What should a leader, any leader, do in today’s AI-saturated landscape?
Have we reached the peak of the hype cycle, or are we just getting started?
Our AI Agent Map is an attempt to capture the current terrain. It’s not a blueprint or a trend report. It’s a snapshot of what’s actually happening, drawn from lived experience and active conversations in the field.
Let’s unpack what it’s telling us.
Are you leading an AI business?
“All anyone wants to talk about is AI. It’s not just Impress my boss, it’s Impress my board.”
That’s intentional in the map. A leader could be anyone from a Senior Engineer to a CEO. Everyone answers to someone, and the very tempting, cheap move in technology is to show your boss something they haven’t seen yet. There is so much excitement that it poses a danger for some leaders to overlook their core business. But AI is complicated, so enter Prompt Engineering. It’s very accessible and you can apply it directly to your problem area – it’s like you invented it. “Check out this cool, new way of Prompting – it saved me days of work”.
Some companies will build tools around prompts, but it’s for short-term gain. There is an inertia point – it doesn’t allow evolution, we need to think beyond the prompt. Ways of working are changing, so today’s cool prompt is tomorrow’s old trick. What happens when (not if) the Prompt is disrupted?
Prompt Engineering Is a Bridge, Not a Destination
“You don’t understand, Prompt Engineering is the future.”
Yes and no.
Prompt engineering is a vital skill in today’s world. But it’s not the endgame. The fundamental shift will occur when prompts fade into the background, absorbed by agents, embedded in tools, and abstracted away from users.
The current chat interface, while novel, is deeply flawed. Human-computer interaction needs a complete rethink. Agents will eventually replace chat as the default mode of interaction, doing the heavy lifting quietly, behind the scenes.
We will have access to the best model.
“Our AI expert has access to the best model around.”
Potentially, but for how long, and are we talking diminishing returns? OpenAI disrupted the industry when ChatGPT gained traction, but many major tech companies had developed similar large language models (LLMs) for years. There will be no model that wins out. They will become like CPUs. There will be a thriving community of specialists, but a small number of large companies will prevail; they will become commodities very quickly, even more quickly than CPUs.
Consider what you build on top of it.
With a basic understanding of Mapping, companies must understand their strength. You already have a team of highly experienced Data Scientists and AI experts, so a new model could make sense. But if you don’t, it’s an expensive game to get into.
The agent vs the API?
“We don’t need APIs, we will use agents.”
That’s not how this works.
Agentic AI is the next evolution. But agents still rely on knowledge, prompts, and APIs. Your APIs are how your core business capabilities are exposed. No APIs, no agents. It’s that simple.
There are two viable paths:
- Become an AI-native business (or join one, as this is not a new concept)
- Use AI to minimize friction for your existing users.
The latter is the more pragmatic and valuable approach for most. Automate internal workflows. Accelerate customer support. Improve personalisation. These use cases offer tangible value. Generating a Pixar-style short for the Monday morning stand-up? Not so much.
Machine-readable APIs are not optional.
“But everyone knows APIs are important!”
Yes, in theory. In practice? Rarely executed properly.
We’ve known how to design clean, atomic, machine-readable functions since the days of Unix. However, API documentation remains wildly inconsistent. Adoption of API-first thinking is patchy. And too many businesses still treat APIs as afterthoughts rather than core assets.
The most innovative companies today aren’t just playing with prompts. They’re the ones exposing robust, well-documented APIs that are usable by agents, developers, and platforms alike.
Final Thought: Don’t Just Chase the Shiny Object
AI isn’t magic. It’s a toolset. The map shows that we’re entering a phase of massive change—not just in capability, but in ways of working. Leaders must stay grounded, avoid the noise, and focus on building systems that evolve, adapt, and compound value over time.
Don’t just impress the boss. Build something that lasts to change the market.
Serverless Craic from The Serverless Edge
Check out our book, The Value Flywheel Effect
Follow us on X @ServerlessEdge
Follow us on LinkedIn