Artificial intelligence (AI) agents are a fast-growing sector, propelled by rapid advances in large language models (LLMs), declining compute costs, and improved integration with enterprise software.

Since 2023, more than 500 startups have been founded across over 20 categories, according to CB Insights.

Booming funding activity has further helped propelled this market. In 2024, AI agent startups raised US$3.8 billion, nearly tripling 2023’s total.

Top AI Agent Trends in 2026

Every major tech player is now deploying agents or agent tooling. Microsoft, for example, integrates AI agents broadly through Copilot across Microsoft 365. Google DeepMind has introduced several agent initiatives including Project Astra, a prototype exploring future capabilities of a universal AI assistant; Project Mariner, which explores the future of human-agent interaction; and Jules, an AI-powered autonomous coding agent.

Most recently, Lenovo debuted Lenovo and Motorola Qira, a cross-device AI “super agent” positioned as a “Personal Ambient Intelligence System”. Designed to work across Lenovo and Motorola devices, including PCs, smartphones, tablets, and wearables, Qira aims to deliver a unified intelligence layer that supports everyday tasks, creativity, and communication, while processing inputs only when users choose to enable them.

Looking ahead, CB Insights expects momentum in the agentic AI landscape to continue. In a report released in October 2025, it shares its predictions for 2026, forecasting accelerating investment in voice-based interfaces, mergers and acquisitions (M&A) consolidation around agentic platforms, and advances in agentic commerce and payments, among the key trends for the year to come.

Voice AI accelerates

In 2026, development in voice AI will accelerate as leading generative AI (genAI) companies continue to concentrate investment in the technology. These ventures are positioning themselves for a future where humans interact with AI via conversation rather than text interfaces.

This trend is reflected in recent moves by major tech companies. Last year, Meta acquired voice AI startups Play AI and WaveForms, two startups developing audio AI systems that can emulate human speech indistinguishably.

Meanwhile, OpenAI and Microsoft have upgraded their AI applications with enhanced voice capabilities and improved audio models, further signaling the shift toward voice-first AI interaction.

Agent AI sees wave of M&A

In Q1 2025, agentic solutions led the top AI exits. Moveworks, Weights and Biases, and OfferFit secured the three largest deals among 85 acquisitions at valuations of US$2.9 billion, US$1.7 billion, and US$325 million, respectively. These transactions underscore agent AI as the primary focus of industry consolidation.

Overall, more than 35 acquisitions took place in the broader AI agent and copilot space in 2025 as enterprise buyers sought to build comprehensive agent solutions to gain a competitive edge.

CB Insights expects consolidation in two more areas: sales and marketing AI agents, which represent a low-hanging fruit for software-as-a-service (SaaS) leaders’ agent plans; and the coding AI agent and copilots market, a fractured space which has seen explosive growth and soaring valuations, making it ripe for consolidation.

Economic pressures mount

In 2025, AI coding tools were the first area to feel shrinking profit margins. Now, that pressure is spreading to  the broader AI agent market.

Smarter and more capable AI models are significantly more expensive to run, leading costs to rise faster than revenue and compressing margins. This forces many startups to change pricing, business models, or sell themselves.

CB Insights expects other agent categories with growing usage to rework their pricing models and contracts as reasoning costs mount. Furthermore, as margins tighten, struggling players will seek exits, a trend already evident in the acquisition of Windsurf by Cognition after the startup’s licensing deal with Google.

Agentic commerce advances

A new class of startups is emerging to support agentic commerce. These companies are building AI-native payment rails and digital wallets that let users authorize spending by AI agents, laying the groundwork for fully autonomous shopping.

Stripe, for example, launched in September 2025 an API for agentic payments. It also introduced with OpenAI the Agentic Commerce Protocol, which is designed to enable AI agents to complete purchases on behalf of users.

Agentic AI is set to influence over US$1 trillion in e-commerce spending. According to research by Boston Consulting Group (BCG), 81% of US consumers expect to use agentic AI tools to shop, shaping more than half of all online purchases in the near future.

Control over data

As AI adoption grows, software incumbents are increasingly walling off access to their customer data, recognizing it as a strategic advantage for building AI features. This creates challenges for AI startups whose products depend on pulling data from multiple different enterprise systems at once in order to automate workflows. It also creates friction for enterprises seeking deep integration across their apps.

In response, a counter-movement toward openness is starting to appear. In September 2025, Snowflake launched a consortium with over a dozen providers, committing to standardized data formats and improving data access. This development signals enterprise demand for openness over vendor lock-in.

CB Insights expects incumbents to continue restricting API access to protect their AI product revenues. Meanwhile, enterprises that want more control over their data will push back by investing in infrastructure that lets them own and manage their data directly.

Against this backdrop, some startups will adapt with agents that work despite limited API access, while others will move closer toward where data is stored and processed.

Agent monitoring tools emerge

Finally, CB Insights expects increased activity in the agent monitoring landscape, especially across voice agent testing, synthetic user generation, and AI productivity measurement. This comes as AI agent reliability remains a challenge in the sector, with failures, hallucinations, and unpredictable behavior creating operational problems.

Several deals in 2024 and 2025  reflect enterprise demand for tools that quantify AI agent ROI, and human-AI workforce productivity. Cekura and Coval, both focusing on voice AI testing and simulation, raised US$2.4 million in July 2025 and US$3.3 million in January 2024, respectively. Larridin, which builds technology to help large organizations measure, manage, and scale their use of AI tools across the company, secured US$17 million in seed funding in March 2025.

Featured image: Edited by Fintech News Switzerland, based on image by thanyakij-12 via Freepik