Hi there!

Welcome to the 34th edition of Work in Beta.

In this edition: you're using AI every day, so why isn't work actually faster? The answer isn't a better AI tool, and it isn't more automation. It's one question almost nobody has asked about their workflow.

So, let’s dive in!

You've used AI all week. Tuesday's report draft took 10 minutes instead of an hour. The meeting summary wrote itself. Yet on Friday evening, your week looked exactly as full as it did before you'd ever opened ChatGPT.

So no, you're not imagining it. The gap between faster tasks and an unchanged week is real. Watch what happens to the time those tasks save. In a six-month experiment across 66 firms and 7,137 knowledge workers, the people using Copilot got two hours a week back from email. The researchers found no sign of those hours turning into more finished work, or into different work. And your own Friday evening tells you it didn't turn into a shorter work week either.

So where does the saved time actually go?

To see where the time goes, look at what a workflow actually is. A workflow is how a piece of work travels from start to finish: a trigger starts it, steps follow, and it ends with something delivered. Then it runs again: next week, next month, every time the trigger fires. That repeat matters, because whatever is slow inside the workflow repeats with it.

Take the monthly report. The trigger is the calendar. Then the steps: you email five people for their numbers and chase the ones who forget, you collect what they send, you write the report, someone checks it, someone approves it, you send it. Then you update the tracker so next month starts clean.

Seven steps, one trigger. The AI tools you've added this year sit on exactly one of the seven: "write the report." The other six take just as long as they did in 2022: you still email and remind, collect what arrives, wait for the check, chase the approval, send, and update the tracker.

That's where the saved time goes. The step got faster. The workflow didn't.

Now look at those six steps again. Every one of them was designed before AI existed. Each was once the best answer to a real problem, and AI has since changed the problem. Nobody has gone back to ask whether these steps still need to exist, or whether the workflow still needs this shape.

So we go back and re-ask. We re-ask this one question for every step of the workflow:

"If AI had always existed, would we have designed this step at all?"

For example, the "summarize the call for the team" step exists because reading a full transcript used to cost someone an hour. Now AI reads the transcript, and anyone who missed the call can get answers straight from it. Ask our question and the answer is no: nobody would have designed the summary step if AI had always been there.

That was one answer: no. Ask the question across a whole workflow and the answers come back in four kinds. Each kind tells you what to do with the step:

Now run the question on our monthly report. The trigger is still the calendar. The emails and reminders for numbers? Gone (remove). AI reads the numbers straight from the five tools they live in (automate). Nobody needs to be asked, and nobody needs to be chased. AI writes the first draft (assist). You give it one real review: judge whether the story is right, question the numbers you doubt, and sign. The check and the approval fold into that same review (protect). The report sends itself, and the tracker updates itself (automate). Same report, same deliverable. Of the seven steps, the only one that still needs you is the judgment.

Something shifts when you see the report this way. You stop asking "where can I add ChatGPT?" and start asking "which of my steps should stop existing?" That's the difference between adding AI and redesigning with it.

One warning before you mark your own list. "Automate whatever repeats" sounds like the obvious rule. It isn't: writing the report repeats every month, and it still needs fresh thinking every month. A step is automatable when its rules never change, not when it merely repeats.

There's one mistake that can undo this whole redesign. In the new workflow, AI does a lot on its own: it collects the numbers, writes the draft, sends the report, updates the tracker. Handing over that much can feel uncomfortable. So most people add a review after every step AI touches, just to be safe. Now count what that did to the workflow. We removed one step, the chase. We added four new ones, the checks. The workflow is longer than it was before AI.

So put your judgment at one gate instead: the point where a mistake becomes expensive. Too many checks eat the time you just saved; too few let errors slip through quietly. For the report, that gate is the one you've already seen: the single real review before the report goes out. A small error from earlier in the workflow gets caught there anyway. One real review beats five thin approvals.

The gate works as long as the month is a normal one. Some months aren't. The team gets reorganised, a big cost lands, the numbers need background that AI doesn't have. In a month like that, AI won't warn you. Its draft will arrive looking as confident as ever. Noticing the unusual month is your job. When you see one, take the work back and do it yourself.

Remember the one step that still needed you? This is what it looks like. One real review in a normal month. The whole job back in your hands in a strange one.

Pick one workflow that repeats: your version of the monthly report. Write down its steps, from what kicks it off to what you update after. Ask our question of every step: "If AI had always existed, would we have designed this step at all?" Mark each one: remove, automate, assist, or protect.

Don't build anything yet. But if a step got marked remove, act on that one now: stop doing it for one cycle and see if anyone asks for it. Deleting a step needs no tool and no setup, and your workflow gets shorter this week.

Everyone is comparing AI tools. Almost nobody is re-asking whether the steps around those tools should still exist. The tools are rented: everyone has the same ones by next quarter. But a rebuilt workflow is yours. Nobody else has it, and nobody can buy it. The advantage doesn't go to whoever adds AI fastest. It goes to whoever redesigns around it first.

Most professionals know AI can do more for them. The gap isn’t awareness - it’s knowing where to start, what to change, and how to make it stick.

That’s what we work on through Work in Beta.

For individuals, we run working sessions, not teaching sessions. You bring a real problem from your actual work; we build the solution with you, live. You’re borrowing our learning curve instead of grinding through your own. You walk out with something that works and the muscle to keep going. When you get stuck later, we’re a message away.

You probably have a version of at least one of these:

Bring that. We’ll build it with you.

For organizations, AI adoption is a people problem, not a technology problem. Your teams have the tools, what’s missing is the translation layer between AI capability and daily work. Which processes to redesign, which habits to break, how to build genuine fluency, not just awareness. We help close that gap through hands-on training, process redesign, and deep adoption engagements. Not advisory, forward-deployed.

If any of this resonates, email us at [email protected] and we will figure out how to work together.