While AI tools are proving vital for worker productivity, many enterprises are reporting serious problems with reliability - and it’s because they’ve been woefully underprepared for integrating the technology.
A new survey of backend engineers and IT decision makers by Temporal Technologies found that 94% are using AI in their workflows - mostly through tools like Copilot or ChatGPT.
However, just 39% are building reliable internal frameworks to support the adoption of the technology. A key stumbling block is the underlying infrastructure required to scale the technology at an enterprise-wide level, the study noted.
"Most teams think they’re modern. Most teams also admit they’re stuck," one respondent said. "Somewhere between post-build ('we built it!') and pre-scale ('oh god, how do we scale this thing?'), the real challenge begins."
Nearly half of large companies rely on custom-built workflow solutions, compared with just a third of smaller organizations, the survey found.
Three-quarters of teams said their workflows are hampered by issues such as insufficient support for long-running processes and high operational overhead, both cited by 35%, along with failure recovery challenges, a problem for 34%.
Reliability is more important than cost, performance, and speed, with 36% of engineering and IT leaders saying that reliability and compliance are their top development priorities over the next 12-to-24 months.
Notably, reliability and compliance are now bigger priorities than automation, cited by 33% of respondents, and reducing technical debt (30%).
Developers are getting stuck with AI tools
According to the survey, nearly a third of developers said that complex, long-running workflows break constantly, which hampers efficiency and slows down development processes.
"Everyone’s ‘using AI,’ but very few know how to keep it from falling over,” one respondent said.
“That knowledge gap isn’t just inconvenient, it’s a dealbreaker when you’re trying to scale agentic systems without a way to retry, resume, or even observe what went wrong.”
Tooling priorities and decision-making power are also misaligned, the survey found. In smaller companies, half said that developers typically lead tooling decisions, while in enterprises, that responsibility shifts to IT managers and CIOs at 53%.
For developers at larger enterprises, this means new tools and solutions are essentially dropped in their lap, and responsibility for implementation is left to them.
"The report tells us a lot of what we hear from partners every day—backend challenges aren’t just technical, they’re also organizational,” said Samar Abbas, co-founder and CEO of Temporal Technologies.
“Engineers and decision makers are prioritizing different things, and that disconnect is driving tooling delays, reliability risks, and rising complexity across the stack. AI is only adding another layer of scale and unpredictability.”
Security concerns are rising
Decision makers ranked security as their top concern, with nearly half saying they were losing sleep over customer churn, their biggest concern during outages.
Another 47% said that downtime drives up operational costs, while only 5% said that failures would have no major impact.
Four-in-ten respondents reported that AI’s biggest impact was in code generation - a growing trend in recent months.
A survey from Clutch revealed that 53% of senior software developers believe LLMs can already code better than most humans. However, many voiced serious concerns about data privacy and security risks.
The findings of the Clutch survey align with a previous study on the topic from Cloudsmith, which warned developers are placing too much faith in AI code generation and opening themselves up to potential security risks.
Make sure to follow ITPro on Google News to keep tabs on all our latest news, analysis, and reviews.
MORE FROM ITPRO
- AI coding tools are finally delivering results for enterprises
- Big tech promised developers productivity gains with AI tools – now they’re being rendered obsolete
- AI coding tools aren’t the solution to the unfolding 'developer crisis’
-
More transparency needed on sprawling data center projects, activists claimNews Activists call for governments to be held accountable when data centers are pushed through without proper consultation
-
Red Hat eyes tighter data controls with sovereign support for EU customersNews The company's new offering will see support delivered entirely by EU citizens in the region
-
Microsoft and Nvidia are teaming up again to support UK startupsNews Agentic Launchpad will offer participants AI expertise, training and networking, and marketing support
-
AI resume screening, recruiter chatbots, and ‘ghost jobs’ are causing havoc for struggling entry-level workersNews A new report shows employers are cutting back on entry-level hiring, but expect things to improve eventually
-
Microsoft’s huge AI spending still has investors sweating despite solid cloud growthNews Capital spending at Microsoft continues to surge, despite previous claims it would cool down
-
Analysts warn AI layoffs could spark a new wave of offshoring – enterprises are rehiring after workforce cuts, but either outsourcing or at lower rates of payNews Analysts expect a wave of rehiring next year in the wake of AI layoffs. That may sound like good news for workers, but it'll probably involve offshoring or outsourcing.
-
UK firms are pouring money into AI, but they won’t see a return on investment unless they address these key issuesNews An SAP report projects increased AI investment, but cautions that too many organizations are taking a fragmented approach
-
Employee ‘task crafting' could be the key to getting the most out of AINews Tweaking roles to make the most of AI makes you more engaged at work
-
Salesforce could become the king of enterprise AI – but only if customers believe in its potentialAnalysis At Dreamforce 2025, Salesforce painted a believable picture for enterprise AI, but shareholders will only be reassured by greater business buy-in
-
OpenAI has a bold plan to pay for its $1 trillion spending spree: Ads, personal assistants, and cheaper subscriptionsNews OpenAI has lined up more than $1 trillion in spending – and now it's trying to figure out how to pay for it all.
