Guide
Prompt Engineering Crash Course for Government
This is just a quick reference for how to use a GPT model
Apr 2, 2024
Basic Prompt Engineering Guide for Government
(mostly acquisition professionals)
I. Introduction
Welcome to this comprehensive prompt engineering guide, tailored specifically for government contracting officers and acquisition program managers. This guide will help you leverage the power of advanced AI models like GPT-4, Claude, Bing Chat, and Bard to enhance your daily tasks, increase efficiency, and make data-driven decisions in the complex world of government contracting and acquisition.
II. Understanding Prompts and Their Components
In the context of AI models, a prompt is an instruction or input that guides the AI on what kind of text to generate. A well-crafted prompt consists of several essential components:
Context: Background information about the topic or task at hand. For government contracting, this might include details about the contract type, acquisition phase, or specific regulations relevant to the task.
Task or Question: A clear and concise statement of what you want the AI to do.
Constraints or Limitations: Any restrictions on the AI's output, such as word count, format, or specific requirements.
Additional Guidance, this can include:
Examples of desired output
Instructions on formatting
Style and tone requirements
Specific terminology or jargon to use
Relevant FAR citations or other regulatory references (verify anything that has legal implications)
Setting: The specific context of your agency or department, the type of acquisition, etc.
Characters: Relevant stakeholders or roles involved in the task.
Narrative: The background, current status, and desired outcome of the task.
III. Crafting Effective Prompts
To create effective prompts for government contracting and acquisition tasks:
Be as specific as possible about the task and desired outcome.
Use relevant keywords and terminology from the Federal Acquisition Regulation (FAR) and other applicable regulations.
Avoid ambiguous language that could lead to misinterpretation.
Provide necessary context about the specific contract, program, or acquisition.
Specify any required format or structure for the output.
IV. Example Prompts for Government Contracting Tasks
Here are five well-crafted prompts for generating a monthly status report outline for a government IT services contract:
IMPORTANT NOTE: If you want the GPT to reference and authoritative source, you should provide that source via upload and reference it. DON'T LET THE GPT HALLUCINATE, IT'S YOUR NAME ON THE LINE.
Prompt 1: Comprehensive Status Report
Context: You are the Contracting Officer's Representative (COR) for a $50 million, 5-year IT services contract with the Department of Defense. The contract is currently in its second year of performance.
Task: Generate a monthly status report outline for this IT services contract.
Constraints:
- The report should be no more than 10 pages long.
- Use standard government report formatting.
Include the following sections:
Executive Summary
Contract Performance Overview
Key Performance Indicators (KPIs)
Challenges and Risks
Corrective Actions
Upcoming Milestones
Financial Status
Subcontractor Performance
Change Requests
Conclusion
Additional Guidance:
- Reference relevant sections of the FAR where applicable.
- Include placeholders for quantitative data on KPIs.
- Ensure the outline allows for discussion of any deviations from the contract terms.
Prompt 2: Milestone-Focused Report
Context: As the Program Manager for a major IT infrastructure upgrade project for a federal agency, you need to prepare a monthly status report focusing on milestone achievements.
Task: Create an outline for a milestone-focused monthly status report.
Include the following sections:
Project Overview
Milestone Summary
Completed Milestones
Upcoming Milestones
Milestone Delays and Mitigation Strategies
Resource Allocation
Budget Status
Risks and Issues
Stakeholder Communications
Action Items
Additional Guidance:
- Include spaces for Gantt chart or timeline visualizations.
- Ensure the outline allows for tracking of both technical and programmatic milestones.
- Include references to relevant OMB guidance on IT project management.
Prompt 3: Performance-Based Acquisition Report
Context: You are overseeing a performance-based acquisition (PBA) for IT support services. The contract is structured with specific performance metrics and service level agreements (SLAs).
Task: Develop a monthly status report outline tailored for a performance-based IT support services contract.
Constraints:
- The report should be concise and data-driven.
- Focus on quantitative metrics and SLA compliance.
Include the following sections:
Executive Summary
Performance Metrics Overview
SLA Compliance Report
Customer Satisfaction Scores
Continuous Improvement Initiatives
Cost Savings and Efficiencies
Challenges and Resolutions
Upcoming Performance Targets
Recommendations
Additional Guidance:
- Include placeholders for charts or graphs to visualize performance data.
- Reference FAR Subpart 37.6 - Performance-Based Acquisition.
- Ensure the outline allows for tracking of both technical performance and business value delivered.
```
Prompt 4: Agile Development Status Report
Context: Your agency is using an agile development approach for a critical software development project. You need to create a status report that aligns with agile methodologies while meeting government reporting requirements.
Task: Create an agile-friendly monthly status report outline for a government software development project.
Include the following sections:
1. Sprint Summary
2. Backlog Status
3. Velocity and Burndown Charts
4. User Story Completion
5. Testing and Quality Metrics
6. Stakeholder Feedback
7. Risks and Impediments
8. Upcoming Sprint Goals
9. Resource Utilization
10. Budget Status
Additional Guidance:
- Incorporate agile terminology and concepts throughout the report.
- Include placeholders for agile artifacts like burndown charts and kanban board snapshots.
- Reference the TechFAR Handbook for agile contracting in government.
Prompt 5: Contract Close-out Report
Context: A major IT services contract is approaching its end date, and you need to prepare a comprehensive close-out report.
Task: Develop an outline for a contract close-out report for a major IT services contract.
Constraints:
- The report should cover the entire contract lifecycle.
- Ensure compliance with FAR Subpart 4.804 - Closeout of Contract Files.
Include the following sections:
1. Contract Overview
2. Performance Summary
3. Deliverables Assessment
4. Financial Reconciliation
5. Property Disposition
6. Subcontractor Close-out Status
7. Outstanding Claims or Disputes
8. Lessons Learned
9. Final Acceptance Statement
10. Archiving Plan
Additional Guidance:
- Include checklists for key close-out activities.
- Reference relevant DFARS clauses for defense contracts.
- Ensure the outline allows for documenting any open items requiring post-close-out action.
```
V. Importance of Good Prompts
The quality of your prompts directly impacts the quality of the AI's response. In the context of government contracting and acquisition (or whatever other work you do), where accuracy and compliance are paramount, well-crafted prompts are crucial.
Remember, while AI can provide valuable assistance, it's essential to review and verify all AI-generated content for accuracy, relevance, and compliance with current regulations and policies.
VI. Practical Applications in Government Contracting
AI models can assist with various tasks in government contracting and acquisition:
1. Market Research: Generate summaries of industry trends, potential vendors, or technology advancements.
2. Acquisition Planning: Draft sections of acquisition plans or develop evaluation criteria.
3. Solicitation Development: Create initial drafts of Statements of Work (SOW) or Performance Work Statements (PWS).
4. Proposal Evaluation: Summarize key points from proposals or generate comparison matrices.
5. Contract Administration: Draft modification requests, create status reports, or summarize contract performance data.
6. Close-out Activities: Generate close-out checklists or draft sections of close-out reports.
VII. Best Practices and Compliance Considerations
When using AI in government work:
ONLY USE AN AUTHORIZED TOOL.
Always verify AI-generated content for accuracy and compliance, particularly click through hyperlinks to ensure they work.
Be cautious with sensitive or classified information.
Use AI as a tool to augment, not replace, human judgment and expertise.
Stay updated on agency policies regarding AI use in contracting activities.
Maintain transparency about AI use when appropriate.
## VIII. Conclusion
Mastering prompt engineering for AI models can significantly enhance your effectiveness as a government contracting officer or acquisition program manager. As these AI technologies continue to evolve, your ability to craft effective prompts will become an increasingly valuable skill. Remember to practice regularly and stay informed about new developments in AI and their applications in government contracting.
Advanced GPT Prompting Techniques for Government
Introduction
As a government contracting officer or acquisition program manager, leveraging artificial intelligence can significantly enhance your efficiency and effectiveness. This comprehensive guide will walk you through progressively more complex GPT prompting techniques, tailored specifically for your role in government contracting and acquisition.
Precision in Prompt Design
When interacting with GPT models, clarity and specificity are crucial. Instead of vague requests, provide detailed instructions to improve response accuracy.
Example:
Summarize the key points of FAR Subpart 15.3 - Source Selection, focusing on evaluation factors and significant subfactors.
Role-Specific Prompts
Leverage the AI's ability to assume different roles to get more relevant responses.
Example:
Assume the role of a seasoned contracting officer. What are the top five considerations when evaluating past performance in a best value source selection?
3. Constraints on Prompt Size
Be aware of token limitations and craft your prompts accordingly to maximize the utility of the response.
Example:
Provide a concise overview (no more than 200 words) of the key differences between fixed-price and cost-reimbursement contracts.
4. Refining Queries
Use the AI to help refine your questions for more focused answers.
Example:
I need information about contract modifications. Can you suggest a more specific query that would be most helpful for a contracting officer dealing with a complex modification scenario?
5. Breaking Down Complexity
For multifaceted questions, break them down into smaller, more manageable queries.
Example:
I need to assess the risks in a major defense acquisition program. Please break this task down into specific areas to evaluate, such as technical risks, schedule risks, and cost risks. Then, provide questions I should ask for each area.
6. Kit Bashing
Combine different prompting techniques to create more specialized or complex prompts.
Example:
Assume the roles of both a contracting officer and a program manager. First, evaluate the following acquisition strategy for potential risks. Then, suggest mitigation strategies for each identified risk. Finally, provide a brief analysis of how these risks and mitigations might affect the overall program schedule and budget.
7. Flipped Interaction
Have the AI ask you questions to test your knowledge or generate new ideas.
Example:
Act as an experienced source selection official and ask me a series of questions to evaluate my understanding of the source selection process under FAR Part 15.
8. Multi-Persona
Generate responses from multiple perspectives to gain a more comprehensive understanding.
Example:
Generate a conversation between a contracting officer, a program manager, and a legal counsel discussing the pros and cons of using an Other Transaction Authority (OTA) for a major acquisition instead of traditional FAR-based contracting.
Ensure the cite the specific FAR and DFARs clauses, and explain in minute detail the pros and cons of each course of action.
9. Few-Shot Prompting
Provide examples to guide the AI in producing targeted and contextually appropriate responses.
Example:
Here are two examples of well-written justifications for other than full and open competition:
[Insert two brief examples]
Based on these examples, draft a justification for other than full and open competition for the following scenario:
[Insert your scenario]
10. Chain-of-Thought Prompting
Use a series of related prompts to dig deeper into a topic.
Example:
Q1: What are the key elements of an Acquisition Strategy?
Q2: How do these elements differ for major defense acquisition programs versus smaller acquisitions?
Q3: What are some common pitfalls in developing an Acquisition Strategy for a major defense acquisition program, and how can they be avoided?
11. Scenario-Based Prompting
Create detailed hypothetical scenarios to get more nuanced and practical advice.
Example:
You're a contracting officer faced with the following scenario: A major defense contractor has just informed you of a potential organizational conflict of interest midway through a source selection. Describe the steps you would take to address this issue, citing relevant FAR provisions and case law where applicable
12. Comparative Analysis Prompting
Use the AI to compare and contrast different approaches or interpretations.
Example:
Compare and contrast the approach to past performance evaluation in FAR Part 15 procurements versus FAR Part 8 orders under the GSA Schedules program. Highlight key differences and similarities, and discuss the pros and cons of each approach.
13. Decision Tree Prompting
Have the AI create decision trees for complex procurement scenarios.
Example:
Create a decision tree for determining the most appropriate contract type for a major systems acquisition. Include key decision points, potential outcomes, and relevant FAR citations for each branch of the tree.
14. Policy Impact Analysis
Use the AI to analyze how new policies or regulations might impact existing processes.
Example:
Analyze the potential impacts of the recently enacted Cybersecurity Maturity Model Certification (CMMC) requirements on the source selection process for defense contracts. How might this affect evaluation criteria, contractor responsibilities, and overall acquisition timelines?
15. Risk Assessment and Mitigation Planning
Leverage the AI to identify potential risks and develop mitigation strategies.
Example:
Identify potential risks in the following abbreviated acquisition plan for a major IT system upgrade. For each risk identified, suggest at least two possible mitigation strategies. Then, rank the risks in order of potential impact on program success.
[Insert abbreviated acquisition plan]
Advanced Multi-Stage Prompting
As you progress you will be able to chain together multiple techniques to build ever more complex promting chains.
Example:
Are you familiar with model-based system engineering?
—model response—
Good, now I want you to assume the role of a systems engineer and describe the process of building system models for a complex platform like the F35 joint strike fighter.
—model response—
Excellent, now I need to add a modification to a contract with a prime contractor, they are currently on task order under a FAR part 16.5 IDIQ with a cost plus fixed fee contract with several more option years, but I need to modify the contract statement of work to include model-based system engineering of their platform, but I'm not sure what all I need to include, so I want you to assume the role of an expert in government contracting and ask me questions for all the information you will need to know in order to write a complete and complaint contract mod, oce you have all the information you need, let me know.
— the model will likely say ok, and then ask a series of questions, then when it has enough info it will say so —
Great, now I need you to explain to me the approach for moding the contract, give me a comprehensive draft of the contract mod and then provide me step-by-step instructions for how we can go about completing the mod.
— the gpt will lay things out —
this seems like a great approach, but I need a second opinion, what I need you to do now is take this approach, and assume the role of an acquisition program manager for this program, a contracts legal expert, and an expert contracting officer, and I want them to discuss the approach, identify the pros and any cons in the approach, and identify one or two viable alternatives that would be equally or nearly as good, perhaps with different advantages. At the end I would like them to reach a consensus on the best value approach to the modification.
— the three will talk back and forth for a while, likely come up with an answer —
great, now please draft me a final contract modification using the following format: <insert the headings for your particular format>
