# Project export: Motherly.ai This document was generated by HackStack to give an AI agent context about a hackathon project. Sections are labeled with their provenance; content marked as truncated was cut to keep this document small. ## Project metadata - Hackathon: TreeHacks 2025 - Tagline: Your mom knows you better than you do. - Devpost: https://devpost.com/software/ai-mama - GitHub: https://github.com/MingkuanY/ai-mommy - Video: https://www.youtube.com/embed/eehiiErNsOM?enablejsapi=1&hl=en_US&rel=0&start=&version=3&wmode=transparent - Result: winner (Delve: Best Agentic Workflow (AirPods + Delve Founder Numbers + Onsite); Scrapybara: Capy Prize (Japan offsite + $1k credits + Scrapybara Pro 6 months [1st] & $1k credits + Scrapybara Pro 3 months [2nd] & $100 credits + Scrapybara Pro 1 month [3rd])) - Team: 4 GitHub contributor(s) — Pranav Tadepalli (25 commits), sophiasharif (23 commits), Sonav Agarwal (21 commits), Mingkuan Yan (12 commits) ## Devpost submission (written by the team) ### Inspiration In the past year, 74% of people have felt overwhelmed by stress, and 30% of young adults in the U.S. experience loneliness multiple times a week. We wanted to create a large-scale solution and turned to what helps us most with both stress and loneliness: our moms. While AI companions like Character AI provide emotional support, they can’t truly take care of you. Our goal was to build an AI agent that deeply understands your emotions and takes action to help you. ### What it does Stress management is deeply personal - what works for one person might not work for another. That’s why we built Motherly.ai, an AI-powered agent that understands how you experience stress and takes real-world actions to support you. Using biometric data like EMG, heart rate, and blood pressure, it detects your mental state in real-time and responds with personalized interventions, whether it’s scheduling a break, ordering comfort food, or adjusting your environment to help you relax. Unlike traditional AI companions, Motherly.ai doesn’t just listen – it takes care of you. Some things Motherly.ai can do for you If reading the news makes you stressed out She can play calming music and surface good news in the world She can play calming music and surface good news in the world If you mention you’re stressed out and constantly doom-scrolling Motherly.ai can interrupt your scrolling, order food for you, give you a massage, and contact your friends so they can reach out to you Motherly.ai can interrupt your scrolling, order food for you, give you a massage, and contact your friends so they can reach out to you To help your child stay motivated while studying Motherly.ai can automatically give them a YouTube break between problems on Khan Academy when they’re stressed, or alternatively nudge them back to studying when they spend too much time distracted. Motherly.ai can automatically give them a YouTube break between problems on Khan Academy when they’re stressed, or alternatively nudge them back to studying when they spend too much time distracted. If you want her to help you stay productive and avoid distractions She can recognize when you’re on Instagram and bring you back to your work She can recognize when you’re on Instagram and bring you back to your work ### How we built it Accomplishments that we're proud of Hardware Hacking: We hacked a TENS unit to give an AI agent the capability of controlling the human body. Cutting-edge research: We used EMGs to create a stress monitoring system based on several papers that correlate EMG activity in the trapezius to stress with p<0.0001 Experimentally Induced Stress Validated by EMG Activity, A Comprehensive Analysis of Trapezius Muscle EMG Activity in Relation to Stress and Meditation The impact of adverse childhood experiences on EMG reactivity: A proof of concept study Experimentally Induced Stress Validated by EMG Activity, A Comprehensive Analysis of Trapezius Muscle EMG Activity in Relation to Stress and Meditation The impact of adverse childhood experiences on EMG reactivity: A proof of concept study Trained an LSTM classifier to predict your emotions based on your stress levels and biometric data Live Biometric Processing: Manage 4 separate real time data streams. Distinct Agent Actions: Created several distinct actions for our agent to take: Close the tab you’re on Switch the music to match your mood Order a surprise Domino’s order Text a friend Order matcha Prescribe a massage (using TENS) Book a yoga class Find fun events in your area Close the tab you’re on Switch the music to match your mood Order a surprise Domino’s order Text a friend Order matcha Prescribe a massage (using TENS) Book a yoga class Find fun events in your area Interactive Avatar: AI mother makes eye contact with your cursor as she helps you ### Challenges we ran into To get the EMGs to have consistent readings, it took SOO many different sensor placements on the trapezius muscle in order to get any sort of reading. After learning a lot about anatomy of shoulder muscles and bones, we finally got the electrodes in the right place to measure stress from these muscles, and got really amazing readings from our sensor! We initially tried to use the same signal processing methods mentioned in the papers (bandpass and butter filters, extracting only low-frequency data, etc.) but these were too slow for real time feedback, so we had to create more optimized methods: We accidentally ordered $20 worth of Dominos while testing our Scrapybara automation Making the AI agent do the right thing based on its prompts was very difficult. ### What we learned How to take readings using EMG sensors and how to give massages using TENS stimulation How to stream data between different services in real time How to create agents with LangChain ### What's next Shrink the EMG sensor to be a 24/7 wearable to enable long-term stress / mental health tracking Allow the user to give Motherly.ai custom personalities, making help more familiar and comforting ## README (from the GitHub repository) No README available. ## Detected evidence (automated analysis) Indexed codebase: 22 recognized source files, 226 KB. - CSS (language) — detected in the code - Next.js (technology) — detected in the code - Python (language) — detected in the code - React (technology) — detected in the code - Tailwind CSS (technology) — detected in the code - TypeScript (language) — detected in the code ## Codebase structure (from repository index) ### Files (42 of 42) ``` agent.py applescripts/sendMessage.applescript book_yoga.py bright.py classifer/emotion_labels.npy classifer/generate_data.py classifer/lstm_emotion_model.pth classifer/synthetic_emotion_data.npz classifer/train_lstm.py client/.gitignore client/eslint.config.mjs client/next.config.ts client/package.json client/postcss.config.mjs client/README.md client/src/app/globals.css client/src/app/layout.tsx client/src/app/mommy/page.tsx client/src/app/page.tsx client/src/lib/useClientRect.ts client/src/lib/useMousePosition.ts client/src/lib/useWindowDimensions.ts client/tailwind.config.ts client/tsconfig.json emg/1_raw_data_13-12_22.03.16.txt emg/read_data.py fun_events.py order_food.py server/.gitignore server/blood_pressure.txt server/body_temperature.txt server/heart_rate.txt server/history.txt server/lstm_emotion_model.pth server/predict_emotion.py server/rules.json server/sample_history.txt server/samples.txt server/server.py server/stress.py shock.txt stress.txt ``` ### Dependencies - client/package.json: @eslint/eslintrc@^3, @types/node@^20, @types/react@^19, @types/react-dom@^19, eslint@^9, eslint-config-next@15.1.7, next@15.1.7, postcss@^8, react@^19.0.0, react-dom@^19.0.0, react-markdown@^9.0.3, recharts@^2.15.1, tailwindcss@^3.4.1, typescript@^5 ### Recent commits (newest first) - Merge branch 'main' of https://github.com/MingkuanY/ai-mommy - Create final_arch.png - classifier! - add labels to sections - Merge branch 'main' of https://github.com/MingkuanY/ai-mommy - book yoga & fun events - Merge branch 'main' of https://github.com/MingkuanY/ai-mommy - Enhanced agent - rebrand - edited actions error - lol - Merge branch 'main' of https://github.com/MingkuanY/ai-mommy - stuf - Update order_food.py - Merge branch 'main' of https://github.com/MingkuanY/ai-mommy - make dependent on stress - Merge branch 'main' of https://github.com/MingkuanY/ai-mommy - Updated server to work with new prompt commands - fake data for terra - Merge branch 'main' of https://github.com/MingkuanY/ai-mommy ## Key source files (fetched from GitHub, selected and truncated for size) ### client/package.json ``` { "name": "ai-mommy", "version": "0.1.0", "private": true, "scripts": { "dev": "next dev --turbopack", "build": "next build", "start": "next start", "lint": "next lint" }, "dependencies": { "next": "15.1.7", "react": "^19.0.0", "react-dom": "^19.0.0", "react-markdown": "^9.0.3", "recharts": "^2.15.1" }, "devDependencies": { "@eslint/eslintrc": "^3", "@types/node": "^20", "@types/react": "^19", "@types/react-dom": "^19", "eslint": "^9", "eslint-config-next": "15.1.7", "postcss": "^8", "tailwindcss": "^3.4.1", "typescript": "^5" } } ``` ### server/server.py ```python import numpy as np from collections import defaultdict import random import json import flask from threading import Thread from time import sleep # from stress import compute_stress_data_from_file from typing import Dict, List from dotenv import load_dotenv from flask import Flask, Response, jsonify, request from flask_cors import CORS from pydantic import BaseModel, Field from langchain_openai import ChatOpenAI from langchain_community.tools import BaseTool from langchain.prompts import PromptTemplate from langchain_core.messages import SystemMessage, HumanMessage, AIMessage from langgraph.prebuilt import create_react_agent from terra.base_client import Terra import requests import threading import time import logging from predict_emotion import predict_emotion logging.basicConfig(level=logging.INFO) _LOGGER = logging.getLogger("app") terra = Terra(api_key="t0PMr4YpxCVtYc0M7bYGSpBuRwujEPvp", dev_id="4actk-aimommy-testing-ntJJIlrzqJ", secret="1a1e999f0665aeda4cf5a92335bce2cf4450f3a34fbb7273") # --- File Paths --- HISTORY_FILE = "history.txt" # Change this to the actual file path RULES_FILE = "rules.json" # Change this to the actual file path HEART_RATE_FILE = "heart_rate.txt" BLOOD_PRESSURE_FILE = "blood_pressure.txt" BODY_TEMPERATURE_FILE = "body_temperature.txt" app = Flask(__name__) CORS(app, resources={r"/*": {"origins": "*"}}, supports_credentials=True) app.config['CORS_HEADERS'] = 'Content-Type' load_dotenv() # --- Monitoring Rule Definitions --- class MonitoringRule(BaseModel): """Structure for a custom monitoring rule""" condition: str = Field( "Conditions based on the current time, stress level, and screen content.") actions: List[str] = Field( "Actions to take based on the condition. Available Action Types: CONTROL, MUSIC, NOTIFICATION, BRIGHTNESS, COLOR") priority: int = Field( "Priority level of the rule (1 (highest) to 5 (lowest))") condition_cute: str = Field( "Conditions field but briefly summarized in southern mothernly language") actions_cute: List[str] = Field( "Actions field but briefly summarized in southern mothernly language") priority_cute: str = Field( "A short priority level for the rule but in southern mothernly language") def create_monitoring_rule(rule: MonitoringRule): """Adds a new monitoring rule to the rules.json file.""" rules = read_monitoring_rules() print("------------------------------------------") print("Creating monitoring rule:") print(rule) print(rule.model_dump()) print("------------------------------------------") rules.append(rule.model_dump()) try: with open(RULES_FILE, "w") as file: json.dump(rules, file, indent=4) return "\n**Monitoring rule created successfully.**\n\n" except Exception as e: return f"\n**Error creating monitoring rule: {e}**\n\n" def read_monitoring_rules(): # make sure the file exists try: with open(RULES_FILE, "r") as file: data = json.load(file) print("Data: ", data) return data except: with open(RULES_FILE, "w") as file: json.dump([], file) return [] def read_history(): print("rh") with open(HISTORY_FILE, "r") as file: data = [[float(x) for x in line.strip().split()[:2]] for line in file] recent_data = data[-10:] # # print("Data: ", data) # # Take the last 5000 points (or fewer if there aren't that many) # recent_data = data[-5000:] # # Downsample by selecting 10 evenly spaced points # if len(recent_data) > 10: # indices = np.linspace(0, len(recent_data) - 1, 10, dtype=int) # recent_data = [recent_data[i] for i in indices] # # print("Recent data:", recent_data) return recent_data def gen_prediction(): try: # Load the last 10 data points for each of the 4 features # stress, heart_rate, blood_pressure, temperature stressData = read_history() biometrics = get_bio_data() stress = [x[1] for x in stressData] heart_rate = [x["rate"] for x in biometrics["heart_rate"]][-10:] blood_pressure = [biometrics["blood_pressure_high"]] * 10 temperature = [biometrics["body_temperature"]] * 10 # Make a prediction prediction = predict_emotion( stress, heart_rate, blood_pressure, temperature) return prediction except Exception as e: print("Error generating prediction:", e) return str(e) # def read_history(): # # File formatted as "time stress" per line # with open(HISTORY_FILE, "r") as file: # data = [] # for line in file: # parts = line.strip().split() # if parts: # data.append([int(x) for x in parts]) # return data # --- Legacy endpoints remain unchanged --- @app.route('/') def hello_world(): return 'Hello, World!' @app.route("/history", methods=["GET"]) def get_history(): print("Getting history rahhh") try: data = read_history() biometrics = get_bio_data() data = [{"time": time, "stress": stress} for time, stress in data] complete_data = { "history": data, "biometrics": biometrics, "prediction": gen_prediction() } return jsonify(complete_data) except FileNotFoundError: return jsonify({"error": "History file not found"}), 404 except Exception as e: return jsonify({"error": str(e)}), 500 @app.route("/rules", methods=["GET"]) def get_rules(): try: rules = read_monitoring_rules() return jsonify(rules) except FileNotFoundError: return jsonify({"error": "Rules file not found"}), 404 except Exception as e: return jsonify({"error": str(e)}), 500 # --- Prompt Template Setup --- system_prompt_template = PromptTemplate( input_variables [truncated — 9704 more characters] ``` ### client/src/app/layout.tsx ```typescript import type { Metadata } from "next"; import { Geist, Geist_Mono } from "next/font/google"; import "./globals.css"; const geistSans = Geist({ variable: "--font-geist-sans", subsets: ["latin"], }); const geistMono = Geist_Mono({ variable: "--font-geist-mono", subsets: ["latin"], }); export const metadata: Metadata = { title: "Create Next App", description: "Generated by create next app", }; export default function RootLayout({ children, }: Readonly<{ children: React.ReactNode; }>) { return ( {children} ); } ``` ### client/src/app/page.tsx ```typescript import Image from "next/image"; export default function Home() { return (
Next.js logo
  1. Get started by editing{" "} src/app/page.tsx .
  2. Save and see your changes instantly.
Vercel logomark Deploy now Read our docs
); } ``` ### client/src/app/mommy/page.tsx ```typescript "use client"; import { useClientRect } from "@/lib/useClientRect"; import useMousePosition from "@/lib/useMousePosition"; import useWindowDimensions from "@/lib/useWindowDimensions"; import { useEffect, useRef, useState } from "react"; import Markdown from "react-markdown"; import { CartesianGrid, Line, LineChart, ResponsiveContainer, Tooltip, XAxis, YAxis, } from "recharts"; interface Props {} interface ChatObject { sender: "user" | "mommy"; message: string; } const EMOTION_TO_MESSAGE = { fear: "Are you feeling scared? I'm here for you.", anger: "I'm sorry you're feeling angry. Let's talk about it.", sadness: "I'm here for you. I know you're feeling sad.", normal: "Seems like you're having a good day!", }; const page = (props: Props) => { const [chatInput, setChatInput] = useState(""); const [chatHistory, setChatHistory] = useState([]); async function sendChat() { // fetch data from localhost 5000 // add the user's chat to the chat history let newChatHistory: ChatObject[] = [ ...chatHistory, { sender: "user", message: chatInput }, ]; setChatHistory(newChatHistory); const response = await fetch("http://127.0.0.1:5000/ask", { body: JSON.stringify({ history: newChatHistory }), headers: { "Content-Type": "application/json", }, method: "POST", }); const reader = response.body?.getReader(); if (!reader) return; let done = false; let data = ""; while (!done) { const { value, done: done_ } = await reader.read(); done = done_; data += new TextDecoder().decode(value); // update the last chat // if the last chat object is from the user // then add a new chat object with the response from the server // else update the last chat object with the response from the server setChatHistory((oldChatHistory) => { if (oldChatHistory.length > 0) { if (oldChatHistory[oldChatHistory.length - 1].sender === "user") { return [ ...oldChatHistory.slice(0, -1), { sender: "user", message: chatInput }, { sender: "mommy", message: data }, ]; } else { const newChatHistory = oldChatHistory.slice(0, -1); return [...newChatHistory, { sender: "mommy", message: data }]; } } else { return [{ sender: "mommy", message: data }]; } }); } fetchActions(); } const [data, setData] = useState(); const [biometrics, setBiometrics] = useState(); const [prediction, setPrediction] = useState< "normal" | "fear" | "anger" | "sadness" >(); useEffect(() => { function fetchData() { // fetch data from localhost 5000 fetch("http://127.0.0.1:5000/history") .then((res) => res.json()) .then((data) => { console.log(data); setData(data.history); setBiometrics(data.biometrics); setPrediction(data.prediction); }); } // interval to fetch data every 5 seconds try { fetchData(); } catch (e) { console.log("Error fetching data"); } const interval = setInterval(() => { console.log("fetching data..."); try { fetchData(); } catch (e) { console.log("Error fetching data"); } }, 500); return () => clearInterval(interval); }, [setData]); const [actions, setActions] = useState(); function fetchActions() { fetch("http://127.0.0.1:5000/rules") .then((res) => res.json()) .then((data) => { console.log(data); setActions(data); }); } useEffect(() => { // interval to fetch data every 5 seconds try { fetchActions(); } catch (e) { console.log("Error fetching actions"); } }, [setActions]); return (
{/* Mommy */}

🤍 motherly.ai 🤍

Don't worry, mommy's here.

{chatHistory.length > 0 && (
{chatHistory.map((chat, index) => (
{chat.message}
))}
)}
setChatInput(e.target.value)} onKeyDown={(e) => { if (e.key === "Enter") { sendChat(); setChatInput(""); // setChatHistory([ // ...chatHistory, // { sender: "user", message: chatInput }, // ]); } }} className="w-full bg-transparent outline-none text-black" placeholder="What's up sweetheart?" >

I know how you're feeling.

{/* @ts-ignore */} {EMOTION_TO_MESSAGE[prediction]}

Your stress levels 🥺

Your heart rate 💓