source tools for AI agent developers.
Cognition Labs
Founders: Scott Wu (CEO)
Funding: $175 million, with a claimed valuation of $2 billion as of April 2024
Work: Devin, an advanced agentic AI, functions as a fully autonomous software engineer. It handles long-term reasoning, planning, and complex engineering tasks independently, integrating common developer tools within a sandboxed environment to collaborate with human engineers and perform tasks autonomously.
Recent Developments: It achieved top performance on the SWE-bench coding benchmark, resolving 13.86% of issues end-to-end without assistance. Early access for engineering work has been opened, with plans to expand, while significant funding and a higher valuation reflect strong investor interest in their technology.
Hippocratic AI
Founder: Munjal Shah
Funding: $500 million
Work: Developing AI healthcare agents for low-risk, non-diagnostic tasks traditionally performed by nurses, such as chronic care management, wellness coaching, and follow-up care. Hippocratic AI claims these agents can outperform nurses in specific tasks while costing significantly less—$9/hour for AI agents compared to $60.26/hour for a registered nurse in California.
Recent Developments: Announced a partnership with NVIDIA to develop super-low-latency “empathy inference” for more natural patient interactions. Launched a staffing marketplace for hiring AI agents in healthcare and began phase three of safety testing with 5,000 nurses, 500 physicians, and various health system partners. Collaborated with over 40 provider groups and digital health companies for comprehensive internal testing.
Adept AI
Founder: David Luan
Valuation: Over $1 billion
Work: Developing AI agents that respond to natural-language commands to control desktop applications, and building a platform for deploying custom agents to automate corporate workflows and integrate various software tools.
Recent Developments: In 2024, Amazon recruited two-thirds of Adept’s team, including CEO David Luan and four co-founders. The remaining Adept team continues to develop agentic AI solutions with proprietary models and custom infrastructure.
SuperAGI
Founder: Ishaan Bhola and Mukunda NS
Valuation: $10 Million
Work: SuperAGI is pushing the boundaries of agentic AI by open-sourcing its technology stack, including LAMs and autonomous agents. This move encourages collaboration and customization. Their SAM-7B model outperforms GPT-3.5 on reasoning tasks, while their VEagle vision model aims to interact with software similarly to human capabilities.
Recent Developments: SuperAGI aims to be among the top three AI research firms by 2024, expanding its customer base and launching a marketplace for AI employees. Their commitment to open-source innovation and monetizable applications is driving progress toward making AGI a practical reality.
Moveworks
Founder: Bhavin Shah
Valuation: $2.1 Billion
Work: Moveworks’ agentic AI solution integrates several elements: autonomous goal-setting and execution for complex tasks, LLM integration for fluent responses, and Retrieval-Augmented Generation (RAG) for improved decision-making. Their cross-system operation, supported by partnerships like Microsoft’s, ensures seamless functionality across enterprise platforms, focusing on enhancing workflows in areas like IT support and HR.
Recent Developments: Moveworks is leveraging agentic AI to enhance productivity by combining autonomous goal-setting and execution with a modular architecture for seamless cross-system operations.
Beam
Founder: Jonas Diezun, Aqib Ansari
Work: Beam.ai specializes in Agentic Process Automation by offering a platform for managing AI agents across business processes, including an AI Agent Hub, pre-trained templates, and customizable solutions. Their focus on operational efficiency, integration with existing tools, and enterprise-grade security ensures robust and adaptable automation solutions.
Recent Developments: Recent developments at Beam.ai include the launch of a platform for creating, testing, and deploying AI agents, expanding their tools and integrations, and developing industry-specific solutions for sectors like insurance and healthcare.
NinjaTech AI
Founder: Babak Pahlavan
Valuation: $4.73 Billion
Work: Developing an AI assistant platform with agentic capabilities, including specialized agents for tasks like coding, advising, and image generation. The platform integrates over 20 external AI models from providers like OpenAI and Google, features a proprietary coding model based on Llama 3.1 405B, and offers personalized AI avatars fluent in over 20 languages.
Recent Developments: Launching a platform for users to access and compare results from multiple AI models, while developing agents that autonomously handle complex tasks through reasoning and planning. They are also implementing AWS AI chips to boost performance and efficiency, and offering a subscription-based model for their AI assistant and tools.