Signed and Dated: Why New Technology Gets Bolted Onto the Old Way of Working

The commonest digital transformation mistake is bolting new technology onto the old way of working. A paper form becomes a PDF that still says "signed and dated". The ritual survives the paper. AI is now being added the same way, speeding up steps that should not exist, which is why many projects disappoint.

What does "signed and dated" mean in digital transformation?

"Signed and dated" is the instruction printed on paper forms. When a business digitises, the form often becomes a PDF or online form that still asks for a signature and date, even when the system already records who submitted it and when. The technology changed. The ritual it was built to serve did not.

Once you look, the pattern is everywhere. Leave requests move from paper to email, yet still need three people to forward them. A report goes from printed to shared drive, yet still gets compiled by hand every Monday. Forms get scanned to PDF, printed, signed and scanned again.

None of this is foolish. Each step once had a reason. The signature proved who agreed, and the date proved when. The new system can prove both on its own, but nobody went back and asked whether the step was still needed.

That is the real mistake in most transformations. The technology works. The process around it was never redesigned.

Why does new technology get bolted onto the old way?

Because it slots straight in. Using new technology the old way needs no one to rethink the work, change roles, or retrain staff. The old process is familiar, someone owns each step, and changing it feels risky. Redesign also needs imagination about how the work could run, which is harder than adding a tool.

Ownership is part of it. A tool has an owner: whoever bought it. A process that crosses three teams often has no owner at all. So the tool gets fitted to the process, because nobody has the authority to change the process to fit the tool.

Existing investment matters too. Businesses have spent years building their current way of working, along with the habits and job descriptions around it. Replacing a step feels like throwing that away, even when the step no longer earns its keep.

What did electricity teach about bolted-on technology?

Electricity showed that a powerful technology can disappoint for decades when it is fitted into old layouts. Factories first swapped steam engines for electric motors and kept everything else the same. Productivity only jumped when factories were rebuilt around what electricity made possible: smaller motors, new layouts and new ways of working.

The economist Tim Harford tells the story for the BBC. By 1900, he writes, "less than 5% of mechanical drive power in American factories was coming from electric motors". Owners who swapped in a motor were "often disappointed with the savings". Yet electricity let power go wherever it was needed, so factories could follow the logic of a production line rather than a drive shaft.

According to Harford, "you couldn't get these results simply by ripping out the steam engine and replacing it with an electric motor. You needed to change everything: the architecture and the production process." When manufacturers finally did, in the 1920s, productivity in American manufacturing "soared".

Computers repeated the pattern. Harford notes research by Erik Brynjolfsson and Lorin Hitt showing the difference was "whether the companies had been willing to reorganise".

How is AI being bolted onto old processes now?

AI is being used to do the old steps faster rather than asking whether the steps should exist. Teams use it to write reports nobody reads, summarise meetings that did not need to happen, and draft long documents a short one would replace. The work speeds up, but the outcome barely moves.

The last row is the quiet one. Giving everyone an assistant and changing nothing else is the purest form of bolting on. Each person rebuilds the business's context in their own chats, and none of it is shared. The guide to giving everyone ChatGPT explains why that stalls.

The evidence points to redesign. According to BCG's 2025 AI at Work survey, about half of companies are moving beyond productivity plays to redesign workflows. Employees at companies in comprehensive redesign also worry more about job security, at 46% against 34%. Redesign is harder, and people feel it.

How can you tell if your AI project is bolted on?

Ask what changes in the outcome, not the effort. If the same steps run in the same order with AI doing some of them faster, it is bolted on. A redesigned workflow drops steps, changes who does what, or produces a different result. Five questions will usually show which you have.

1. Why does each step exist? If nobody knows, it may be a ritual.
2. Would we design it this way today? Starting fresh often removes half the steps.
3. Does the outcome change, or only the speed? Faster rituals are still rituals.
4. Does the AI know the business? Without shared knowledge, it can only copy the old way.
5. Who owns the whole process? Without an owner, nobody can redesign it.

Research on failed projects points the same way. RAND's 2024 study lists chasing the technology rather than the real problem among the five root causes of AI project failure. The guide to why AI projects fail covers the rest.

How do you redesign the work instead?

Start from the outcome the work exists to produce, then rebuild the steps around what AI and your systems can now do. Keep the steps that genuinely protect the business, such as a named person approving what goes out. Drop the ones that only made sense on paper. Do it one workflow at a time.

The approval step is worth singling out. Not every ritual is empty. A person approving anything sent, published or spent is a real control, and it should survive any redesign. The difference is that it now sits at a deliberate point, with a named owner, rather than a box on a form.

Redesign also needs the AI to know the business. MIT NANDA's 2025 study blamed most failed AI projects on systems that do not retain feedback, adapt to context or improve over time. A redesigned workflow gives AI the facts and methods it needs, then keeps them current.

Skein builds a Company OS, one shared system underneath everyone's AI, starting with the workflow that pays back fastest, seen running before more is committed; the Skein home page shows how the layers fit. That suits a growing B2B business whose staff already use AI but whose processes have not changed. A business that wants AI added to its current process without changing anything may find a simple tool rollout enough.

Frequently asked questions

What is the most common digital transformation mistake?

The most common mistake is fitting new technology into the old process without redesigning the work. Paper forms become PDFs that still ask for a signature and date, and AI gets used to speed up steps that no longer need to exist. The technology works, but the outcome barely changes.

Why do digital transformations fail?

Many fail because they digitise the existing process instead of redesigning it. Nobody owns the end-to-end process, the old way is familiar, and redesign needs imagination and authority. Technology bolted onto old steps delivers small gains, while the larger gains come from changing how the work runs.

What is the productivity paradox?

The productivity paradox describes powerful new technologies that fail to raise productivity for years. Factories using electric motors and businesses using computers both saw slow gains at first. In both cases the big improvements came once organisations reorganised their work around the new technology, rather than just adding it.

How should AI change business processes?

AI should prompt a redesign, not just speed up existing steps. Start from the outcome a process exists to deliver, ask why each step exists, drop the rituals, and keep real controls such as human approval. Give the AI shared business knowledge so it can do the work properly.

How do you know if AI is just automating a bad process?

Check whether the outcome changes or only the speed. If the same steps run in the same order with AI doing some of them faster, the process has not been redesigned. Ask why each step exists and whether you would design it the same way today.

Related reading