Searching for a job is frustrating. There are multiple job platforms, which means you may end up with a flood of emails from each with job listings based on your search criteria. Often these will display the same jobs. Also often, those jobs will be repeated in job listing emails day after day until the job is no longer available.

What if there was one platform that pulled in all the job listings from all these platforms? And what if you could use a simple swipe-right or swipe-left action to dismiss the listing or add it as a job of interest? Maybe then, once the job is dismissed, some smart tech could ensure you don’t see that job again in future listings?

That was an idea I pursued for about 5 minutes.

When pulling in data from other sites, the most common techniques are APIs (Application Programming Interfaces) or web scraping.

In the case of APIs, it relies on 1) the job listing site to have an accessible API that can be used by third parties, and 2) it being free, or affordable enough to make it a worthwhile endeavour.

In the case of job listings on LinkedIn, for instance, the API favours job postings, rather than job retrieval or job queries. This is for one simple reason - all the large platforms are slowly becoming walled gardens. They’re making it increasingly difficulty to use the web in the way it was first intended by Tim Berners-Lee. Instead of the open web, we now have a handful of massive tech companies competing for your eyeballs and advertising dollars. If you click on a link in these platforms, it will often open an in-app browser, sometimes with no option to open the link outside of the app in your own browser. These platforms wants to keep you in their walled garden because that’s how they make money. This is fair enough - they’re not a charity - but it clearly benefits them more than the user.

The other way of pulling data is web-scraping. The problem with this is it’s legally dubious, at best. It almost certainly breaks the terms and conditions of the platform, and if you attempt to commercialise that data in any way, you will probably end up with a cease and desist letter from a highly paid lawyer.

Unless, of course, you’re already a massive tech company and have essentially scraped the entire web to use for your AI models.

This is where we may have a future answer - AI, and specifically - agentic AI.

Agentic AI refers to artificial intelligence systems capable of perceiving their environment, making autonomous decisions, and taking actions without human intervention.

The idea is, with a simple text prompt, you can ask AI to perform computer-based tasks on your behalf while you go off and do something more interesting.

OpenAI ChatGPT has demonstrated this with ‘Operator’. Microsoft has introduced agents for Co-Pilot. Claude AI (Anthropic) has ‘computer use’, and Google has unveiled Project Mariner. Smaller players are also producing demos of similar advances every week.

Here’s how these demonstrations typically go:

Imagine you’re researching your next trip. With a simple prompt, you tell AI roughly where and when you want to go, your budget, the kind of things you like to do, etc,. It then autonomously opens a bunch of web browser windows and starts doing the searches for you, and pulls the results into a nicely summarised report with recommendations. You could event prompt it to go ahead and make all the bookings for you. I must have spent months of my life doing travel planning manually, so I can easily understand the appeal of handing this over to AI - an autonomous travel agent.

When you first see it, it’s creepy as hell. The biggest barrier to this taking off is clearly going to be trust. I’m certainly not yet ready to give control of my computer over to AI. I know a lot of early adopters are experimenting with it. I’m being a bit more cautious.

However, it’s coming, and this could solve the job search issue I described earlier.

I foresee a solution where, in the first instance, you hand over control of your email inbox to AI. Again, this is not the future, it’s already happening. You ask AI to summarise new job listings that match your criteria, AI then take those listings and add them to a list in another application, which could be simply in a notes app. You then review the listings and delete any you’re not interested in. The AI takes note of what you’ve deleted and ensures they’re omitted from any future summaries from your email inbox.

This is a fairly simple use case.

The thing about agentic AI is it gets around the web-scraping and AI issue. You can even take out email entirely, and just rely on searching the platforms directly. When prompting the AI, you can give it credentials to log in to LinkedIn as you. The same goes for Indeed, Monster or whatever is your job search platform of choice.

The question is, what will the platforms do to stop this? Is there anything they can do? If the platform owns the agentic AI solution you’re using, they’re unlikely to object. As for the others, it will be interesting to see what happens.

I know that there are many, like me, that are longing for the death of Google. The ‘don’t be evil’ mission statement was ditched a long time ago at the alter of pure greed. The result is that Google Search, as a product, is absolutely trash. I stopped using it a long time ago because of privacy concerns. When I look at it now, I don’t see the best search results at the top. In fact, you have to go ‘below the fold’ to get organic search results (results that aren’t paid for by advertisers). They have ruined their own market-leading product.

Just last week, the co-founder of Google - Sergey Brin, sent out an internal memo saying that AI engineers should work 60-hours per week to hit the productivity ‘sweet spot’. So engineers are expected to work long hours, to create a system that will potentially replace their own job, so that the giant corporation can maintain their huge margins. Capitalism at its finest!

Google was a leader in AI. They purchased the British company Deep Mind who was at the forefront of this stuff. Google Deep Mind is doing great things, yet Alphabet (the parent company), still makes most of its money from Google Search’s targeted advertising. Google still has massive market dominance in search. That currently remains at 90% of the global search market. Only a company with a such a monopoly would actively go out of their way to make their product worse for users.

Things are starting to change. For many of us, Google is no longer the first stop for searching the web. I now use AI. Sooner or later, everyone else will. Google is trying to counter this by integrating AI into Google Search, such as with AI Overviews. It may be too late. Once the masses understand the utility of AI, they will gravitate to AI-first search. That could be Google, but they’re no longer the only player in town.

It has been revealed in anti-trust lawsuits that Google paid Apple as much as $20 billion to be the default search engine on the iPhone. Yet, if you buy an iPhone now, Apple markets ‘Apple Intelligence’. The deal they have done to make up for the limitations in Siri is not with Google, it’s with OpenAI. They have said that they may support other AI providers later, but you have to wonder if Google has missed their opportunity to be the market leader in AI. I really hope so.

Google have already discovered in their experiments with AI Overviews that the summary is often good enough so that users will not click on a link to take them through to the source website. If you’re a company paying for a prominent link in Google, what’s the value in paying any longer if you’re not getting the click-through to your own site? It’s a problem for Google, and it’s a problem for advertisers.

This is where the big AI platforms may also become their own walled gardens. You currently only get live web search on a paid plan in OpenAI’s ChatGPT. Your search includes links to the source articles, but if all you are looking for is information, nine times out of 10 there may be no incentive to click and review the entire source article. This is a big problem for website owners everywhere.

Some websites are already going as far as adding known AI web crawlers to the robots.txt file on their website to block them. Not that it always works. Some of the major AI players have already been accused of ignoring the robots.txt file. The file is a convention for web technology that crawlers are supposed to abide by - it’s not a real preventative measure to actively block or allow web crawlers.

Besides the possibility of destroying humanity, AI is another advancement where it’s in the balance whether it will advance the open web, or further entrench the power of the tech oligarchs.

For anyone reading my post in the walled-garden that is Facebook, please, take a stand. There are other options.