Back to insights

You can’t replace a job that never existed

Founder & CEO, Kydosa

We’re using AI with several clients and seeing 10x to 100x more knowledge work than their existing teams could previously cover. Yet they aren’t replacing any staff, and it isn’t a noble gesture from leaders. The reason is simpler than that.

With one client, we reviewed invoices from their facilities vendors. One invoice was for $2,500, far too small to earn a close look when attention goes to the big ones. It had passed the prior check, yet contained 5 discrepancies, including a $1,750 contract-rate error - 70% of its value! It wasn’t an outlier. Across hundreds of small invoices like it, similar findings added up to material savings and a firmer footing with vendors.

A deeper invoice review connects the bill to contracts, technical records, work logs and applicable rules.

The work nobody was doing

The big invoices got extra scrutiny. The small ones passed because the checks only reviewed whether they looked normal for their size, not whether they were right in the first place. Checking every invoice in detail would have needed a team no company could afford.

However, a “digital co-worker” was able to do the more detailed checks that would have crossed several departments in the company. For each invoice, it went back to the contract, matched technical reference numbers against specifications, reconciled work logs and related customer cases, and checked whether local laws applied. That’s roughly 50x the knowledge work, at less than half the cost of a person doing the standard review.

The client didn’t cut a single finance role. The team now acts on what the co-worker finds, correcting rates, holding vendors to their SLAs. The work AI took on had never been done by anyone.

You can’t replace a job that never existed.

When people talk about AI productivity, they usually picture today's work done with fewer people. However, what we see more often is work that was never done, now finally getting done.

We confused prioritization with focus

Why was it never done? We learned to filter.

I love two books that helped me grasp this. In The Case Against Reality, Donald Hoffman argues that evolution tuned our perception for fitness, not truth. Our senses simplify and our brains take shortcuts, because that’s what kept us alive. In The Creative Act, Rick Rubin describes creative work as tuning back into the raw signal those filters screen out.

The same is true for organizations. We pick the top 10 accounts for a deep-dive, audit only 5% of transactions, or average target customers into segments. We call it prioritization, but much of it was just rationing attention, because examining the full signal cost too much and took too long. Over the years those filters hardened into how we run the business, even as the data and the pace of change moved on.

Prioritization and focus aren’t the same thing. Focus is choosing what matters, whereas prioritization, too often, was deciding what we could afford to look at. AI widens the aperture. It doesn’t replace focus; choices still have to be made. Now we can look first and then choose, with much stronger evidence.

My friend Shubber Ali sums it up succinctly. He says “It’s not about making the impossible, possible. It’s about making the impractical, practical.”

Making the impractical, practical

Seeing risk before it's news. We have a manufacturing client that is widening its risk analysis to every product and its entire distribution network. The resulting risk terrain flagged a proposed law in one of its state markets, buried on page 16 of 67 of legislative minutes, that could change how its goods are sold. It wasn't in the news, in any of their feeds, or on their research team's radar. They’re building it into strategic planning instead of scrambling later.

A plan for every store. A CPG food client is aiming to be category captain at a national retailer (the supplier the retailer trusts to shape the whole category). Previously, the client would go through a time-consuming process to build a dozen plans for how the shelf should be laid out (planograms) and then apply them across hundreds of stores. Now with a digital co-worker, it will build one for each store, factoring in local demographics, weather and hyper-local trends. It expects around 5% higher category sales, 20% less spoilage, and deeper strategic investment from the retailer.

Building trust as co-workers

Trust doesn’t happen overnight. The category manager and the digital co-worker learn from each other. The category manager is starting by reviewing each planogram closely, correcting it and giving feedback. As the results hold up, the digital co-worker will take on a handful of stores, then a region, then hundreds. The category manager stays accountable for the results and will be able to cover every store individually instead of relying on a dozen templates.

Where to look now

An invoice checked against the contract, not just its size; a law spotted on page 16 before it made the news; a shelf plan for every store, not a dozen templates. None of this was impossible before. It just wasn’t practical, so we filtered it out.

So, with your attention less rationed, where would you choose to focus?