SkillCurio Dispatch · Issue #5
The distance between an idea and doing something with it is shrinking.
AI is making it easier to build, automate, and experiment. Here’s where your existing skills give you a head start.
Get the next Dispatch · The complete issue, every two weeks.
Remember when building a useful application meant finding a developer? Or automating a business process meant writing scripts, getting approvals, and spending weeks connecting systems?
Those barriers haven't disappeared.
But AI is starting to knock some serious holes in them.
And the exciting part isn't just what developers can accomplish.
It's what people who understand a problem can now begin doing about it.
This week's developments show AI moving beyond generating answers toward helping people build solutions, test ideas, and make things happen.
You don't need to master every new AI tool.
You need to recognize where one could help you accomplish something you couldn't easily do before.
What changed
1. GitHub is letting AI choose the AI
Choosing the right AI model can feel like shopping for a new computer every time you want to get something done.
GitHub wants to remove some of that friction.
Its newly released Copilot Fusion can select and coordinate different AI models depending on the coding task.
For straightforward work, it might use one model. For something more difficult, it can use one model to produce an answer and another to critique it.
GitHub reports that Fusion Max achieved competitive coding benchmark results while using roughly half the API cost of a leading comparison system. That's a benchmark result, not a guarantee for every project. Fusion is available across Copilot plans, with usage-based model charges.
Here's why this matters beyond software development.
We're beginning to see AI systems handle more of the complexity involved in using other AI systems.
You focus on the outcome. The software increasingly figures out how to get there.
For developers, that could mean less time choosing models and more time building.
For aspiring builders, it's another glimpse of a future where technical complexity becomes less of an obstacle to trying something new.
Explore GitHub Copilot Fusion →
2. The next automation expert might already work in your department
Enterprise automation company Serval just raised $150 million to expand its AI-powered workflow platform.
The funding is interesting.
But the company's approach to automation is much more interesting.
Serval's Catalyst system allows people to describe business processes and have AI help turn those requirements into executable workflows.
Think employee onboarding, software access requests, equipment approvals, or routine IT support.
These are processes that experienced employees understand intimately.
And Serval is betting that those employees can increasingly help build the solutions themselves.
The company also introduced Simulations, which lets IT teams test how an AI help desk agent handles different scenarios before allowing it to interact with live systems.
Imagine testing how an AI agent processes a laptop request from a contractor versus a full-time employee.
You could discover that it sends the request to the wrong approver before anybody receives an incorrect approval request.
That is a practical use of AI.
And it highlights an opportunity for experienced professionals.
You already understand how work gets done. AI can help you explore better ways of doing it.
The valuable skill isn't necessarily writing every line of automation code.
It's understanding the workflow, spotting the exceptions, and knowing whether the outcome is correct.
Read about Serval's automation approach →
See how Serval tests AI agents before deployment →
3. Google makes AI-generated content easier to identify
AI-generated images, video, and audio are becoming remarkably convincing.
That's exciting for creators.
It's also making digital verification more important.
Google has expanded access to its SynthID Detector, allowing people worldwide to check supported images, video, and audio for invisible AI-generated watermarks.
The tool can identify supported watermarked content from Google and participating partners, including OpenAI and NVIDIA.
This isn't a universal fake-content detector. A file without a detectable watermark isn't necessarily human-made or authentic.
But it gives more people access to a practical verification tool.
And that matters for professionals handling digital information.
Researchers, educators, journalists, marketers, and business teams increasingly need to determine where material originated before using it.
Explore Google's SynthID announcement →
Your next learning step
Find one thing you could automate
Time: 30 minutes. No coding required.
Think about something you do repeatedly.
Maybe you're copying information between documents, categorizing requests, preparing a weekly summary, or reviewing the same information every morning.
Now ask yourself:
What would happen if I didn't have to perform every step manually?
Start with a simple workflow:
Something happens → Information is collected → A decision is made → An action follows → You check the result
For example, imagine a system that receives a support request, checks its category, and prepares a suggested response.
You still review the response before sending it.
But the repetitive preparation is handled for you.
Write down how that process would work.
Identify the information required, which steps could be automated, and which decisions should remain yours.
Use a hypothetical example or sample data rather than confidential workplace information.
Then, if you're interested, use the n8n course above to explore how you might build it.
Your goal isn't to automate your entire job.
It's to identify one task where AI or automation could give you back time.
That's a meaningful first step.
Future Radar
Google DeepMind's Genie 4: What if AI could build a world you could explore?
Here's something worth getting excited about.
Google DeepMind unveiled Genie 4, a new AI system that creates interactive simulated environments.
Instead of generating a static image or prerecorded video, Genie 4 can simulate an environment that responds as someone explores it.
It can generate movement, environmental sounds, and interactive dialogue.
And users can influence the simulation using natural-language instructions.
The technology remains in a limited research preview, and Google acknowledges that physical accuracy and long-term consistency still need improvement.
But consider the possibilities.
A technician could eventually practice diagnosing equipment problems in a simulated environment.
A student might explore a historical setting.
A safety team could rehearse a complicated scenario.
A designer could experiment with an interactive environment before anything gets physically built.
These are possibilities, not established Genie 4 applications.
Still, they suggest an exciting direction.
AI may eventually make experiential learning much more accessible.
Instead of only reading about a problem, you might be able to practice solving it.
The question worth asking: If you could create a realistic training environment for something you do professionally, what would you want to practice?
Explore Google DeepMind's Genie 4 announcement and demonstrations →
One thing to remember
The most exciting AI developments aren't necessarily the ones announcing another powerful model.
They're the ones making it easier for people to do something useful.
Build a prototype.
Improve a process.
Test an idea.
Learn a skill.
Solve a problem that's been bothering you for years.
You already bring something valuable to that equation: your experience and your understanding of what needs to be done.
AI can help you act on it.
You don't have to know how to build everything.
Start by learning how to build one thing.
Then see where that takes you.
SkillCurio Dispatch #5 · November 5, 2026. This web edition preserves the approved newsletter text and source links.