My father sold kitchens for years. When a customer signed, his whole job was to walk downstairs, hand the paper to a colleague, and move on. A few days later the kitchen was at their door.
After a few rounds of restructuring and digitization, the same sale meant screens, forms, fields, handoffs between departments, and a delivery that took longer than it used to. More software. More steps. More work for everyone involved. The customer waited longer to get the same cabinet.
That story stuck with me because it breaks the promise we were sold about modern work. Computers were supposed to make things faster. Instead, a lot of people spend their days feeding systems so data can move.
How we built the slowdown
Over the last few decades, most companies went through the same arc. Restructure. Centralize. Digitize. Standardize. Each step made sense in a slide deck. Fewer errors. Better visibility. One source of truth.
The cost showed up somewhere else. Every handoff needed a format. Every department needed its fields filled. Every process needed compliance before the next person could act. The sale still happened in a conversation. But the data had to travel through a maze before anything moved.
People stopped being the ones who made things happen and became the ones who made information flow. Copy this into the CRM. Attach the signed PDF. Open the ticket. Update the status. Wait for the system to catch up.
What we lost along the way
Speed, for one. The kitchen story is not an edge case. How many simple outcomes in your own work now require five tools and a checklist that did not exist ten years ago?
Flexibility, for another. When everything runs through a fixed process, the process becomes the product. Does it really matter that every customer fills the same form if they all want to buy the same thing? Or that every answer lives in the same field if the useful detail is scattered across an email, a call note, and a message someone sent at 11pm?
And retrieval. Need to find what was agreed, what changed, or who signed off? Good luck. The information exists. It is just locked behind the structure we built to organize it.
What AI changes about the structure
Here is the shift worth paying attention to. For years, we designed around the idea that humans must normalize data before a system can use it. Same fields. Same format. Same pipeline. That made sense when only structured data could move between tools.
AI is good at a different job. Pull the useful detail from an email, a PDF, a Slack thread, a handwritten note on a scanned form. Merge it. Summarize it. Answer a question across sources that were never designed to talk to each other.
That does not mean ripping out every process overnight. It means asking a question we stopped asking: does this step exist because the work requires it, or because our tools required us to shape the work a certain way?
Maybe the customer form matters less than the assistant that can read what they already sent you. Maybe the status update matters less than the thread that explains why the status changed. Maybe the real work is the conversation, and the paperwork was always just a tax we accepted because we had no other way to move information.
We are building Zenpa around that bet. Not another system to fill in. A private assistant on your Mac that reads the tools you already use, connects the fragments, and gives you answers without forcing your work into one rigid shape first.