Back in 2017, our agency began work with Fetch.ai, an organisation focused on combining Artificial Intelligence (AI) and blockchain.
As the chief technology officer would describe it “we want to create a world where millions of agents will live, interact and exchange value”. These autonomous agents would learn about our needs and then go out to ‘fetch’ the information or services we need, undertaking tasks on our behalf.
At the time it was a very bold vision. The world hadn’t yet seen ChatGPT and AI agents were still an academic endeavour. I must admit, Fetch was always a difficult concept to grasp, and it was a vision that required careful explanation to journalists.
When working with Fetch we intentionally targeted journalists we knew had deep knowledge and an interest in AI, ultimately securing some excellent outcomes e.g. Hal Hodson’s write-up for The Economist and several articles in TechCrunch. But back then, the vast majority simply weren’t curious enough (IMHO). Kudos to those that were.
Fast forward six years and it’s impossible to avoid agents. They’re everywhere. Having been shown the advances emerging from Open AI, Google DeepMind, Anthropic and others, journalists are (quite rightly) covering the subject from every possible angle.
I’ve always wondered why more tech journalists don’t go straight to the R&D labs to get the jump on major new themes. As former BT technology chief Peter Cochrane once explained to me, those in the R&D labs can see how technology will develop up to 20 years into the future with a high degree of confidence.
What’s going to be possible with agents?
Well literally as I write, PhocusWire’s newsletter has hit my inbox explaining that OpenAI has just tested ‘Operator’ an agent capable of using the web in the same way we do, moving the mouse, navigating to a site and booking travel. That’s pretty much how our Fetch spokespeople were saying it would play out.
OpenAI’s blog says: “Operator can ‘see’ (through screenshots) and ‘interact’ (using all the actions a mouse and keyboard allow) with a browser, enabling it to take action on the web without requiring custom API integrations. If it encounters challenges or makes mistakes, Operator can leverage its reasoning capabilities to self-correct.” That’s wild. Particularly so when you extrapolate it to other tasks.
Look to crypto for a glimpse of what’s to come
When it comes to crypto people get distracted by market fluctuations and bad actors. What do many miss? Thousands of talented developers are rapidly building with open source, fuelled by easy access to capital through tokenization. Culturally, this group are also intent on changing the world. It’s a recipe for innovation.
I’ve recently descended back down the crypto rabbit hole and there’s a lot to see. AI Agents are the hot narrative, deservedly so.
Using blockchains is still difficult. You need to initiate transactions from a wallet. These days you’ll need to bridge between blockchains too. If you’re interested in decentralised finance you may need to swap tokens on a decentralized exchange or move them around in search of yield generation. Even researching a new token requires skill. Trading it profitably? Extremely difficult.
Crypto itself (tokens) are really a natively digital settlement layer that’s faster, cheaper and arguably better than traditional payment rails. Smart people have long noticed these qualities make it ideal for machine-to-machine economic activity. With AI agents we’re starting to see this happen.
Today, right now, it’s possible to use agents to:
Using these agents to ‘do crypto’ is not difficult. It’s either natural language or drag and drop. When you step back and consider this for a moment, it truly is a glimpse into the capability of agents.
With a single query you can shortcut research on social media and transaction analysis on the blockchain itself – previously the job of a skilled researcher. You can devise a trading strategy and have the agent execute it for you, sending the tokens to your wallet – previously the job of a skilled trader. You can deposit money with an agent and it will make early-stage investments autonomously – previous the preserve of VCs. You get the idea, these tools automate difficult tasks, and they level the playing field.
If crypto has made this much progress you can be sure other industries will gradually get there too. The agents are coming…
Image: Steve Johnson / Unsplash