By now, if you have suffered through the relentless onslaught of our LinkedIn content, you know we do things in Generative AI.

We’re essentially an artificial intelligence layer that sits on top of your data belonging to Crewing, Chartering, and Commercial Teams.

Once we integrate with your data, we use AI to turn it into actionable intelligence, helping you make intelligence-led decisions.

But those same AI capabilities solved another problem for one of our clients.

Introducing Ocean Oracle

So what have we done?

We’ve enabled teams talk to complex technical manuals running into hundreds of pages with Ocean Oracle.

The challenge of data overwhelm does not exist just in your database. It exists in your crew. No matter how much time you put into writing your Safety Management System and Stand Operating Procedures, it is for nothing if your team doesn’t read the manual.

And it is easy to understand why they wouldn’t.

Manuals are hundreds of pages long and increase in length and complexity yearly. It is unrealistic to expect a person to study the entire manual, particularly when updated frequently. It's a huge problem in business with complexity around machinery and operations - not just maritime. Enabling your team members to access the right information, promptly, to operate your business correctly is hard to do when you rely on them to open up a 700-page document on their computer and read it.

If you know someone struggling to get team members to engage with lengthy manuals with in-depth schematics and diagrams, Ocean Oracle may be the right solution.

A customer asked us to solve a problem whereby they had complex cargo and machinery operations manuals, and they wanted a better way to read them. They mentioned the challenge engineering and deck officers faced to stay current on the latest procedures. This was even more challenging if they worked across multiple vessel types. Of course, we were happy to help.

From our perspective, there must have been an off-the-shelf generative AI solution that we could plug into to deliver this. It seemed like the AI components required to consume complex diagrams would be out there. We could find one, plug it in, and bingo, we’d solve the problem.

We expected something 20 to 30 pages long.

24th November 2023 - The Manuals Arrive

When multiple manuals, running 500-800 pages long, turned up, I was slightly embarrassed and expected that we would not be able to help.

After all, the manuals were dense with text and interlaced with different diagrams and schematics. To navigate the document, you might have to reference 10-15 pages dispersed throughout the document to reach a definitive answer.

After the engineering team saw it, whilst surprised, I think more than anything, they were excited about the challenge. They outlined how several hard problems needed to be solved. Not just problems for Greywing but problems that were yet to be solved in Generative AI more widely. But, they sensed that it was most likely something they could deliver. I was surprised; I was on the cusp of telling the customer we might have over-promised on this one. But instead, I let them know we would put something together.

5th December 2023 - POC Released

Within a few days, we had something compelling working; you may remember the original demo we ran through here, where we demoed the capability without a front end. We sat down on a call with our customer and ran through it. The core value was there, and we agreed to take the next step. Add a front end and move it into production.

12th January 2024 - Delivery

49 days later (including a week off for Christmas and New Year's), we delivered Ocean Oracle and, with it, the ability to converse with technical cargo and machinery operations manuals.

But that wasn’t all that had happened.

450,000 views on Twitter

Unbeknownst to them, the Greywing engineering team were opening up a new front within Generative AI. Hrishi's post demonstrating the break through he was working on for Ocean Oracle had garnered over 450,000 views on Twitter. We received a steady stream of other AI companies and companies facing the same problems of having data trapped in complex documentation. Data that their teams couldn’t access efficiently.

The response from outside of the maritime industry was been overwhelming.

I think this is something that will happen more and more as innovation continues to emerge in the maritime industry and as the leading vessel operators compete with one another to remain on the cutting edge of innovation, the maritime industry will begin to define what is the cutting edge of technology, not only for maritime but also for adjacent industries operating with similar levels of complexity.

If we had not faced the unique challenges found in the maritime industry, a core generative AI technological breakthrough with applications across petrochemicals, Oil & Gas, medical devices, automotive, defence, and hardware industries might not have been built.

A Flight Operations Director has already contacted us. They want to use the same technology to keep their pilots up-to-date with complex manuals.

Whoever thought that technology, built for maritime, might once pave the way for aerospace?

Want to stay updated on all things AI we're building for Maritime? Sign up to our monthly newsletter here.