accountability_agent

Inspiration

Myself and ian often log/journal/use accountability groups, and we'd like to have a proactive agent that knows our preferences that can proactively reach out, knowing us and our preferences, but also retaining it's own identify, to keep us accountable.

What it does

1. User enters preferences for bot behavior (the bot identity)
2. User enters goals
3. Bot knows goals and reaches out programmatically according to user timelines and goals.

How we built it

We used an agentic tool that allows the user to specify the behaviors of an LLM. There are multiple profiles that we can assign personalities to the agent with by different methods.

Challenges we ran into

SMS messaging (US is very strict to prevent spam) General setup of API calls Tailoring of agent

Accomplishments that we're proud of

This is our first time using agents! Which was pretty straightforward retrospectively. Generally linking our outputs to other functions is satisfying, and seeing it as a message in your phone is cool.

What we learned

Using openai api calls, using SMS messaging via python, and managing an agent.

What's next for accountability_agent

Tailoring of timeline preferences for how it operates. So that it can proactively (but within user preference) keep in touch with the user.

Built With

• google-sms
• openai
• python