Free 40-page Claude guide — download today
April 15, 2026Claude Skills Hubclaudepromptsestimation

Estimate with Claude

Optimize estimation with Claude using L99 and /deepthink prompts

Estimating with Claude: Why Your Current Approach Isn't Working

If you're searching for a Claude prompt for estimation, chances are you're trying to use AI to simplify complex calculations or predict uncertain outcomes. Perhaps you're a project manager trying to estimate the time and resources required for a new project, or a business owner trying to forecast sales and revenue. Whatever your use case, you're likely finding that your current approach to estimation is time-consuming, prone to errors, and often inaccurate. This is because estimation requires a combination of data analysis, domain expertise, and critical thinking – skills that are difficult to replicate with traditional prompting methods.

The Pattern that Works

To overcome these challenges, we've found that stacking the L99 and /deepthink prompt codes can be highly effective. The L99 code provides a framework for breaking down complex problems into smaller, more manageable components, while /deepthink encourages Claude to think critically and consider multiple perspectives. By combining these codes, you can create a prompt that not only estimates outcomes but also provides a clear explanation of the underlying assumptions and calculations.

Before and After Example

Here's an example of how this pattern can be applied in practice: Before: What are the estimated costs of developing a new mobile app? After: Using L99 and /deepthink, estimate the costs of developing a new mobile app, considering factors such as development time, team size, and technology stack. Provide a detailed breakdown of the costs and explain any assumptions made in the estimation process. In response to the After prompt, Claude might provide a detailed estimate of the costs, including:

  • Development time: 12 weeks
  • Team size: 4 developers, 1 designer, 1 project manager
  • Technology stack: React Native, Node.js, MongoDB
  • Estimated costs: $200,000 - $250,000
  • Assumptions: The estimate assumes a moderate complexity app with a standard set of features, and that the development team has prior experience with the technology stack.

Anti-Patterns that Don't Work

While the L99 and /deepthink combination is effective for estimation, there are several anti-patterns that people try which don't work. For example:

  1. Using the /trim code alone, which can oversimplify the estimation process and lead to inaccurate results.
  2. Stacking /hook and /simplify, which can create a prompt that is too focused on a specific aspect of the problem and neglects other important factors.
  3. Applying the SENTINEL code, which is better suited for anomaly detection and monitoring rather than estimation.

When NOT to Use this Approach

While the L99 and /deepthink combination is a powerful tool for estimation, there are certain situations where it may not be the best approach. For example:

  • When dealing with extremely complex or nuanced problems that require human expertise and judgment.
  • When the estimation requires a high degree of precision or accuracy, such as in financial modeling or scientific research.
  • When the underlying data is incomplete, inaccurate, or biased, which can lead to flawed estimates and incorrect assumptions.

Next Steps

To learn more about the L99 and /deepthink codes, as well as other prompt codes that can be used for estimation and other tasks, see all 120 codes tested over 3 months in the Cheat Sheet. This comprehensive resource provides a detailed guide to using Claude for a wide range of applications, from estimation and forecasting to text generation and conversation.

Want the full research library?

120 tested Claude prompt codes with before/after output and token deltas.

See the Cheat Sheet — $15