Get better AI results with 2 questions

Today’s models can solve complex mathematical problems, develop software applications, and find relevant information across huge datasets in seconds. Yet many users still get only mediocre results.

1. “What do you need from me to achieve the goal?”

With today’s models and agent capabilities, carrying out a digital task is no longer the main bottleneck.

The real bottlenecks are usually:

1. Clarity about the goal

2. The context required to achieve it

You are responsible for setting the direction, defining the goal, and determining what a good result should look like. amber can then help identify the information, criteria, and constraints it still needs.

Try this prompt:

“I want to achieve [describe the desired goal or outcome precisely]. What do you need from me to guide me toward this result?”

This simple question can be more effective than spending excessive time trying to write the perfect prompt from the start. Once the goal and context are clear, amber can help develop the right approach.

2. “How can you improve your instructions?”

One of our principles is: Perfection comes through iteration. This is especially important when working with amber.

When an output does not meet expectations, users often either add dozens of follow-up prompts or endlessly edit the original instruction. A better approach is to explain what is missing and ask amber to improve its own instructions.

Try this prompt:

“I am not satisfied with the following points: [describe the feedback or errors]. How should your instructions be revised to achieve the expected result?”

Review the proposed changes, make any final adjustments, and let amber revise its approach. This works well for chat prompts, skills, agent instructions, and workflows.

A repeatable workflow

Use both questions as part of the same process:

1. Define the goal
   Clarify what the result should achieve, who it is for, and what success looks like.

2. Identify missing context
   Ask amber what information, criteria, or constraints it needs.

3. Create and review an initial solution
   Treat the first output as a starting point, not necessarily the final result.

4. Describe the desired impact
   Explain what is missing and how the result should be different, rather than prescribing every individual correction.

5. Improve the instructions and approach
   Let amber revise its working logic and test the improved approach against the original goal.

This workflow is particularly useful for strategic decisions, workshop concepts, document and meeting analyses, project planning, offer preparation, and customer onboarding.

Conclusion

The most valuable question is not always:

“How do I write the perfect prompt?”

Often, these two questions are more effective:

1. “What do you need from me to guide me toward this goal?”

2. “How should your instructions be revised to achieve the expected result?”

The first makes missing context visible. The second turns mediocre results into a better process instead of an endless correction loop.

Try both questions the next time you work on a complex task with amber.