I've moved to create metadata driven applications. This approach works well with AI as AI can be used to generate the metadata which powers the application.
For example I created an open source metadata crm called IceburgCRM iceburg.ca (available in Django / Laravel) recently. Without AI involved you could configure the crm by inserting data in the database. For example you add an entry for accounts, contacts, etc in the modules table and your crm will have menu links and a module section setup for accounts, contacts.
With AI you can now generate metadata needed to create custom crms. For example typing in "create a wine crm" in the AI prompt I build I can generate:
Using AI in traditional applications opens up a new dimension for us crud developers. I think we will see a shift to more metadata applications in the future as the benefits become clear. It's a new field using a different type of creativity and some old school design patterns.
I've created an open source crm / crm creator (iceburg.ca). I can create a crm from any existing database, from text using AI or custom parameters or use a number of premade templates.
I was looking at adding creating a crm from any website/page.
I'm interested in discussing steps 1 - 3. Using scrappingbee is expensive. I've brought down the price of a crm creation through AI to about 1 1/2 cents (+3 cents if you want a custom cover image). Have you looked at running something locally. I'm using laravel and was considering using Dusk to retrieve the contents.
What are you using for cleaning? Regex removals of tags?
OpenAI API for Structuring Data is new. What are your experiences? How does pricing compare to gpt3.5?
Any plans on open sourcing any part of your stack. What does your sass look like?
I'm not sure what you mean re: ScrapingBee being expensive. I am scraping some pretty heavy pages and it's costing about a half of a cent for a page. Less again with OpenAI. All-in-all, maybe sub 1c per page. Maybe there is more you are doing with this CRM creation process you have.
Right now in terms of open source or SaaS ... I think it is likely at some point I will replace ScrapingBee, maybe even OpenAI, with my own version of these and then take it down to a single API key and bill users myself. However I am leaving it free (albeit with BYO keys) for the moment as I am still building things out. I will likely continue to offer free to those who were already using it before I switch to a different model.
As mentioned, this is just the beginning of this project. The next features include logging in to accounts, giving instructions, so that ideally you can say things like 'book me a uber for 3pm to take me home from work', and the API gets it done and returns you the confirmation and whatever else you asked to be returned.
I built IceburgCRM (available in Django or Laravel). You can create a specialized crm with only a few words. For example: "create a stamp collecting crm". Modules, Fields, Relationships, Subpanels, default data and even a image for your homepage is created.
Data can be imported/exported into 6 different formats. Supports unlimited many to many relationships, 26 different themes, etc.
I've seen working with Laravel for years. I recently (Friday) released a Django version for my AI CRM builder (provide a few words
and use ChatGPT to build a CRM)
I used packages with similiar functionality to Laravel (orator, Django Breeze)
Everything translated well. The choice of backend didn't matter as much as I thought it would.
Using VS Code compared to PHPStorm was the only part of the experience that felt degrading. Using PyCharm might have changed that.
Large companies hire people to work on open source. Some use their clout to influence standards.
As a developer I like that companies hire for open source. As a user the quality increases with full time employees. What we don't notice is many projects morph to fit corporate needs because they become important stakeholders.
I create a CRM in Django CRM converting it from my popular Laravel CRM project. The default CRM mirrors SugarCRM/Suite structure. It also has some new features like unlimited relationships between modules.
The cool feature is it is also an AI powered CRM that lets you create a CRM by describing it.
This is my first Django project. I tried to reuse as much code as possible so the application mirrors a Laravel application using orator, Django Breeze.
I would love any feedback and suggestions for best practices.
Description
Iceburg CRM is a metadata driven CRM with AI abilities that allows you to quickly prototype any CRM. The default CRM is based on a typical business CRM but the flexibility of dynamic modules, fields, subpanels allows prototyping of any number of different tyes of CRMs.
Features
[Unlimited Relationships between any number modules without common fields]
[Metadata creations of modules, fields, relationships, subpanels, datalets, seeding]
[Ability to Import/Export in 6 different formats (XLSX, CSV, TSV, ODS, XLS, HTML]
[25 different input types, Laravel field validation, Maska field masking]
[26 themes with light and dark themes available]
[Module based Role permissions (read, write, import, export)]
[Calendar, Audit logs, Vue3 Charts, Convertable modules, Related Fields (related to another module)]
Could be as simple as the cost and lack of money. Weak content / weak promotions. Covid: packing into a small area with many strangers inside for hours seems risky now.
That becomes a branding issue. Many large companies own many smaller brands that are kept independent. You see this a lot in food also with mobile where each smaller brand can focus a smaller group. This gives customers choice but keeps profits with the entity.
For example I created an open source metadata crm called IceburgCRM iceburg.ca (available in Django / Laravel) recently. Without AI involved you could configure the crm by inserting data in the database. For example you add an entry for accounts, contacts, etc in the modules table and your crm will have menu links and a module section setup for accounts, contacts.
With AI you can now generate metadata needed to create custom crms. For example typing in "create a wine crm" in the AI prompt I build I can generate:
https://wine.iceburg.ca
"create a crm for bee keeping": https://beekeeping.iceburg.ca
"create a coffee lovers crm": https://coffee.iceburg.ca
Using AI in traditional applications opens up a new dimension for us crud developers. I think we will see a shift to more metadata applications in the future as the benefits become clear. It's a new field using a different type of creativity and some old school design patterns.