The 5 Levels
of Task Execution

A map of how tools for hands-on work have evolved — from static, offline guidance to fully autonomous systems. It helps an organisation see where its current tools sit, plan the next step, and judge new ones. Heylper aligns with this framework, offering adaptive and proactive guidance for real-world tasks.

The current research paper — Executable Procedural Memory Graphs — describes where this framework leads.

1.Why a framework

Tasks are getting harder. With increasing task complexity and widening skill gaps in homes and workplaces, traditional tools such as manuals and videos are insufficient.

A roadmap is needed. The 5 Levels give a structured path for developing and adopting tools that meet these challenges. Each level is a milestone in technological and procedural advancement.

2.The five levels

• Level 1Static guidance
• Level 2Sequential guidance
• Level 3Adaptive guidance
• Level 4Proactive guidance
• Level 5Autonomous execution

3.Each level in detail

Level 1 — Static guidance

Offline tools like manuals, checklists, or videos. Users must interpret the instructions themselves.

Example: a printed manual for assembling furniture.
Limitation: static; no adaptability, real-time interaction, or feedback loop.

Level 2 — Sequential guidance

Real-time, step-by-step instructions delivered in order, without adaptation.

Example: a cooking app that advances to the next step after manual confirmation.
Limitation: no context-awareness; cannot adjust to errors or unexpected changes.

Level 3 — Adaptive guidance

Hands-free, real-time voice guidance that adjusts dynamically to the user's progress and errors.

Example: AI detecting incorrect assembly during furniture building and suggesting corrective actions.
Limitation: reactive rather than predictive; relies on detected inputs to adjust.

Level 4 — Proactive guidance

Tools that predict potential issues and alert the user before errors occur.

Example: AI tracking tool positions through a camera and warning about misalignment before the user continues.
Limitation: cannot execute the task; requires the user to act on the alert.

Level 5 — Autonomous execution

Fully autonomous systems or robots executing tasks independently, with minimal user input.

Example: a robotic arm assembling parts on its own and sending progress updates to a connected app.
Limitation: expensive and domain-specific; lacks general adaptability across diverse tasks.

4.Applying the framework

In homes, the framework covers DIY tasks such as cooking, repairs, and gardening. Users progress from static tools (Level 1) to predictive systems (Level 4) for safer, more efficient execution.

In workplaces, frontline teams gain efficiency through adaptive (Level 3) and predictive (Level 4) guidance, eventually enabling autonomous execution (Level 5) in industrial settings.

5.Where Heylper sits

Today: Heylper operates at Level 3 — adaptive guidance — using language models to provide real-time, context-aware support.

Next: Level 4, proactive guidance, by training skill-sharing AI on anonymised task data.

Beyond: Level 5, autonomous execution, in specific domains — particularly where robotics and IoT intersect. The two films on the homepage are that path made visible: the same checklist, verified the same way, first for a person and then for a humanoid.

6.Conclusion

The 5 Levels offer a clear pathway for addressing task-execution challenges. Heylper exemplifies the progression, advancing from adaptive to proactive guidance and beyond — a structured approach that scales across users and industries.

© Heylper Inc. Heylper™