Project ECHO: Breathing Life into Office Notes with AI

Artificial intelligence (AI) is technology that enables computers and machines to simulate human learning, comprehension, problem solving, decision making, creativity and autonomy.

Source

The idea for this project began when we noticed numerous small pieces of paper in the office on which people wrote various notes, such as how they felt that day, current events, and so on. This inspired us to create a personality from these notes. Using AI, we aimed to display how this imaginary person—whom we later named ECHO—would feel at any given moment. The goal was to bring this personality to life and share it with the office.

To bring our project to fruition, we identified the following technologies:

To allow users to view the generated personality of ECHO, we installed a touchscreen on a Raspberry Pi. After setting up the touchscreen, we used the official Raspberry Pi Imager to install the Raspberry Pi OS Lite (64-bit), which doesn’t include a graphical user interface (GUI), making it faster and sufficient for our needs. We configured the OS to launch Headless Chromium in kiosk mode, displaying our site where the generated personality is shown.

With the Raspberry Pi configured, we developed the necessary website endpoints:

• An index page to display the generated response.

• A page to upload a file with the generated response.

• Pages to interact with the database (get and set data).

• A page to view all data from the database.

• Pages to set and view suggestion data.

Here are screenshots of the frontend pages we’ve implemented:

Ollama is an open-source platform that simplifies running large language models (LLMs) locally on your computer. It allows users to download and execute LLMs for personal use. More information about Ollama can be found on their official GitHub repository and their official website here.

We utilized a PowerShell script for most of the REST API calls in this project. The script performs several steps:

1. Retrieves all data from the database and saves it for generating the persona and suggestions.

2. Sorts the data by date to prioritize the most recent entries.

3. Sends a REST API call to the Ollama API to generate the persona using the model gemma3:27b on our local server

Our prompt for the LLM looks like this:

4. The generated response is sent via a REST API call to our endpoint, where it is saved in a file that our index page        reads from.

The final steps are similar, but instead of generating the persona, we prompt the LLM for suggestions and send that response to the suggestion data setter. The prompt for this LLM call looks like this:

The Task Scheduler enables you to automatically perform routine tasks on a chosen computer. The Task Scheduler does so by monitoring whatever criteria you choose (referred to as triggers) and then executing the tasks when those criteria are met.

Official Documentation

To automate the process, we used the Windows Task Scheduler to run our PowerShell script at regular intervals. The Task Scheduler allows us to define what task to perform, when to perform it, and for how long. Below is how we configured our task in Task Scheduler:

Project ECHO transforms everyday office notes into a dynamic, AI-generated persona using technologies like Raspberry Pi, PHP, MySQL, PowerShell, and Ollama’s local LLM. This integration fosters a unique blend of automation and personalization, enhancing workplace engagement. Despite challenges in user interface design and system reliability, the project demonstrates the creative potential of AI in human interaction. ECHO serves as a testament to the innovative possibilities when technology is harnessed to enrich workplace culture.

And here is the final result of this project: