HFS Horizons: Agentic Technology, 2026

The HFS Horizons: Agentic Technology, 2026 report is for CIOs, chief AI officers, and enterprise transformation leaders evaluating technology partners to move agentic AI from pilot into scaled production across hyperscaler, SaaS/ERP, and AI-native environments.

Overview

HFS Horizons: Agentic Technology, 2026 assesses 12 technology firms powering enterprise agentic transformation, spanning hyperscalers, enterprise SaaS/ERP providers, and agentic AI-native firms. The report evaluates how effectively each firm advances the shift toward Services-as-Software™, an HFS Research paradigm in which services are delivered primarily through software rather than human effort, with outcome-based pricing and near-full machine-led execution. Data collection spanned September 2025 to January 2026 and included 24 reference checks for AI-native participants.

The chart below plots all 12 technology firms across the three HFS Horizons, showing which firms have reached the Market Leader status and which are maturing as Enterprise Innovators. 

Key findings

1. Seven firms: AWS, Ema, Google Cloud, Microsoft, Rhino.ai, Salesforce, and ServiceNow reach Horizon 3 as Market Leaders for embedding agents directly into enterprise workflows, governance layers, and systems of record.
2. Five firms: CrewAI, Lyzr, Otera, SAP, and WRITER are placed in Horizon 2 as Enterprise Innovators, with no firms in Horizon 1, reflecting the capability maturity of the cohort.
3. Hyperscalers are increasingly moving up the stack, shifting from selling tools that enable others to build agents to delivering agents and outcomes themselves. SaaS/ERP vendors are beginning to shift from selling software licenses that require human operation to offering agents that can execute defined business processes, moving gradually from tools that assist to services that act. Agentic AI-native firms win on depth and specificity by solving a narrow problem with exceptional intensity for a defined set of enterprise clients, targeting high-value workflows and delivering measurable efficiency gains in production.
4. Some of the most compelling results in this report come from firms with fewer than 100 employees. Agentic AI-native firms move faster and typically provide more direct leadership involvement than large platform vendors.
5. Agentic AI is not eliminating headcount in the firms building it but rather amplifying it. AI-native technology firms continue hiring to support agent operations, governance, and client success, but every person is using agents to multiply their output. The result is a new human-agent dynamic where growth decouples from headcount, individual capability scales and work shifts toward oversight, orchestration, and judgment, with agents handling the execution.
6. Otera is targeting 90% autonomous operations across claims and invoice processes spanning 1,000+ FTEs at Allianz Partners, while ServiceNow is achieving 80% ticket deflection at CANCOM. Salesforce Agentforce autonomously resolved 70% of customer chats for 1-800Accountant during peak tax season, and Microsoft’s banking assistant resolved 75% of customer requests across 30,000+ monthly conversations at Commerzbank.

HFS finds that every firm in this cohort has production-grade agentic capabilities available today, yet true Services-as-Software remains emergent across most of the cohort. HFS assesses that the bottleneck is no longer the technology but enterprise operating models, data readiness, and governance maturity.

The Bottom Line: The agentic technology market has stratified between firms delivering agentic AI as a platform capability and those genuinely operationalizing Services-as-Software, with commercial models shifting from subscription to usage-based pricing as revenue decouples from headcount.

Research methodology

Vendor briefings with AI leadership and other public information such as websites and publication across 12 technology firms augmented by 24 client, partner, and employee reference checks for AI-native participants, conducted globally between September 2025 and January 2026.

To read the rest of the research, click the download button below.