AI accelerates content creation but consumes saved time in editing, strategy and client review at US technology company in 2026

This has been part of my AI experience in 2026. I use it, I like it, and it has helped me do things that would have taken considerably longer before. Somehow, my workday hasn’t received the good news.

In the first quarter, we were sometimes asked to produce up to 15 pieces of content a day, alongside strategy, media relations, campaign management and client counsel. AI-generated scopes arrived, designers received AI visuals to emulate, and delivery expectations started travelling at the speed of a rocketship leaving Earth for a more peaceful universe.

I understand how we got there because I contributed to it. You finish something faster and think, wonderful, I can tackle that other thing. While that is generating, you open another project. By lunchtime, you have several promising beginnings and one human being attempting to bring them all to a useful conclusion.

Berkeley researchers spent eight months studying work at a US technology company and observed employees taking on broader tasks, working faster and extending work into more hours of the day, often without being asked. It was one company, but I recognised the pattern. I had saved time and spent it before getting out of my chair.

That might be manageable if everything we started was as quick to finish. But getting a first version from AI and getting it ready to use are two different things. There is usually more thinking, checking and reworking to do than the speed of that first response suggests.

I’ve received excellent AI-assisted briefs that helped us understand a client’s needs sooner. AI has also helped me explore more possibilities when developing a strategy. As the suggestions arrive, though, I still need to keep asking questions. Would that platform reach the right people? Could we sustain that content schedule? What are those projections based on? A useful suggestion can take some working through before we know whether it belongs in the plan.

Designers and developers have their own questions to ask. A generated visual can explain an idea wonderfully, but how would it work within the corporate identity, and what would it cost to build? Generated code may need considerable work before it functions reliably. That first version gives everyone something to work with. We still need time for people to apply their experience, spot what needs attention and work through it. That time is easy to overlook when we’re already moving on to the next thing AI has helped us start.

In my own website briefs, SEO writing and UX implementation, AI has helped me work through substantial material. Even with detailed instructions and source documents, I’ve had to correct errors and restore missing information. Spotting those gaps requires an understanding of the client’s offering and what people need from the website. Otherwise, an omission can carry through into the brief, search content, and the journey toward an enquiry.

Clients have budgets to manage and people asking them to do more. Of course they are exploring tools that promise to help. We are doing exactly the same. The difficulty comes when the speed of producing that first version becomes the basis for estimating everything that follows.

And much of our work doesn’t begin with something to produce. A client may need help navigating a sensitive issue, approaching stakeholders or understanding why a campaign isn’t gaining traction. We might advise against a public statement, recommend a conversation first or question whether an announcement offers journalists anything worth covering.

AI helps me research the background and explore approaches. I still have to understand the people involved and take responsibility for my advice. After more than three decades in communications, experience influences what I notice and the questions I ask. Sometimes I recognise a problem because I’ve seen a version before. Sometimes we simply don’t know enough yet.

Even writing up a recommendation can expose an assumption that needs testing, sending me back to the client with another question. The temptation, when a platform is offering to finish in seconds, is to keep asking it to fix what you don’t like. I’ve done that. Five versions later, I’m still unhappy because I haven’t resolved my own thinking. The anticipated saving has disappeared into a lengthy discussion with something relentlessly willing to try again.

A 2025 study of 319 knowledge workers found that greater confidence in AI was associated with less self-reported critical thinking. That gave me reason to consider my own habits and whether we’re leaving enough room for people to question what it produces.

They learn through developing recommendations, working with clients and seeing what happens when their advice is implemented. I want them to question my thinking when they see something I’ve missed. There is little room for that if every available minute goes toward getting through everything we can generate now.

So, I’m looking more carefully at what we commit to and how we estimate the work. Clients should benefit from faster delivery where we can achieve it. They also need us to recognise when a request requires investigation or a difficult conversation before we proceed. We need to allow for that when agreeing scopes and deadlines, with input from the people responsible for implementation.

And yes, I worked with AI on this piece, from researching the evidence to debating the wording. It’s a great partner. We still had some spirited discussions about what I meant and what had gone missing along the way. Which rather illustrates the point.

I love communications and intend to keep exploring what AI can help us do, while keeping my own brain well clear of mashed potato. That means stopping myself from automatically filling every saved hour and considering where it would do the most good for Mango and our clients. Sometimes we can deliver sooner. Sometimes an idea deserves more attention, or someone on the team needs time to work through a recommendation. Those are business decisions I want to make properly, preferably before I’ve opened another five projects.