Tool sprawl limits AI integration for 70% of enterprises

You don't know what "too many cooks in the kitchen" really means until you collide with your sous chef during dinner prep and send an entire cutting board's worth of sliced onions flying ceilingward. 

A lingering odor in the kitchen for a few days is one thing—but when you apply the concept to AI tools, too many is a lot more dangerous. While two AI apps working independently might not cause onions to fly, when you start pushing toward double digits, things get messy. 

Shadow AI, data silos, and integration headaches: all of these can plague your business as a result of tool sprawl. To find out how deep the problem runs, Zapier asked over 500 enterprise leaders about the number of AI tools they're using, whether those tools are properly connected, and what challenges AI sprawl is bringing to their organizations.

Key findings

• Only 35% of enterprise leaders say the AI tools used in their organization go through proper approval channels. 

• Over a quarter of enterprises (28%) now use more than 10 different AI apps.

• 70% of enterprises haven't moved beyond basic integration for AI tools.

• Most enterprises (66%) plan to raise their AI tool count over the next 12 months.

• Three in four enterprises (76%) have experienced at least one negative outcome because of disconnected AI.

• 9 in 10 enterprise leaders say having a central AI orchestration platform is critical or important.

Table of contents:

• Only 35% of leaders say all AI tools go through proper approval channels

• 70% of enterprises still haven't moved beyond basic AI integration 

• Most enterprises plan to increase their AI tool counts despite integration struggles

• AI tool sprawl costs businesses time, money, and data security

• Enterprise leaders say a central AI orchestration platform is critical for business success

• Tame AI tool sprawl with Zapier

Only 35% of leaders say all AI tools go through proper approval channels

Even AI's most bullish advocates probably couldn't have expected it to become so widespread so fast. But here we are: AI is built into pretty much every piece of software now—not to mention our phones, computers, tablets, and watches. 

For enterprises, this means employees can access hundreds of different AI apps instantly, with zero training or oversight. And according to the data, they're doing just that. Only a third (35%) of leaders say the AI tools used in their business go through the proper approval channels. 

Shadow AI, or the use of unapproved artificial intelligence, has become a real concern, with 31% of enterprises discovering "rogue" AI tools in their organizations every month. 

Broken down by department, Operations and IT teams (23%) are most likely to use unapproved AI, followed by customer service (21%) and HR (20%). On the other hand, 14% of enterprises have no visibility into what AI tools are being used by employees.

70% of enterprises still haven't moved beyond basic AI integration 

There are officially too many cooks in the kitchen: over 1 in 4 enterprises (28%) now use more than 10 different AI applications. But despite rising tool counts, a whopping 70% of enterprises still haven't moved beyond basic integration for their AI tools.

Of this group, 43% say they've developed basic connections between some tools, 22% say most of their apps operate independently, and another 6% are still in the planning stage, with no connections deployed.

Without meaningful integration, nearly a quarter of enterprises (22%) are stuck manually transferring data between siloed AI. And while over a third of executives (36%) say they've built (or plan to build) custom integrations between tools, this becomes increasingly difficult and costly as you add more software to your stack.

Most enterprises plan to increase their AI tool count despite integration struggles

With AI as hot as it is right now, asking businesses to pump the brakes is like asking a very frisky cat to climb back into the metaphorical bag.

Even as integration challenges multiply, the majority of enterprises (66%) plan to raise their AI tool count over the next year. This leaves a much smaller minority who expect their AI tool count to decrease (3%) or stay the same (28%).

Still, many leaders do want to cut back on AI sprawl, with 39% saying they've tried (or plan to try) to standardize with fewer tools. Another 38% conduct regular audits to catch shadow AI, and 44% say they've started creating detailed AI usage policies. 

With these safeguards in place, businesses should have better AI visibility and governance even as tool counts grow. But that doesn't solve the problem of siloed AI you can't connect to the rest of your tech stack. 

AI sprawl costs businesses time, money, and data security

Tool sprawl doesn't just sound bad on paper. It has real costs, reckoned in dollars, time, and even peace of mind. Compliance concerns loom especially large, as 36% of enterprise leaders claim AI sprawl is increasing security and privacy risks for their businesses. 

In total, 3 in 4 enterprises (76%) have experienced at least one negative outcome due to disconnected AI. A third (34%) of leaders believe tool sprawl makes training employees on AI a major challenge, and 30% say they're wasting money on redundant AI software.

And 29% claim manual data transfers are eating employees' time, showing that AI can actually add to the workload problems it's meant to solve—if it's not properly integrated. 

90% of enterprise leaders say having a central AI orchestration platform is critical or important for their business

Rome wasn't built in a day, and neither are fully connected tech stacks. But even if most businesses haven't tamed tool sprawl, they're at least on the right track. Today, 9 in 10 enterprise leaders think having a central AI orchestration platform is critical (45%) or important (45%) for success.

AI orchestration connects AI tools, agents, and workflows into unified, end-to-end business processes. While this level of integration is a North Star goal for many, only 35% of enterprises have actually invested (or considered investing) in AI orchestration software.

Big picture, this means businesses know what they need, but aren't quite ready to go all-in on making it a reality. (That, or they just don't know Zapier exists.) But with 98% of enterprises planning to integrate multiple AI tools, orchestration won't be optional for long.

Tame AI tool sprawl with Zapier

AI is still a shiny new technology, so people are going to adopt the latest tools whether they know what to do with them or not. Plus, the problem isn't so much the number of apps a business uses, but how unified they are.

With Zapier, the most connected AI orchestration platform, you can take a sprawling AI stack and integrate it with all the other software your teams use daily. Automate data syncing across more than 8,000 apps and create powerful AI workflows without writing a line of code.

Zapier lets you approach enterprise tool sprawl from multiple angles. Use as many AI apps and agents as you like, but keep them out of silos by adding AI directly to your workflows. Or cut down on unused software subscriptions and consolidate with Zapier's suite of built-in AI tools.

Related reading:

• AI at Zapier: How we use AI to streamline work

• Only 50% of enterprise leaders say AI benefits are widespread

• 78% of enterprises are struggling to integrate AI with legacy systems

• The AI tools graveyard: 4 ways to prevent it