Codesolvent Blog

We previously released our multi-modal AI assisted work space capabilities. Today we release some additional updates showcasing human-in-loop integration.   Vernon     Billie     Combined Demo    

AI tools are the tentacles exposed to AI models in order to allow them to access and drive external functionality. The primary form this takes for LLMs is function calling. We have also seen so-called "computer use" and web browsing designated in similar fashion. For code generation, we could also consider the ability to integrate dependencies in the form of external packages as a form of tooling. In other words anything that extends the capabilities of the AI/LLM model could be thought of as a tool. We can then extend this notion into the runtime environment for the code that LLMs generate by generating code that depends on runtime dependency injection. Late Binding Dependency injection is accomplished via "late binding", meaning the actual implementation of a given functionality isn't specified at the time code is written, until at some later point before execution of logic. Late binding can be implemented at the language level, such as in C++ you have virtual...

We have released updates to the Solvent-Botworx platform that includes the introduction of automation programs and the addition of multi-modal capabilities. Multi-modal speech capabilities in particular are an intriguing development for integration into "deep-work" work-spaces.  One could imagine a future where anyone doing heavy-duty cognitive work  will be working within a software workspace with seamless multi-modal AI integrations. With the right type of integrations, multi-modal capabilities offer the possibility of true AI aided assistant capabilities. Below are demo videos. Speech modality demo:   Image and speech modality demo:  

In this post we introduce using the fork/join concurrency pattern combined with the git-style cherry picking to manage and use LLM conversations.   These are loose conceptual connections, so please put down the pitchfork (pun intended). Just as in git you can cherry-pick a commit from one branch to create a new head on another branch, you can pick conversations across sessions (including between different models) to build context to drive LLM conversations. You can also create a fork of a session, this creates a parallel set of messages that are detached from, ie forked off the main session within which they run. You can create new messages that join an existing fork. This is a nice capability, the degree to which it will prove useful is yet to be determined. One of the possibilities is giving the user an ability to create long term "memory" from AI conversations and being able to easily reuse that memory (set of messages) without the need to rebuild context. The ability to...