Analyse the Workato Job Log with MemGPT and AutoGen

Ferry Djaja
8 min readDec 11, 2023

In this tutorial I would like to go through my use case to get the Workato recipe job log real-time and instruct the agent to conduct the analysis and subsequently generate a chart based on the obtained data. For this use case, I will be using MemGPT with AutoGen.

Memory-GPT (MemGPT)

MemGPT enables LLMs to manage their own memory and overcome limited context windows. Large Language Models (LLMs) face a limitation in terms of their context window, which restricts their effectiveness in extended conversations and document analysis. For more details, please refer to https://memgpt.ai/.

AutoGen

AutoGen is a framework that facilitates the creation of Large Language Model (LLM) applications by employing multiple agents capable of engaging in conversations with each other to collectively address various tasks. For more details, please refer to https://microsoft.github.io/autogen/.

Conversational Flow Scenario

For this use case, I have the following scenario, which represents the dialogue sequence among the agents.

  1. Here is the sequence of communication in which the individual is requesting a list of Workato recipes with folder ID xxx.
  2. The Workato Analyst receives the task from the human proxy and instructs the MemGPT coder to fetch the information.
  3. The MemGPT Coder then performs the function to retrieve real-time data from the Workato platform and furnishes the information to the Workato Analyst for analysis and commentary.
  4. Subsequently, the Workato Analyst shares the data with the UI Designer.
  5. Upon receiving the data, the UI Designer creates the Python code to construct the chart and forwards the code to the Analyst.
  6. The Analyst proceeds to execute the Python code to visualize the chart.

--

--