It is said that any system that rely solely on manual entry such as manual metadata usually fails over time. Under heavy workloads, humans skip fields, use inconsistent terminology, or neglect tagging entirely, making search functions nearly useless.

Typing endless variations of titles, descriptions and keywords is deeply tedious. Typos, inconsistent naming conventions, and duplicated efforts quickly degrade any possibiity of finding an image in files of millions.

Looking back, it is difficult to pinpoint exactly when the frustration really started to build up.

There was no dramatic moment.

No single upload.

No particular photograph.

Instead, it crept in slowly over time.

At first, metadata felt like just another part of the job and the buzz in the early days was seeing the work accepted by the assessors and added to our portfolios

Photographs were edited. That in itself is enough work. But then what?

Titles were written and entered on the keyboard into the right box. That’s too long, that’s not catchy enough, thats too quirky, that doesnt demand attention, thats not descriptive enough.

Descriptions were added on the keyword into the right box. That’s badly written, that’s keyword stuffed, that is repeating the title, that sounds like keyword spam, that’s not long enough, that’s too long, that doesn’t really describe the image – that got rejected last time.

Keywords were created. struggling to find 50, scraping the barrell, trying to prioritise the most important. Too conceptual, no concepts, too spammy, not relevant enough. Boring, boring boring. That will do, well if its rejected its rejected.

Images were uploaded with a sinking feeling of how long the metadata processing will take.

The process was accepted simply because it was the way things had always been done.

For years, we worked that way.

And for years, it seemed perfectly reasonable.

The problem was that photography never stands still.

As our portfolio grew, so did the workload.

One shoot became ten.

Ten became fifty.

Hundreds of photographs became thousands.

The photography itself remained exciting.

The metadata did not.

The more photographs we produced, the harder it became to give every image the individual attention it deserved.

A title would be written.

A description would be created.

A collection of keywords would be assembled.

Then, more often than not, that metadata would be copied across the rest of the batch with a few adjustments along the way.

The photographs were similar but they were essentially different.

The metadata was rarely different.

At first, that felt efficient, almost a smart way to tackle the issue

Practical. Yes

Necessary. Yes

The job got done.

The images were uploaded.

The agencies accepted them ( usually – not always) .

Everything appeared to be working.

There was no real mystery about what was happening.

But we knew. Deep down we knew.

We knew the metadata was not reflecting every photograph as accurately as it could.

We knew the tiny details that made one image different from another were often being lost.

We knew some photographs were receiving the same descriptions, the same keywords and the same titles despite telling very different individual stories.

The problem was not recognising it.

The problem was having the time, energy and motivation to do something about it.

The joey emerging from a pouch.

The nervous look, the curious glance.

The subtle behaviour.

The moment captured that had lasted only a fraction of a second.

Each photograph told its own story.

But increasingly, those stories were being squeezed into the same standard descriptions, the same keywords and the same titles.

And there was a consequence to that.

Great photographs did get uploaded.

Great photographs did get accepted despite the generic metadata ( usually, not always) .

But that did not necessarily mean they ever got seen.

Some of our favourite images disappeared into vast libraries containing thousands, sometimes millions, of photographs of the same subjects.

Somewhere on page thirty of a hundred pages of search results sat photographs that deserved far better visibility than they were receiving.

Not because the photography was weak.

Because the metadata was no stronger than the hundreds of images surrounding them.

Good enough metadata had helped the photographs get online.

It had not necessarily helped people find them.

The trouble with good enough is that it rarely stays good enough forever.

As the years passed, our patience, motivation and self discipline began to wear thin.

Backlogs started to grow.

Photographs sat waiting to be uploaded and tagged.

Not because we did not value them.

Not because we had lost interest in photography.

But because we knew what was waiting for us at the end of the process.

Metadata.

The same repetitive task.

Again.

And again.

And again.

Sometimes a shoot would sit untouched for weeks.

Sometimes months.

The photographs had already been taken.

The editing had already been completed.

Yet the final hurdle still felt enormous.

When uploads eventually happened, there was often a sense of relief rather than satisfaction.

The photographs were online.

The task was complete.

But deep down we knew there was a difference between finishing a job and doing a job well.

The photographs deserved better.

We had known it for years.

The problem was that better required time we did not have.

Time was the one thing we never seem to have enough of.

Around a year ago, that growing frustration finally led to a simple question.

Like many people we had started early experimenting with ChatGPT.

Mostly out of curiosity.

It was new, exciting.

Interesting, thought provoking.

Occasionally surprising.

Then one day, the conversation drifted towards photography.

More specifically, towards the part of photography we enjoyed the least.

Metadata.

What if there was a better way?

The rise of AI-augmented admin work is revolutionizing professional life. By taking over routine tasks, AI allows employees to focus on strategic and creative efforts. This redistribution of responsibilities not only makes work more stimulating but also leads to unexpected opportunities for growth and innovation.

WRENCH AI

What if every image could receive the individual attention it deserved without hours of repetitive writing, copying and pasting?

What if great photographs no longer had to disappear into page thirty of a hundred pages of search results?

What if good enough no longer had to be the compromise?

At the time, it was nothing more than a conversation.

A passing thought.

A question asked out of frustration.

We had no idea that question would eventually become a project of its own.

We simply knew that after years of accepting good enough, we had started looking for something better.

“AI does not replace human intelligence. It reveals how much of it we were wasting on tasks beneath our potential.”

Bernard Marr (Author and Futurist):

---