The trend: 76% of US workers say their favorite genAI tools don’t have access to company data or business context, per a Salesforce and YouGov survey, which derails them from completing work-specific tasks. 

That gap is showing up in small and medium-sized business (SMB) adoption patterns too: Only 13% say they’re using agentic AI, and just 19% have built custom AI internally, per Mailchimp/WARC—indicating that most teams are experimenting at the surface, not plugging AI into the key business systems where decisions are made.

Zooming in: “AI that’s disconnected from enterprise systems never earns sustained use,” said Srini Tallapragada, Salesforce’s president and chief engineering and customer success officer.

The lack of business context helps explain why many AI pilots stall. AI tools can summarize, rewrite, and brainstorm, but they rarely become daily essentials because they can’t reliably complete business-specific tasks.

Ways secure access to company data would change work:

• 62% of US workers say it would reduce time spent searching, per Salesforce and YouGov.
• 60% say it would improve work quality.
• 59% say it would help them finish tasks faster.

The challenge: Many businesses are hesitant to connect AI to core systems because a single mistake can leak sensitive data, create compliance and security problems, and risk brand damage. Finding a safe middle ground will remain an ongoing hurdle.

Implications for brands: Human teams will lose hours tracking down the right assets, verifying the latest approved claims, and pulling in customer context to tie into desired business output. Solutions could include:

• Securely connect AI to your systems of record (CRM, DAM/PIM, CMS, analytics, and brand rules). 
• Standardize product data, claims, pricing, and naming so outputs stay consistent, accurate, and reusable.

Teams that can safely integrate AI with trusted business data will have better odds to accelerate AI use beyond pilots while wasting less time connecting the dots.