Autonomous Data AI Agent
DataBy AI is a data servicing platform that features an autonomous AI agent, Gaby AI that automates the data-cleaning process.
Gaby AI Objective
The intent behind Gaby's development is for him to handle the temporal and long-term memory of my main agent, Tom - like a database gatekeeper.
In summary, Gaby is a set of a self-maintained data workflow that is responsible of managing the dataset prior and post model fine-tunings / training steps. Besides his data management capabilities, he comes with an in-built machine learning model that ranks data subsets, selects certain highlights from Tom's observations, and allocates them to their respective data storages.
Gaby’s ML model is motivated by the Google PageRank algorithm and other Markov chain–inspired models. It is still being actively experimented with, particularly in terms of how it might outperform current KV methods used for retaining caches during chat sessions, and how it integrates into Tom’s overall decision-making framework.
The primary network being explored is a Self/Multi-Head Attention architecture, where contextual information vectors are retrieved through transformation functions. At present, the only conclusive finding is that these transformation functions enable efficient inference over large text corpora by projecting them into lower-dimensional vector spaces—significantly reducing compute costs.
This reduction in complexity makes algorithms such as self-organizing maps more feasible for real-world applications. The remaining challenge lies in regularizing cost functions relative to the observational data gathered from Tom’s family members.
On a more enthusiastic note, his Cloud Development and API architecture has been much more trivial .. and fun.
Gaby AI Data Services
The current targeted areas are:
User's database connection: requires designing secured routes to the user's database like MongoDB, SupaBase etc. to assist data wrangling procedures.
Data Documentation and Modelling: requires external platforms API integrations and connections and fine-tuned autoregressive models on visualising charts (e.g. mermaid coder)
Data Cleaning & Processing procedures: data wrangling procedures on messy datasets.
Production:
http://databy.ai
Last updated
Was this helpful?