AI doesn’t sit still. In fact, it barely walks. It sprints. And now that OpenAI’s Deep Research is officially available in the UK as of Friday 7 February, the potential for serious business use is picking up speed.
We’ve been exploring this new feature at Blue Llama and ran a real-world test using mortgage providers here in Jersey. The results were genuinely impressive, with a few expected bumps along the way.
Let’s break it down.
First things first. What is Deep Research?
Most people are familiar with AI assistants like ChatGPT that respond to prompts in a single chat window.
Deep Research works differently. It behaves more like an autonomous AI agent. You give it a broader challenge, and it sets off to solve it. It searches the internet, checks sources, follows links and even changes direction based on what it finds. Think of it like hiring a virtual research assistant who doesn’t need constant instructions.
In short, it doesn’t just answer. It investigates.
What we asked it to do
To put Deep Research to the test, we gave it this task:
Find all mortgage providers in Jersey, Channel Islands, and collect their available mortgage products in a spreadsheet. Include: provider name, product name, type (fixed, tracker, etc), interest rate, loan-to-value ratio, fees, a short description and the date collected.
It’s a big ask. Especially since much of Jersey’s mortgage data is tucked away in PDFs or hosted on clunky websites.
How it performed
It found the right sources
Deep Research quickly identified mortgage providers including HSBC, Lloyds, Skipton International, Santander, Barclays and Butterfield. It accessed their websites and third-party platforms to collect data.
It understood the task structure
Without needing detailed formatting guidance, it grouped the data sensibly: provider, product name, type, interest rate, LTV, fees and notes.
It retrieved a lot of data
On the second run, it pulled back 86 mortgage products from eight different banks in around eight minutes. If you’ve ever tried doing this manually, you’ll know that’s fast.
It identified gaps and tried again
When we noticed some product gaps and inconsistent naming, we asked it to refine its results. It handled the feedback well and delivered a cleaner, more complete dataset.
Bonus
It even flagged mortgage types like “tracker saver” and “fixed saver,” giving us insight into naming patterns across banks.
Where it struggled (as expected)
Spreadsheet creation was hit and miss
Despite the prompt, it didn’t always generate a clean, downloadable spreadsheet. Some manual formatting was needed, though it handled CSVs fairly well.
Date stamps were inconsistent
It listed the collection date as March 2025 in some entries, likely pulled from product PDFs referencing future availability.
Incomplete product ranges
Some banks only displayed limited product types, and other details like phone numbers or email contacts were missed entirely.
Still, considering this was gathered autonomously, it’s an impressive result. Especially for a feature still in preview.
Cost and access
At the moment, Deep Research is available only on the ChatGPT Pro Plan at $200 per month and is limited to 100 uses monthly. That might sound restrictive, but considering each run could replace hours of human work, it quickly adds up in value.
OpenAI has indicated plans to roll this out to the Plus Plan at $20 per month in the near future. Possibly even to free tiers eventually.
Where this could go next
From market research and product comparisons to competitor tracking and lead generation, the possibilities are endless. Imagine being able to:
- Automatically compare insurance or pension products
- Monitor regulated markets for price changes
- Analyse hundreds of suppliers to find the right match
And this is just the beginning.
OpenAI CEO Sam Altman recently said a small percentage of the world’s work output could already be handled by AI. That doesn’t mean job losses. It means efficiency. Businesses that embrace this now will move faster and smarter.
So should you be using Deep Research?
If your team relies on repetitive online research, data collection or market comparisons, then yes. It’s absolutely worth exploring.
Even in its early form, Deep Research shows how far AI has come. And it’s only getting better.
Feeling unsure where to begin? That’s completely normal. We’re happy to help you explore the possibilities and identify where this tech could make a meaningful difference in your organisation.
And if you’re curious about what’s coming next, keep your eyes open. The next AI release might just raise the bar again.
