Royei Shaabon

• Product manager2025–2026

  Matrix DNA

  A B2B data platform. I joined a pilot with few users and a poor reputation and took it past 1,000 active users: ×12 in users and ×46 in active data sources, in 15 months.

  Read →
• AI agent · in production2026

  Backlog agent

  A Telegram bot that takes a voice note, checks for duplicates, proposes a feature, and writes it to the backlog and spec. The agent proposes, a human approves, a deterministic pipeline writes.

  Read →
• Agent orchestration2026

  Agent dev team

  An orchestrator, frontend and backend agents, two-layer QA, and a GitHub merge only after approval in Telegram. Including the experiment that failed and what it taught.

  Read →
• Full-stack · in use2025–2026

  Operational systems

  Two systems I built and deployed solo: logistics data collection for field teams (50 active users) and shift management for about 60 users. Next.js, Firebase, Cloud Run.

  Read →
• Founder2024

  BLISSe

  An AI nightlife app. Research, prototype and an MVP I built solo with GenAI tools. Selected for AWS Founders Club. Stopped when the paying side wouldn't pay.

  Read →
• Research & strategy2026

  EZpresence

  User and market research for an Israeli startup building an autonomous social media manager for small businesses. Three personas, 10 interviews, 10+ competitors mapped.

  Read →

And more

• PM tooling2026

  Backlog / Roadmap Dashboard

  A backlog with ID, quarter, tech review, technical risk and RICE, flowing into a roadmap and Google Sheets.

• Team project · ProductGO2025

  Duolingo: a growth proposal

  The course capstone, in a team of six: product and freemium analysis, competitors, and a proposal for a new growth engine. Deck · PDF

• Course project · market research2025

  Soft-skills training market

  About $86B in 2024. Demand is clear; the hard part is measuring behaviour change and proving value.

• Royei consistently demonstrated strong product thinking, strategic vision, and a deep understanding of both user needs and business objectives. He has a unique ability to connect the bigger picture with day-to-day execution. One of Royei's most significant contributions was driving the adoption of new product and collaboration methodologies across Product, Design, and Engineering.

• Royei consistently demonstrated a broad perspective and a level of professional maturity that usually takes years to develop. He is deeply connected to the product and highly committed to its success. Royei combines sharp analytical thinking with high motivation, taking full ownership of his tasks while never losing sight of the big picture.

• He showed sharp product thinking, strong analytical skills, and a structured, consistent approach. He was receptive to feedback, challenged himself, and improved iterations quickly. Royei brings high commitment, independence, and a fast learning ability.

1. The agent proposes, a human approves, a pipeline writes

   The agent has no write access. In a real system, one wrong row costs more than all the time automation saves.

2. Correctness before latency

   A faster, cheaper model was dropped because its tool calls weren't reliable. 20 correct seconds beat 5 guessed ones.

3. Set thresholds before the test

   What counts as success, what as failure, what we can spend. Written after the results, you always find a reason to continue.

4. Stopping early is a result

   At BLISSe the research showed the paying side wouldn't pay. I stopped before full development. Cheaper than any other version of the story.