I've read a lot, and written a little, about how AI Vibe Coding will impact low-code platforms like the Power Platform. It's quite clear where Microsoft wants it to go, as someone said "they want natural language to be the language of the platform". Microsoft are all in on this, because a natural language UI future:
- Is a world where they could lose their legacy dominance
- They can profit from the infrastructure needed
- It moves to the new cash flow model- PAYG
- It "in theory" opens up new users
The first one is the big one for them, they missed the mobile revolution and now they are full FOMO (fear Of Missing Out). At least mobile had limited impact on enterprises, but AI could create the first real break that encourages organisations to invest in a big infra move.
Nvidia is a big winner, as in a gold rush don't dig for gold, sell shovels. And the infrastructure to use all those shovels/graphic cards are large data centers, and Microsoft with Azure has that. So they are in a good position to have the economies of scale to be cheaper and profit from competition (worked well for AWS).
When everyone moved to the cloud companies like Adobe figured out very quickly the peaks and troughs of sales through direct purchases could be replaced with steady dependable and often sticky cash flow with subscriptions. It feels that Microsoft has drained that cash cow now, there are no new licenses to sell, additionally they are no easy way to track ROI, making keeping existing licenses harder. Pay As You Go could be the answer, linking a ROI to direct consumption, now organisations are incentivised to spend more, as every dollar spent with Microsoft is 2 dollars saved.
And the final one is that there is a whole new market. Currently Power Platform makers have to have a combination of skills and drive to create solutions. Now with AI abstracting away the skills more people can make solutions. I think this one is the weakest, because I would expect for every new user you will lose (or at least decrease the output) of a current maker.
With this in mind we should always be taking what Microsoft says with a small pinch of salt, they need their AI offerings to be your future, even if they (in my opinion) don't quite live up to the hype.
So we know Microsoft is going all in, so how will that impact the Power Platform, and what will it look like in a years time.
Years Time
This one is controversial, but I suspect in a years time very little will have changed. There will many (many) new AI features, and although these will be talked about a lot, they will not have trickled down to large organisations.
Large orgs are like oil tankers, they take miles to turn. All the new AI features will need to go through legal, governance, and most importantly security. Smaller agile companies will risk it, but large orgs with lots to lose wont.
I would also expect a little swing back, more investment back in core features like PPAC and less forced on AI features (it's amazing how the threat from large paying customers can have on Microsoft).
Don't forget in most cases these AI features are not solving a problem, they are improvements, and although organisations want that productivity boost to help their bottoms lines, they are happy to let someone else be the first adopter and take the risks.
After Next Year
The truth is AI is not going away, it is the future, so the way we work will change.
And this is where I think there are 2 possible paths we could end up down.
Path One - Microsoft's Happy Path
The AI tools start delivering what was promised, the following key things happen:
- Vibe coding delivers robust solutions
- LLM token costs drop significantly
- Users switch from visual to text base UI
- Security gets ahead in the arms race
Now Power Apps looks very different, Canvas Apps are still here, but now you start with a prompt, and after a few rounds of prompting the app is available to be updated in the UI. That's right, prompting will get you so far very quickly but th 80/20 rule strikes, and the last 20% is so much easier to do in the UI with a simple click/drag/type (My favourite example of this is car production lines, robots will do most of the work but there will always be a human on the production line to do the fiddly bits).
Flows will leverage the simplicity of prompts to create automations, but then gain the core requirement of deterministic outcomes. If you put a and b in, you always get c out.
Copilot Studio will be just like Power Apps, vibe coded. So you start with prompts and it creates all of the topics and connections. Again the UI will come at the end so you can do the last 20% targeted.
Power Pages is the interesting one, as its in a very difficult situation with Power Apps taking on React. So my guess would be that they unify Power Apps and Power Pages (going full circle back to Power Apps Portal). The developer builds the same and then at point of deployment decides if its external or internal.
Dataverse won't change much, its benefits are not just UI deep so I can see still being the AI database of choice.
Dataverse Low-Code plugins will disappear, and I vibe coded version of standard plugins will be created (Im kind excited by this, as the terrible buggy nature of Low-Code plugins ruined the opportunity).
Although AI has taken over the platform, there are still a plethora of legacy apps and flows, all of which continue to run, generating value, (as if its not broken why fix it).
Path Two - Microsoft's Unhappy Path
As you saw in the happy path, ther 4 key requirements are not easy. So I could quite easily see
- Vibe coding still delivers spaghetti and unreadable solutions
- The continued investment in larger LLM models means token costs staginate/rise
- Users push back and want to stay visual
- Prompt injections are not beaten and new vulnerabilities are discovered
In this world organisations simply don't pay for the AI features. They go the other way, forcing them off. The current crop of Low-Code tools deliver all the value they need, and security win the PR war with the senior leaders.
Microsoft is stuck as its invested heavily and needs to double down or pivot. They pivot, splitting all the AI features out into their own area.
Power Apps keeps the visual UI with a little bit of AI on the side. Special AI components and/or a way to use AI to create PCF components. Deep down I suspect this will happen even on the happy path, as vibe coding as a premise isn't right for organisations. They need understanding and stability, nothing of scale has ever been made vibe coding (well maybe fly.peter.com is an exception lol). The real value is AI assisted development, where the developer already has the skills and supervisors the AI, AI a tool not a delegation.
Power Automate doesn't change, the sprinkle of AI stays in certain connectors like now. The truth is Power Automate built on the very established Logic Apps, for me it's kind of at peak evolution (especially when inline UI launched properly, hurry up Microsoft).
Copilot Studio leaves the Power Platform going on to join M365 Copilot and other Agents that are designed to compete with the og Office suite.
Power Pages becomes the vibe coded part of the platform. It takes on the React code instead of Canvas Apps, but the code is visible and editable. It becomes more of the Low-Code to Pro-Code bridge, not used by citizen developers, just accelerating pro devs.
Dataverse starts appearing more and more outside of the Power Platform.
Dataverse Low-Code plugins start to work 😎
The current tools all get a blast of investment, all those features put on hold suddenly start being delivered.
Microsoft is also ironically in a good position, as even though they invested heavily in AI, so did their competition. And they are able to fallback om existing venue streams to steady the ship.
The Take Away
Although AI is the future, the future is not tomorrow, and no one knows exactly what the future looks like. It could be everything, or it could be a niche, it could be the new Smartphone or the new Blockchain.
Either way the Power Platform will still be here, bigger and better, and for me that's super exciting 😎
Top comments (1)
The components of power platform are someone else code. the beauti of the space is how creative you can be with your problem solving space but still work with the constraints or limitation of what drag and drop components are available.
I am excited to see how the whole next year unpacks.