First Step

The description states that First Step is a productivity tool for people with ADHD. It allows users to capture thoughts, notes, links, emails, and files without needing to categorise them first. These raw captures are stored before being processed by AI (GPT-5.6) into validated tasks or memories.

It offers a “Focus” surface that recommends one task at a time based on user-defined capacity, time, and mode. The system supports optional adaptive learning from user actions but does not override explicit inputs.

The product includes a “Brain” feature which stores captures, evidence, history, and open loops using PostgreSQL full-text search and pgvector hybrid retrieval.

It is built with Next.js, TypeScript, Supabase (for authentication, data storage, and vector search), and uses Codex for development with GPT-5.6 models.

Inference The tool appears to be a lightweight, AI-assisted note-taking and task management system tailored for individuals who struggle with organisation due to ADHD.

The author claims First Step was built from personal experience with ADHD and an attempt to solve the problem of “capturing a thought still created work.” It positions itself as a quieter alternative to traditional task managers, where users do not have to decide whether something is a task or note, assign projects, estimate effort, or sort items.

The product evolved from a form of self-rescue into a tool tested against various personas (creators, founders, researchers, business owners, operations workers). The author notes that it is not a diagnostic or clinical product.

Inference Positioning has shifted from personal utility to broader applicability through persona testing, though no evidence exists of actual customer adoption or market validation beyond the author's own use case and limited user feedback.

The description states that First Step is intended for people with ADHD. It was tested against five personas including creators, founders, researchers, business owners, and operations workers.

Inference While the core target is identified as individuals with ADHD, the author suggests it may also appeal to others who struggle with task management or cognitive overload. However, no specific ICP definition or segmentation data is provided.

There is no mention of pricing, monetisation strategy, or business model in the description. The project is described as a personal solution and a hackathon submission.

Inference No commercial structure is evident from the self-reported information.

The product is built using Next.js, TypeScript, Supabase (for authentication, data storage, PostgreSQL search, pgvector), and Codex for development. It uses GPT-5.6 models (Luna and Terra) for structured extraction and grounded recall.

It implements deterministic ranking combined with deadline and momentum signals to recommend tasks that fit the user’s current capacity. The system retains raw captures before AI processing and avoids mutating tasks or overriding explicit inputs.

Inference The technical stack suggests a modern, serverless SaaS-like architecture with strong emphasis on privacy and control over data flow. However, no evidence of scaling, performance metrics, or production deployment beyond the developer's environment is given.

There is no evidence of revenue, customers, user base, or adoption metrics. The project was submitted to a hackathon and described as a personal solution.

Inference No traction signals are evident. The only validation mentioned is internal testing with personas and the author’s own experience.

The description mentions trying various task managers, note-taking apps, and productivity systems before building this tool. It contrasts itself with tools that require upfront organisation and decision-making.

Inference It competes with traditional task management tools (e.g., Todoist, Notion, Trello) and possibly AI-powered assistants like ChatGPT or Obsidian, though no direct competitor names are listed.

• Lack of external validation: No evidence of real-world usage or customer feedback beyond the author’s own experience.
• Unverified claims: The product is described as tested against personas but lacks data on actual user engagement or satisfaction.
• No commercial viability: No pricing, monetisation, or business model details are provided.
• Single-person team: The project is built by one person (nikos stasinopoulos), raising questions about scalability and long-term maintenance.
• Limited scope of testing: Persona testing does not equate to market validation.

1. What specific feedback did you receive from the personas tested, and how was it incorporated?
2. Are there any early adopters or users beyond your own experience and persona testing?
3. How do you plan to scale beyond a single developer’s capacity?
4. What is your intended monetisation model, if any?
5. Have you considered privacy implications of storing personal data in the cloud with AI processing?