A Coach for students to organize their tasks well

GoalSync AI is an innovative AI goal and progress coach designed to assist a friend in staying accountable to their long-term career objectives while managing daily tasks. This personalized tool breaks down significant long-term visions into manageable weekly milestones and specific daily tasks. Rather than functioning as a conventional task manager, GoalSync AI engages in daily conversations with its user, reviews completed tasks, verifies alignment with weekly milestones, and offers supportive, structured progress summaries.

The AI agent interacts with the user evening after evening, analyzing the tasks completed, assessing their adherence to weekly goals, and delivering encouraging, organized progress updates. The live demo showcases the AI agent conversing with the user, assessing daily tasks in relation to weekly milestones, and maintaining a contextually relevant multi-turn dialogue.

GoalSync AI is constructed on an open-source visual agent framework, complemented by fast, open-weight model inference. This includes Agent Harness / Framework: Flowise AI, a drag-and-drop platform for connecting LLM nodes, memory blocks, and prompt templates to create an agentic workflow. Fast open-weight LLM core is hosted on Groq Cloud, enabling high-speed inference.

The project employs open-weight models, such as Meta's Llama 3, to handle reasoning, planning, and progress tracking. Contextual Memory is integrated via a Buffer Memory node linked to a Conversational Chain node, ensuring the agent retains multi-turn chat context through daily check-ins. The system is guided by a Structured Coaching Prompt, which sets the tone for an encouraging, detailed style of response encompassing daily, weekly, and long-term tracking logic.

The concept of open innovation played a crucial role in the development of GoalSync AI. This approach emphasizes data privacy and personal control, ensuring that sensitive daily goals, habits, and career targets remain within controlled pipelines. Additionally, the use of open-weight models via high-speed inference infrastructure results in the tool being free of cost.

The visual, open canvas design of the project allows for flexibility in model selection if the need arises to switch models or adjust prompt parameters, without the need for extensive code rewriting.