Agentic AI refers to AI systems that can plan, take multi-step actions, and interact with tools or applications on behalf of the user. They’re an emerging system within the broader landscape of Gen AI, which has traditionally been a multimodal reactive generator of text, image, and video. The same Gen-AI models become agentic when relevant modules for agency (e.g., additional components or capability layers) are added, such as planning, task decomposition, and function calling.

Companies are using these systems in areas such as customer support, software development, research, and day-to-day operations. These are typical areas where they expect to see noticeable efficiency gains or lower costs.

Agentic AI was widely billed as a 2025 breakthrough, but the data indicate uneven progress. On real-world assistant tasks that require tool use and web interaction, the GAIA benchmark still shows a significant capability gap: human respondents score about 92%, whereas advanced systems like GPT-4 with plugins achieve roughly 15% in the original paper, indicating fragility in multi-step reasoning and tool orchestration.

Enterprise returns remain concentrated in a small subset of GenAI deployments. The “State of AI in Business 2025” report, affiliated with MIT, describes a growing “GenAI Divide,” estimating that 95% of organizations have not yet realized financial returns from GenAI initiatives. The study attributes this to persistent struggles with context management, memory, and integration. These bottlenecks are particularly relevant to agentic AI, in which systems must autonomously plan, adapt, and act in response to changing inputs. While the report focuses on GenAI broadly, its findings mirror the real-world frictions organizations face when deploying agent-based systems.

At the same time, technical capability is improving in narrower domains relevant to agents. OpenAI’s 2025 report showed substantial gains on SWE-bench Verified, a human-validated software engineering benchmark often used as a proxy for autonomous tool use in coding workflows, indicating that planning and multi-step execution can advance rapidly when tasks are well-scoped. This suggests headroom for agentic systems in constrained environments, even as general reliability lags.

Spending forecasts imply continued investment regardless of the ROI bottleneck. International Data Corporation projects worldwide artificial-intelligence outlays of roughly $632 billion by 2028, with the US generative-AI spending alone expected to reach about $108 billion that year. Europe-focused releases indicate similar double-digit compound growth, reinforcing a multiyear capital-expenditure cycle for model development, orchestration, and safety tooling required for dependable agents.

Taken together, the picture is consistent: agentic artificial intelligence is scaling in budget and improving on targeted, benchmarked tasks, but broad, cross-workflow reliability and thus returns remain limited. Benchmarks such as GAIA and enterprise surveys align on the same constraint: multi-step robustness, memory, and integration are the choke points that will determine how fast agentic systems move from pilots to profit.

With that background, let’s explore our selection of the 10 best agentic AI stocks to buy now.

Our Methodology

For our list, we screened for stocks that offer agentic AI or agentic AI infrastructure as a core product. From these, we selected stocks with the highest number of hedge funds holding stakes as of Q3 and ranked them accordingly.

Why are we interested in the stocks that hedge funds pile into? The reason is simple: our research has shown that we can outperform the market by imitating the top stock picks of the best hedge funds. Our quarterly newsletter’s strategy selects 14 small-cap and large-cap stocks every quarter and has returned 427.7% since May 2014, beating its benchmark by 264 percentage points (see more details here).

10. Veritone, Inc. (NASDAQ:VERI)

Number of Hedge Fund Holders: 11

Veritone, Inc. (NASDAQ: VERI) is one of the best agentic‑AI stocks to buy now, given its focus on AI infrastructure and data tokenization.

Recently, D. Boral Capital reaffirmed its bullish stance on VERI: on December 2, 2025, its analyst, Jason Kolbert, maintained a Buy rating with a $23.00 price target, and again on December 9, 2025, reiterated that Buy rating and target, signaling no change in conviction despite market turbulence.

In another development, on December 2, 2025, Needham & Company LLC analyst Joshua Reilly reiterated his Buy rating on Veritone, keeping a $10.00 price target. Reilly explained that his positive stance reflects confidence in Veritone’s strategy around unstructured‑data management and AI licensing: following Veritone’s recent Virtual AI and Data Economy Forum, Needham highlighted how the company’s approach to converting unstructured data into tokenized, licensable assets could give it an edge, especially with hyperscalers.

Beyond those two firms, the broad analyst consensus on VERI as of December 10, 2025, leans toward “Buy” or “Moderate Buy.”

The average 12‑month price target sits around $11.00–$11.20, with a low of about $6.00 and a high of roughly $23.00. That spread underlines the divide between cautious and bullish estimates, but even the median suggests nearly 100% upside from the current price level.

Veritone, Inc. (NASDAQ: VERI) develops AI‑powered software that converts unstructured data—video, audio, sensor output, and similar sources—into actionable intelligence via its aiWARE platform.

9. C3.ai, Inc. (NYSE:AI)

Number of Hedge Fund Holders: 21

C3.ai, Inc. (NYSE: AI) is one of the best agentic AI stocks to buy now.

On December 9, 2025, the company announced that the U.S. Army’s Rapid Capabilities and Critical Technologies Office (RCCTO) selected C3.ai to deliver an “AI contested logistics” solution for combat operations.

The deal tasks C3.ai with building a system to manage and forecast the critical logistics needs of forward‑deployed Army units operating in contested, high‑risk environments. That entails predicting requirements for spare parts, fuel, and munitions in advance. The software will plug into brigade‑level Command & Control networks so supply and resupply efforts can keep pace under duress, helping maintain operational tempo and readiness when disruption or unpredictability is highest.

C3.ai isn’t unfamiliar with this space. The contested‑logistics project leverages its existing products, such as the “C3 AI Contested Logistics” and “C3 AI Readiness” applications, which have reportedly been deployed for other U.S. defense clients, including the Defense Logistics Agency and the U.S. Air Force.

This win for C3.ai suggests a broader strategic shift: militaries recognize the value of real‑time, data‑driven logistics powered by AI. If the system performs as designed, it could give deploying forces a decisive edge in managing supply chains during periods of chaos.

C3.ai, Inc. (NYSE: AI), based in Redwood City, California, is an enterprise‑AI software company behind the C3 Agentic AI Platform, C3 AI applications, and C3 Generative AI, offering integrated AI tools across a wide range of industries, from defense to healthcare.