Workers are getting to grips with agents, says Slack data; not really, says Lucidworks agentic researcher - take your pick!

Back in October last year, two-thirds of desk workers had not tried AI tools. Flash forward to today and 60% reckon to now use AI at work, with daily usage up 233%.

That’s according to the outcomes from the latest Slack Workforce Index, based on polling of 5,156 workers across Australia, France, Germany, Japan, UK and the US.

Who’s using it? What they’re getting out of this? Daily AI users are more likely to report ‘very good’ or ‘good’ productivity (83%), being able to focus (81%), being flexible (79%) and managing stress (77%).

There are the inevitable regional variations at play. In the US, 88% talk about better productivity and 87% cite ability to focus being good or very good, while in the UK the ability to access people, files and resources comes out at 88%, followed by good or very good productivity,

Meanwhile, counter to the ‘robots coming to steal our jobs’ narrative, 80% report a good or very good sense of belonging, the same percentage point to more job satisfaction, while 77% cite a good or very good work/life balance. In the UK, job satisfaction that is good or very good is pointed to by 87% of respondents, while in the US it’s 83%. In comparison, for non-AI users in the UK, that percentage drops to 68% and in the US to 77%.

The top three ways AI is said to have increased worker productivity are eliminating need for extensive research (51%), assisting with writing/communication (50%), and helping brainstorm to overcome creative blocks (45%. US respondents cite providing real-time feedback/improving performance highly (52%), while their UK counterparts cite this even more highly at 54%.

And 96% of AI users overall have used AI to perform tasks they didn’t have the skills to do themselves.

Interestingly, AI is said to enhance rather than replace connection among workers. For example, half of Gen Z respondents say AI hasn’t changed how often they collaborate with colleagues, while 29% say it’s actually increased their engagement as a team.

Agents incoming

Given the current rise of agentic tech, it’s interesting to note that 40% of respondents say they work with agents, with 30% of Millennials claiming to “thoroughly understand” AI agents, more than Gen Z  on 22%. This is most likely attributable to Millennials probably holding more senior positions within organizations - some 43% of executives report daily AI use overall, along with 35% of senior managers and 23% of middle managers. These people are using AI in a business context rather than a personal one and hold decision-making roles that Gen Z employees do not yet.

That said, one conclusion from this study is that desk workers - as opposed to execs/management - are becoming more comfortable with both AI in general and agents in particular. Some 40% of desk workers polled say they are now using agents. The more they use them, the more trust in the tech will grow, argues Lucas Puente, Slack VP of Research:

As workers actually use and experiment with AI agents, their trust and enthusiasm in this technology grows – and we see them leveraging agents to unlock new skills and opportunities in their everyday work.

As to what early adopters of agents are looking for:

• 72% want AI agents to augment or automate brainstorming.

• 80% want AI agents to assist with research for deal support.

• 82% want AI agents to create presentations.

On the other hand…

Of course, it’s 2025 and barely a day goes by without a fresh piece of AI/agent-related research appearing in our in-boxes. Co-inciding with the Slack ‘glass very much half-full’ worldview, along comes a new study from Lucidworks into agents that lends itself to a more 'glass half-empty' reading - only 35% of organizations meet the minimum requirements for successful use of agentic AI.

And according to this study, Dawn of the Agentic AI Era, agentic AI adoption is very low, with only six percent of companies having implemented more than one Agentic AI solution, and most of those that have are in the technology industry sector.

It should be noted at this point that this downbeat assessment of agentic adoption comes from none other than…er, an actual agent. As well as tapping into the opinions of 1,600 CEOs, CTOs, and other AI strategy decision-makers from 15 industries across eight countries, Lucidworks basically sent out its own agent, Guydbot, to top this data up:

While other reports rely on self-reported data and aspirational roadmaps, our agentic AI tool has autonomously evaluated actual digital experiences across 1,100+ companies to document which AI capabilities are in production today. This complements insights from our survey of 10,000+ AI leaders over three years.

Overall, the conclusion that Guydbot comes back with is one of widespread anxiety and confusion among AI adopters. Some 83% of respondents to the Lucidworks poll reported feeling ‘Major’ or ‘Extreme’ concern over their AI progress, an eightfold increase on the position two years ago.

Mike Sinoway, CEO of Lucidworks, is quoted as arguing:

If you think of Agentic AI - AI that performs tasks - as a car, then you can imagine Generative AI as the engine, and data as the fuel. Our report finds that too many e-commerce companies are trying to build Formula One racers around go-kart engines - and they might not even have enough gas to fill their tanks. To put it another way: companies have been trying to run before they can walk because of top-down pressure. Now panic is setting in. Many are reluctant to admit how far behind they really are, but we have the receipts to prove it thanks to Guydbot.

My take

As noted, another day, another piece of AI/agentic research. Which to take on board? Well, ultimately that's up to you. The Slack data makes for interesting reading as much for the acceleration of interest in and acceptance of AI in the workplace that it indicates, particularly the suggestion that desk workers are engaging, at a very early stage, with agents. The Lucidworks study is a different beast and covers a lot more ground across a more segmented study base, so it isn't ultimately a case of comparing apples and apples. If neither of them suits your worldview, never mind. There will be another report along tomorrow, no doubt.