- The Angle
- Posts
- What Do We Mean When We Say AI Agent?
What do we mean when we say AI agent?
Gil Dibner
In our conversations with startups these days, the word “agent” has rapidly become one of the most commonly used and most confusing. There does appear to be a commonly accepted technical definition of “agent” as applied to software (or “agentic software”). This definition refers to any system that is both autonomous and goal-seeking. Those two attributes appear to be essential for something to qualify as an agent. Beyond that, however, there is a lot of confusion in how the term is actually used.
A quick search for agent taxonomies reveals various taxonomies that can be used to categorize agents by various attributes. Agents are sometimes categorized by their logical approach to decision-making (simple-reflex, model-based, goal-based, utility-based, or learning), by their degree of autonomy (fully autonomous, semi-autonomous, manual), by their internal complexity (single-agent, multi-agent), by their architecture (deliberative, reactive, hybrid), or other by any of several other dimensions that can get increasingly specific.
But in the context of most of our discussions with entrepreneurs, these definitions and categories are rarely mentioned. In our experience, when a technologist today says “agent” he or she is usually using the word in one of four ways:
Anthropomorphic (or literal) agents are AI agents that have a visible digital presence that users can directly interact with such as virtual assistants. Crucially, their interface is designed to mirror that of a human “agent,” usually via text or voice chat. These agents perform tasks visibly and predictably, allowing users to observe and control their actions in real-time. Good examples of visible agents would be customer service chatbots, robotic SDRs, virtual market research panels, or virtual travel agents.
Functional agents are specialized AI agents that perform specific tasks behind the scenes, often without direct user interaction. They focus on one or a few functions, optimizing workflows and enabling efficiency in automated processes. They can replicate human labor or other software systems, but they are typically autonomous enough that they are not interacted with directly. Examples of functional agents would include sales lead enrichment, marketing content generation, or procurement management.
Adaptive agents are highly versatile AI systems designed with broad capabilities to handle a wide variety of tasks, including those not known or defined in advance. These agents autonomously adjust their strategies, leveraging advanced learning algorithms and adaptability to tackle new challenges as they arise, making them well-suited for dynamic and unpredictable environments. The best examples of adaptive agents are usually offshoots of the foundational LLMs produced by the hyperscalers such as OpenAI, Google, or Meta. These foundational models (and companies building on them) can engage in a wide-range of tasks without any a priori configuration.
Composable agents are autonomous, modular building blocks within complex software systems, distinct from traditional API microservices by virtue of their ability to make decisions and manage tasks independently. Designed to be invoked and combined by other agents or function calls, composable agents provide reusable functionalities that can adapt to various contexts and support larger workflows without continuous oversight, allowing systems to scale and flexibly orchestrate multiple agents. For example, in a cloud-based architecture, a composable agent might autonomously monitor and handle data encryption, deciding on encryption levels based on data sensitivity, rather than merely responding to API calls. Similarly, in an e-commerce platform, a composable agent for order tracking could proactively manage tracking updates and optimize delivery routes in response to varying conditions, enabling other components to leverage these functions seamlessly.
We see tremendous potential across all four types of agentic systems - both on the application and the infrastructure layer. We’ve already invested in one functional agent company (still in stealth) and two infrastructure companies that can be used to assemble composable agents into next-generation agentic architectures.
But given how quickly this space is evolving, we are far from certain that the categories above are correct or even useful. And so I’ll throw the question back to you: do these categories make sense? Are we missing something? Is there a different definitional framework we should be applying? Most crucially - if you are building either an agentic application system or infrastructure for enabling such systems, please reach out. We’d love to learn about how you see the world.
FROM THE BLOG
Am I Thinking About AI the Right Way?
Gil shares the four AI themes and questions he's thinking about.
The Venture Apocalypse
The venture world is deeply debating its future, but core principles remain unchanged.
No Sleepwalk to Success
Engineering success in a technical startup.
Revenue Durability in the LLM World
Everything about LLMs seems to make revenue durability more challenging than ever.
EUROPE AND ISRAEL FUNDING NEWS
UK / SaaS. Granola raised $20M, led by Spark Capital, for its AI-powered note-taking and meeting recording solution.
UK / SaaS. Genie raised £13.3M, led by GV, for its “agentic legal editor” software platform.
Italy / Cybersecurity. Cyber Guru raised $25M, led by Riverside Acceleration Capital, for its cybersecurity training platform.
Portugal / E-commerce. Sensei raised €15M, led by BlueCrow Capital, for its AI-based online store solutions for offline retail stores.
UK / EnergyTech. Measurable.energy raised £4M, led by Vertex Exploratory Fund and Clean Growth Fund, for its AI-powered plug sockets that reduce electricity costs and carbon emissions by up to 50% in offices.
WORTH READING
ENTERPRISE/TECH NEWS
New valuation heights for Perplexity. Perplexity AI, backed by Jeff Bezos and Nvidia, is seeking to raise $500 million in a new funding round that could value the company at $8 billion, more than double its previous valuation. The AI search company, which currently generates about $50 million in annualized revenue, has recently faced challenges from media organizations over content usage, leading to the implementation of a revenue-sharing program. The company's technology provides users with instant, sourced answers by leveraging various large language models from providers like OpenAI and Meta.
Stripe is a stable(coin) genius. Payment giant Stripe has made a major push into cryptocurrency with its $1.1 billion acquisition of Bridge, a rapidly growing stablecoin infrastructure startup that processes billions in payment volume. The deal comes amid increasing institutional adoption of digital currencies, with major players like PayPal and Robinhood making similar moves into the crypto space through stablecoin launches and acquisitions. Bridge, which started by offering cross-border payment solutions and evolved to support government aid disbursement, will help Stripe advance its mission of making money easier to move across borders with minimal costs.
AI takes the wheel. Agents, agents, everywhere. Anthropic has announced major upgrades to its AI models, including an enhanced Claude 3.5 Sonnet with improved coding capabilities and a new, cost-effective Claude 3.5 Haiku model that matches the performance of its predecessor's top-tier model. The most significant announcement is the introduction of "computer use" in public beta, allowing Claude to interact with computers like humans do - viewing screens, moving cursors, and operating software interfaces. While still experimental and sometimes error-prone, this groundbreaking capability is already being tested by major companies like Asana and DoorDash, with Claude 3.5 Sonnet scoring notably better than other AI systems in computer interaction tests.
HOW TO STARTUP
Tips for AI PMs. Google's NotebookLM is a breakthrough in AI-generated audio content. The team developed a system that transforms written content into engaging, conversational podcasts that feel natural and entertaining rather than just text-to-speech summaries. On the latest Latent Space podcast, the NotebookLM team discusses their approach to building the product. They discuss how they incorporated human-like elements such as interjections, disagreements, and creative metaphors They also outline five key principles for AI product managers, emphasizing simplicity, real-time feedback, embracing variability, taste-based curation, and hands-on experimentation in product development.
The end of ARR. This presentation from Dave Kellogg (of Kellblog fame!) explores the transformative impact of artificial intelligence (AI) on SaaS pricing models. As AI integrates into software solutions, traditional metrics like ARR become less relevant, necessitating a shift towards usage-based and hybrid models. The rising costs of AI development and the diverse pricing drivers it introduces further contribute to this evolving landscape, where pricing strategy will play a pivotal role in capturing value. For a more in depth discussion, listen to this conversation between Kyle Poyar and CJ Gustafson.
ICP marketing makes a comeback. Kyle Poyar presents a comprehensive guide of over 40 marketing strategies to effectively reach and convert ideal customer profiles (ICPs) in today's evolved GTM landscape. The article breaks down these strategies into four key stages: identification of target accounts, creating awareness, generating interest, and converting prospects into real pipeline opportunities. The approach emphasizes moving beyond traditional marketing-sales divisions and MQL metrics, instead focusing on coordinated, multi-channel campaigns that leverage next-generation GTM tools and AI-enabled content creation to reach the right customers more efficiently.
HOW TO VENTURE
2% is all you need. The latest from Jamin Ball examines a growing misalignment in venture capital where fund managers (GPs) are increasingly prioritizing management fees (the "2%") over investment success (the "20%") due to ballooning fund sizes. This shift has created a problematic dynamic where VCs are more focused on rapid capital deployment than company success, leading to over-valued rounds, less engaged investors, and a more transactional approach to founder relationships. The author warns founders to be wary of this trend and suggests they should distinguish between "2% firms" (focused on deployment) and "20% firms" (focused on company outcomes) when choosing investment partners.
PORTFOLIO NEWS
Tensorleap was selected as one of the 45 startups to join Israel’s Innovation Dream Team by The Peres Center for Peace and Innovation.
Reco’s CISO Merritt Baer recently spoke on the N2K podcast about the complexities of AI security and the challenges CISOs and security leaders face.
PORTFOLIO JOBS
Vault Platform
Full Stack Software Engineer (Remote UK)
Reply