By now, where I stand on LeoAI is clear.
Over the last few months, I laid out where I thought things needed to go and what the area of focus should be. During that time, I posted a "roadmap", if you will, of some ideas. It was based upon the progress I saw taking place and a guesstimate about what the compute costs might be.
Things are advancing so rapidly that I was caught off guard. What was sensible a couple months ago is no longer.
For this reason, let us update some things and I will explain the reason why.
Image generated by Ideogram
LeoAI: Full Service
We are awaiting the arrival of LeoAI.
To me, the most important feature that LEO can roll out is the chatbot. This is crucial and should be done as soon as possible. Naturally, we are dealing with something that has to be trained and edited so rushing the process is not feasible.
That said, I believe this is the first iteration of Leo as an AI platform. If Leo does not achieve this, it is dead in the water. That is true, by the way, for any site.
Obviously, most who are reading this have tried chatbots in the past. What they can do with text is pretty amazing.
This is, however, only the bare minimum.
After that, we have image generation. Anyone who used application such as Ideogram know what I am referring to.
Finally, there is video generation. This was something that I projected out at least deep into 2025, with 2026 even being possible. Here is where the surprise enters.
Take a look at this video. It is what Meta just released last week.
@taskmaster4450le/re-leothreads-zzzarnyd?referral=taskmaster4450le
No longer is just a chatbot sufficient. LeoAi must include image and eventually video generation. Naturally, there is a trade off with cost as inference is not cheap. However, having a full suite of services inherent on Leo is crucial.
Get Ahead of the Curve or Get Crushed
I project that, by the end of this decade, if the pace of advancement keeps up, we will see the death of the website.
This is stunning to some yet this is the transition.
With our present form of the Internet, the webpage is the basic element. This is where everything stems from. It is what people visit, how they receive their information.
How can this disappear?
Simple. We are looking at the next iteration where AI agents are the norm. Website will not be required to provide information since that will be accumulated by the agent.
This might seem like an aggressive forecast yet, as the basis of this article shows, the pace of disruption cannot be underestimated.
For this reason, digital platforms have a choice: get ahead of the curve (as best they can) or get crushed. There is no middle ground.
LeoAI: Prime Position?
Fortunately, when it comes to Leo, the model is being trained on Llama. MovieGen is a Meta creation. They also have integrated multimodal into Llama 3.2, allowing images to be fed into the chatbot.
All of this could potentially be integrated into LeoAI. We will have to see what is accessible in the open source end of things. Meta is opening up a lot, but not everything.
The point here is to provide as much AI utility on Leo as possible. This is what is required to be a successful platform. To think otherwise is risking ending up like the Luddites.
We can see the moves Meta, X, and Google are making. These are social media platforms (along with search) that is going all in AI. Those platforms are having AI incorporated into everything.
There might be a great deal available to Leo simply through what Meta is creating. Here is where piggybacking off that company's output can be of benefit.
When we couple this with all the data that we are feeding it on Hive, especially through Threads, we can see how this can produce a useful AI system.
This is where I stand...for now
With how things are going, we might have to lay out a new plan in another 6 months.