Ai Agent Adoption And Budgets Will Rise Significantly In 2026, Despite Challenges

Iana Kunitsa/Moment/Getty Images

Follow ZDNET: Add america arsenic a preferred source on Google.

ZDNET's cardinal takeaways

• Most IT leaders (82%) judge that information integration is simply a large AI challenge.  
• Agentic AI pioneers usage 10% much package applications.
• IT leaders will walk 19% of their IT budgets connected AI agents this year.
  

Organizations are quickly adopting autonomous systems. The mean endeavor now manages 957 applications -- rising to 1,057 for those furthest on successful their agentic AI journey -- pinch only 27% of these applications presently connected, creating a important situation for IT leaders aiming to meet their near-term AI implementation goals.

That's 1 of nan cardinal takeaways from nan 2026 Connectivity Benchmark Report from Salesforce, MuleSoft, and Deloitte Digital. The eleventh yearly study is based connected study information and interviews pinch 1,050 IT leaders globally. 

Also: Microsoft and ServiceNow's exploitable agents uncover a increasing - and preventable - AI information crisis

Despite nan wide inclination of agentic transformation, organizations look important obstacles. The superior challenges stem from complex, tightly coupled IT environments and accumulated bequest method debt, which necessitates contiguous process streamlining. 

A important consequence is that 71% of respondents reported that their systems are overly interdependent. This coupling critically hinders AI adoption, pinch information integration being cited arsenic a awesome hurdle by 82% of respondents.

Companies that person afloat embraced agentic translator are utilizing importantly much applications than those still successful nan readying phase. On average, companies pinch afloat agentic take reported 1,057 applications, a important quality compared to nan 669 applications utilized by those successful nan readying stages. 

Also: AI is collapsing nan profession ladder - 5 ways to scope that activity domiciled now

Yet contempt this precocious measurement of applications, connectivity remains debased crossed nan board. The wide mean for connected applications was only 27%. The bully news is that companies that person achieved agentic translator demonstrated amended integration, pinch astir one-third (32%) of their applications connected. 

Many organizations person information trapped crossed their exertion ecosystem. Trapped information translates to trapped worth creation for employees, customers, and business partners. Business leaders must admit that AI projects are information projects. The biggest threat to organizations optimizing their effect utilizing AI will beryllium silos -- data, context, knowledge, and contented siloes. 

Agents are everywhere 

The study revealed that 50% of AI agents run successful silos alternatively than arsenic portion of a multi-agent system. That said, only 2% of IT leaders said they person nary plans to adopt AI supplier protocols, while agent-to-agent (A2A) and model discourse protocol (MCP) were utilized by 40% and 39% of respondents, respectively. Thankfully, integration is simply a cardinal attraction area -- 86% of IT leaders said that, without due integration, AI agents tin present much complexity than value. 

Organizations are quickly accelerating their take of AI agents, pinch nan existent mean utilization opinionated astatine 12 agents per organization. This complaint is projected to summation by 67% complete nan adjacent 2 years, reaching an mean of 20 AI agents. 

Also: Forget nan main AI serviceman - why your business needs this 'magician'

AI agents person achieved near-universal take wrong organizations, pinch 89% of teams and functions reporting their use. This wide implementation has led to a important summation successful IT fund allocation. Digital leaders expect dedicating an mean of 19% of their IT budgets complete nan adjacent twelvemonth to initiatives focused connected AI agents and agentic transformation.

Companies instrumentality AI agents successful 3 superior ways: utilizing pre-built software-as-a-service (SaaS) agents (30%), integrating embedded agents successful existing endeavor platforms (36%), aliases processing civilization agents internally (34%). Effective AI supplier implementation hinges connected seamless information integration, a belief held by almost each IT leaders (96%). 

Without due integration, a ample mostly (86%) of IT leaders said they interest that AI agents could summation complexity alternatively of delivering value, underscoring nan captious relationship betwixt information and successful AI supplier adoption.

Roadblocks to agentic success

The study suggests AI translator is hard. Nearly each (97%) IT leaders reported facing challenges pinch their agentic translator projects. Access to trusted information is simply a cardinal obstacle. IT teams now walk an mean of 36% of their clip designing, building, and testing caller civilization integrations betwixt systems and data. Custom activity will not thief standard AI adoption. 

The apical challenges to agentic translator included: 

• Risk management, compliance, security, and ineligible implications -- 42%
• Lack of soul expertise successful AI and aliases AI supplier creation -- 41%
• Legacy infrastructure aliases systems incompatibility -- 37%
• Integration of siloed apps and information -- 35%
• Business and IT alignment -- 32%

The astir important obstacle to successful AI aliases AI supplier deployment is information quality, cited arsenic nan apical interest by 25% of organizations. Furthermore, almost each organizations (96%) struggle to usage information from crossed nan business for AI initiatives. Other cardinal challenges included cybersecurity risks and nan trouble presented by outdated IT architecture aliases infrastructure. Most IT leaders (94%) noted that AI agents will require an IT architecture that is much API-driven. 

Value creation astatine nan velocity of need 

AI agents are delivering important value, pinch 94% of respondents confirming they dramatically summation nan velocity and ratio of improvement teams. 

This ratio summation allows nan aforesaid percent (94%) of IT teams to attraction connected higher-value tasks. Moreover, AI tooling is changing developer preferences: a ample mostly (90%) favour AI-assisted coding devices (like 'vibe coding') complete accepted methods for creating integrations and APIs. Despite this beardown preference, organizations still look sizeable obstacles, arsenic 72% of respondents reported difficulties erstwhile attempting to leverage AI for strategy integrations.

Also: Want section vibe coding? This AI stack replaces Claude Code and Codex - and it's free

IT leaders said they expect AI agents to present respective awesome benefits, including:

• Boosting worker productivity by automating tasks.
• Accelerating workflows and regular task executions.
• Simplifying analyzable systems and processes to heighten their management.

Furthermore, adopting AI agents provides organizational advantages that widen beyond method teams. When non-technical users usage AI agents, organizations spot accrued operational efficiency, decreased IT workloads, and amended customer restitution and experience. 

IT leaders are prioritizing an integrated end-user experience, pinch 83% already providing a connected acquisition crossed astir of their channels. This accent is mostly owed to nan anticipated effect of AI agents, which astir each organizations (96%) expect to amended nan worker experience. 

Also: Turn AI chaos into a profession opportunity by preparing for these 4 scenarios

Organizations that person afloat adopted AI agents are astir 5 times much apt to study a important betterment successful worker acquisition (59%) compared to those still successful nan readying shape (12%). This integration of end-user experiences is already delivering measurable benefits, pinch complete half of IT leaders reporting outcomes for illustration accrued customer engagement and improved operational visibility.

While 98% of organizations are committed to implementing AI agents and a beardown statement (95%) exists that these devices will summation improvement squad velocity and efficiency, successful implementation remains a important concern. Nearly two-thirds (64%) of respondents said they are worried astir their organization's expertise to execute its AI supplier implementation goals. 

Agentic translator tin impact a messy beginning, middle, and ongoing journey, but location is nary extremity for now, arsenic we are only astatine nan very opening of this shift. In an property of abundant innovation, nan captious occurrence factors will beryllium culture, people, processes, and technology. 

To study much astir nan 2026 Connectivity Benchmark Report, you tin sojourn here.