Project GROOT!
It reminds me of https://youtu.be/TYsulVXpgYg
We are living in a sci-fi novel
InterestingâŚ
Is AI making your code worse?
Ben and Ryan are joined by Bill Harding, CEO of GitClear, for a discussion of AI-generated code quality and its impact on productivity. GitClearâs research has highlighted the fact that while AI can suggest valid code, it canât necessarily reuse and modify existing codeâa recipe for long-term challenges in maintainability and test coverage if devs are too dependent on AI code-gen tools.
https://stackoverflow.blog/2024/03/22/is-ai-making-your-code-worse/
Using the wrong tools We are at the beginning except it doesnât take AI many years to learn how to become a better programmer- it takes weeks, if not days. No doubt there will be resistance and self denial but everyone has an opportunity to start learning how to use these tools properly, because in 1 to 2 years (if not sooner) the programmers world is going to be a very different place!
Weâll be lucky if we make it another one to two years with all this Ukraine nonsense, and if that doesnât kick off then we are well overdue a significant global cyber conflict, which will definitely put a nail in the coffin of all things tech related.
Quote from my life story!
I always keep an eye on this thread to see if the new Beta version is going to have a handy AI agent integrated. I recently chose to build a project using Replit - which has come a long way since I posted this and was blown away by how fast and how dam complete it was. Not only the building of the application but the planning, explanation, learning and deploying - well everything really. I like working with Wappler because it was fun, but having experienced the help of an AI Agent and Code Assistant I canât go back. I can see how awesome it would be in Wappler and wonder if there is resistance to include AI in its feature set. Not just to help build out an applications foundation but to answer questions about the code, processes, errors and help with finding issues. Anyway, AI is not going away and my hope is the team is listening.
George has mentioned in other posts that they have explored AI capabilities, but they don't see value in integrating them yet. I am paraphrasing, so take my words lightly. You could search for his exact comments if desired.
The race is speeding up though. Many companies and creators of AI products say they will replace various roles and functions with AI. I read an article today about Replit's CEO saying they "don't care about professional coders anymore" and cut their staff in half due to their "Agent" product.
It's scary to think about. So many jobs are going to be affected and it will only result in the further concentration of wealth by those who are mainly driven by money and power. Only our AI overlords will be able to save us then. Better be nice to them.
The approach Wappler is taking of 'waiting until it's better' is wrong in my opinion.
AI is getting, and will continue to get better at a remarkable pace. Once it's 'good enough' for the Wappler team, it'll be too late. Platforms like Replit will be better, significantly. You should implement AI knowing it will be better, and continue to expand functionality rather than waiting.
I believe Wappler is uniquely positioned to offer something different.
If they integrate generative AI to the platform Wappler has the visual builder to view the API's/actions created and enhance them - right now, this would be a major value add over Replit. Replit can get stuck in loops trying to build complex integrations/workflows, and you really then need to inspect/question line by line for really complex stuff. Wappler can do this visually. It's a really good match with Wappler so i'm still surprised at the reluctance.
They should also look at implementing tools that allow users to build AI apps, e.g.
https://sdk.vercel.ai/ - Svelte I assume would be needed to be included and bundled, but this would allow users to build very slick front-end AI interactions like chat.
Langchain - this would give users the power to create agentic applications (langchain best for reasoning/automatic decision making)
Agents - LlamaIndex - strong in automatic indexing / retrieval.
As I know Wappler won't add AI anytime soon I'll be looking at the vercel SDK quite soon.
Wappler could really empower users by implementing GenerativeAI that's editable with a visual editor and integrate tools that Wappler users can use to build AI/Agentic applications.
I for one would advocate for a price increase so they can hire an AI expert to help them achieve this.
I think the problem with adding AI to Wappler is 'What do you expect it to do?'
I can't think of a possible scenerio in Wappler that I would need AI that I can't already get from ChatGPT on the rare occasion I need to look something up.
What are some specific examples of tasks you would do in Wappler with AI?
Wappler helps you to develop in a no-to-low-code way, visually. You can do everything you do in Wappler by writing code yourself, but you yourself choose Wappler as it helps you to achieve what you want to achieve faster, and/or it makes it possible (if you can't code). When the team implement a new feature (like the new Data panel) it's an improvement in our workflow, or what we can achieve. AI does the same.
Wappler is great for CRUD apps, but we are moving away from that to AI-Layer applications so offering AI functionality makes total sense.
Why do you use Wappler? You could do exactly the same with boilerplate apps, and coding.
If you look at Replit, I can describe an app, it plans it, and creates it VERY quickly. So the value add here is a huge time saving. Huge. It struggles a little still with super complex ideas, but it's getting better.
Why introduce AI:
- Tooling for the new generation of applications being built
- Speed to market of features, apps, ideas
- Time to develop anything reduced
- Develop much more complex functionality (e.g. I implemented BullMQ recently with the use of AI, I could not have done it without. If the AI had app-or-Wappler context it would have been even easier).
That's it from me on the topic as I know the Wappler team is unlikely to budge in the short-term.
- Allow for chat-based creation and modification of entire projects. Make the AI create all of the files, but let us view/modify along the way.
User - "Hey Wappler, I want to create a new marketplace app. It needs to have a database for buyers and sellers, a product table, and other related tables that you believe are important for an app like this.
Wappler - "Certainly. I created all the necessary Wappler files and committed to git. Please review the database structure in the Database Manager"
User - "That looks great! Now I need the base pages for the app created."
Wappler - "got it. Based on common designs for marketplace apps I've created the main page, listing page, messaging between buyer and seller...."
- Allow for chat-based creation and modification of custom extensions.
Even though these can be accomplished with a lot of back and forth via external tools, it would be helpful if it was built-in and understood how Wappler creates files and where it stores them, so the tedious manual efforts that a lot of us go through were abstracted away.
Replit I can write a detailed spec about what I want the more specific the better And it will create it! And I mean everything! 95% with a few tweaks on behaviour. So the expectation is I do the same with wappler. And we are talking minutes instead of weeks!
Woah, now that's impressive!
Have you tried Windsurfe (now codium)? They have Agent Flows, it works the same as Replit, but it's looking more reliable, and can handle more languages like iOS.
o3-mini out in < 2 weeks, then they'll work on o3 pro, which will be much smarter than o1. Wappler really needs to look at this, now. Right now Wappler has an advantage in that it could have a visual builder + AI.
@mgaussie Looks like it is now qodo.ai but looks impressive.
Here is a great article on Agentic Agents used in development. Agentic Coding: How I 10xâd My Development Workflow | by nicolas | Jan, 2025 | Medium . As a single example we are talking about cutting down weeks of work into minutes. 2025 will be the year of the AI agents, with companies like Salesforce and and Meta not hiring this year but instead expanding the use of itâs proprietary Agentic development.