Hello community, I’m repeating the strategy I used when I launched my previous module (Ticket Plus & HelpDesk & AI Chatbot) with this new module “AI Assistants”. That is, offering it at a very ** reduced price of 10€ ** to you who are always attentive to the community in the forums, and also to my clients of my previous modules, to whom I have sent this same announcement by email.
What this module consists of
It’s very simple, but very powerful: being able to use the different generative AIs that have already come to market within our own Dolibarr without having to pay $10-20 USD for each user and month to these providers, but using/paying the API of these providers.
In other words, each user/employee in your Dolibarr will be building their own conversation history, just as one does in chatGPT or Perplexity or Claude, etc. but without depending on a specific AI provider:
- being able to use ANY provider in the SAME environment
- being able to CHANGE model/provider IN THE MIDDLE of a conversation, without losing its progress
- building a UNIQUE history of conversations in which to semantically search (using embeddings) for previous “similar” conversations
- directly STORING in our Dolibarr server not only text conversations, but images generated by text-to-image AIs
- and in the FUTURE: being able to interact with our Dolibarr database, to query, summarize, retrieve, etc.
AI Providers
So far, I have developed these integrations:
- with Openai, chatGPT in its GPT-4o and GPT-4o-mini family
- with Perplexity, opensource llama 3.1 models for chat or INTERNET SEARCH
But in the coming weeks I will integrate:
- image generation with FLUX through fal.ai, which works wonderfully (I’ve already integrated it into another app)
- audio transcription with OpenAI’s Whisper
- conversation with Claude
- through fal.ai’s API, all kinds of generative AIs can be incorporated at very economical prices (image to video, text to video, text to audio, audio to text, image to image, etc.)
As you must know, this is a technological sector in explosion, like a supernova!
Limitations
Obviously I know it’s difficult (impossible?) to emulate the complexity of the interfaces of many of those providers that offer when you sign up with a fixed monthly payment. So I don’t think a module like this will ever be a 100% replacement for what they give us. But certainly we can get 90% of what they offer by paying barely 10% or much less.
The intention is not to fall behind and get the most out of it for our teams and for our clients. And to do it quickly.
Privacy
I know that many people distrust those bigtech companies that offer these AI services, even if it’s through an API and despite promising that the data processed by their servers is not kept. But friends, if you didn’t know, our data (ALL?) is already in the hands of a number of companies that we can’t imagine. Those who made your web browser, your Office, your operating system, your antivirus, your apps, even your internet connection company. Obviously many times that information is encrypted, but many other times it’s not, and most of the time it has to be decrypted at the destination where caches, backups, etc. are kept.
I’m not justifying it. I’m putting in the balance the risk of giving a little more of our data and how much we receive in return when we use this type of AI technologies.
Anyway, as many know, it will soon be more common to host a generative model on your own server (in fact you can already do it) and in a very short time, those of us who are more distrustful will be able to connect modules like this to those “endpoints” OF OUR OWN as we now connect with OpenAI, Perplexity or Anthropic.
I count on your collaboration
You are always exceptional allies for being more intensive users of Dolibarr and its ecosystem of modules. So I thank you in advance for your collaboration with any suggestions you may have. I’m going to try to work very intensively for the rest of September to try to complete all the ROADMAP that I have proposed for this module.
Price 10€ SEMI-ANNUAL
When you go to the module’s file on Dolistore you will see that the price is only 10€, but this time I have decided that access to updates will last only 6 months.
In 7 years it’s the first time I don’t offer lifetime updates of my modules. The difference is justified in that this module will require me to WORK ON IT EVERY MONTH OF THE YEAR, FOR YEARS (hopefully). So, I find a periodic payment for updating the module fairer.
The final price will be 20€. I’ve put it at 10€ for this month of September, to encourage you to try it and help me leave it as “finished” as possible in these first weeks.
User guide and DEMO
This time I’ve rushed to publish a guide before launching the module:
User manual of the AI Assistants module for Dolibarr
At the end of that guide you have a form to request access to a DEMO portal that I have set up so that you can try it live. Although I have limited it to the use of the most economical models of both OpenAI and Perplexity (there is a difference of 25x!!!) but you will see that they respond sufficiently well in most cases, despite being the most economical model.
Currently both Perplexity and OpenAI are around $0.20 USD for processing ONE MILLION TOKENS which is equivalent to more than TWO THOUSAND PAGES of an average book.
Sorry for the length of this post, but I think this announcement required clarifying all these details. Anyway, go ahead with your comments.