Solveit: A Polya-Inspired, Human-in-the-Loop AI Workspace for Deliberate Coding

Answer.AI’s Solveit blends Pólya’s problem-solving method with an iterative coding environment and AI that shares the developer’s context. The platform prioritizes human control, fast feedback, and dialog engineering to keep AI helpful and learning effective. A five-week program starting Oct 20 offers full access and lessons, including sessions with Eric Ries.

Key Points

• Solveit operationalizes Pólya’s four-step problem-solving framework for coding and beyond, emphasizing small, verifiable steps and reflection.
• Fast feedback loops via a REPL/notebook-like workflow reduce errors and accelerate learning compared to big-bang coding.
• AI works best with shared context; in Solveit the AI sees what the human sees, can use the same tools, and acts as a collaborative partner.
• Dialog engineering (editable histories, pin/hide, context control) counters autoregressive drift and improves AI session quality over time.
• The tool is designed for human-in-the-loop mastery—not AI replacement—with a five-week course starting Oct 20 featuring Eric Ries and full platform access.

Sentiment

The community is notably split. Supporters — especially former cohort participants and fast.ai fans — are enthusiastic about the methodology and vouch for its value. However, a significant portion of commenters are confused by the messaging, skeptical of the pricing model, and some are outright dismissive. The founders' heavy engagement in the thread helps clarify confusion but also reinforces the perception that the product needed better upfront communication.

In Agreement

• Former cohort participants vouch strongly for the course's value, saying it changed their programming approach and taught proactive rather than reactive learning
• The human-in-the-loop philosophy resonates with developers who noticed AI tools like GitHub Copilot hinder learning when used without discipline
• fast.ai alumni praise the team's track record and recommend the approach based on their transformative experience with Jeremy Howard's previous courses
• The platform's unique features like editable AI responses, Python function-as-tool conversion, and collaborative notebook coding are seen as genuinely novel additions beyond existing tools

Opposed

• The article fundamentally fails to explain what Solveit actually is, requiring multiple founder clarifications in the comments before anyone understood the product
• The $400 price for a 5-week course raises skepticism, with critics questioning whether it's just selling a methodology that experienced developers already practice
• Some view it as a grift — selling lectures and a glorified Jupyter notebook with AI features, not a genuinely new platform
• The incremental bottom-up approach may be inferior to a top-down methodology where AI generates complete constructs that are then reviewed and understood
• The simplest antidote to AI fatigue is just using less AI, not buying another AI-adjacent product