How to spot the patentable feature in your own work

A practical method

Start from the problem, not the product. Write down the technical problem in one sentence, with a number in it if you can. Then write down what a competent engineer in your field would have done about it, using what was already known.

Now find the step where you did something different. That difference is the candidate invention. Everything around it is implementation.

A useful test: if you removed that step and used the conventional approach instead, would the system still meet its specification? If yes, that step is not the invention. If no, it probably is.

What engineers systematically undervalue

The unglamorous fix. Teams pitch the architecture and forget the two-line change to the control loop that made it stable - which is frequently the only genuinely inventive part.

They also undervalue things that were hard to find but look obvious in hindsight. A threshold, a timing relationship, a sequence of steps, a material choice that nobody would have predicted. Once you know the answer it looks trivial; before you knew it, nobody had it.

What to actually do

• Write the technical problem with a number in it.
• Find the step that departs from what the field would have done.
• Test it: does the system still work if you substitute the conventional approach?
• Do not discard a difference because it looks obvious now.