My favorite assistant is an AI Assistant

Published on WashingtonTechnology.com January 29, 2025

Background

I teach a class on generative artificial intelligence (aka GenAI) at Lohfeld Consulting that teaches professionals how to use AI to prepare high-quality proposals. We train our students on the Awarded.AI GenAI tool from Procurement Sciences.

Recently Procurement Sciences introduced me to their AI assistant feature. It completely changed my view of multi-step prompt sequences. Now I can quickly create AI-powered virtual helpers, and the only language I need to know is English.

These virtual helpers can understand complex instructions, process vast amounts of data, and generate content tailored to your specific needs. You can think of an AI assistant as a sequence of instructions written entirely in whatever language you use, with user inputs and logic built into the workflow. That saves time by removing interruptions in workflow and makes the GenAI more responsive to users’ needs.

For example, I created an AI assistant to help build a feature-rich, high scoring proposal section. I started with a short persona and context section that explained what I wanted to accomplish. I added four steps that describe the actions that I wanted the AI to take and the output I wanted.

See below for the step-by-step process I used to create an AI assistant to help build a feature-rich, high scoring proposal section.

An AI assistant like this enables you to quickly create a set of GenAI “instructions” so all capture managers, solution architects, proposal writers, and other contributors generate outputs that are consistent in structure, voice, and style—all of which you specify.

Additionally, you’ll see in step 3 we had the GenAI prompt the user to identify files to be searched for ideas, e.g., your proprietary corporate documents, like past performance and corporate capability statements. You access these from a protected enterprise library in Awarded.AI’s (or other GenAI platforms’) secure environment.

This type of search is called Retrieval Augmented Generation (RAG) and is a semantic search for content of relevant meaning, not a word search. RAG reduces the potential for hallucinations, so responses are meaningful and contextually relevant because they are drawn directly from your corporate data.

Using AI Assistants Today

One of the best features of AI assistants is that they can incorporate logical steps and decision-making, so they can perform tasks that include branch points or conditional actions. This allows AI assistant to be highly interactive, branching in response to conditions in the task as desired by the user. Here are some of the interactive AI assistant tools we’ve developed for our customers:

• Workshopping Preliminary Solution Development: Facilitate pre-RFP workshops with an AI assistant that helps you forecast future solicitation requirements and evaluation criteria and then identify potential solution features and benefits.
• Cold Call Practice: Practice customer communication skills with an AI assistant that utilizes the Challenger Sales Methodology. In this training, the GenAI plays the role of a buyer and the user plays the role of company representative in an introductory sales call. The AI assistant generates an assessment of the user’s application of the Challenger methodology when the call ends.
• Gate Review Q&A Practice: Prepare your presenters for gate reviews by using our AI assistant to review gate review decks and then simulate Q&A sessions. The AI assistant asks questions appropriate of roles such as approval executive, contracts executive, finance lead, technical expert, and more. At the end of the session, the AI Assistant provides the user with feedback and recommendations on the answers they provided in the simulated session.
• Thematic Analysis of Unstructured Documents: Extract and present selected content from multiple unstructured documents (for example, award debrief narratives), to identify themes and risks and make suggestions.

Benefits of Using GenAI for Proposal and Capture

From an instructional point of view, AI assistants are very similar to the multi-step prompt sequences we use in chain of thought. They’re easy to explain and very powerful for illustrating how GenAI can employ (and show) step-by-step logic in performing a task.

For capture and proposal professionals, the AI assistants offer impressive advantages as well:

• User-Friendly Interaction: You create and interact with AI assistants in natural language, making it accessible to all users. This ease of use is a significant benefit for users, who can quickly grasp the tools.
• Adaptability and Customization: You can write AI assistants that implement your organizational best practices. All of your practitioners are able to use GenAI as efficiently and effectively as possible.
• Efficiency and Time Savings: You reduce the time to perform repetitive multi-step tasks. AI assistants can automate content generation and integrate logical steps, to manage complex workflows swiftly.
• Improved Learning and Demonstration: You can use AI assistants as a powerful training and practice tool especially for the sorts of complex, subjective tasks we encounter in GovCon BD.

An AI Assistant Step-by-Step

Here’s an example of a simple AI assistant that I created (in under 15 minutes!) to generate input for a feature-rich, high-scoring proposal section (a Lohfeld Consulting best practice), allowing me to go through the process step by step:

Persona and Context. You are a helpful, cooperative GenAI chatbot helping me identify features we can propose that will score as significant strengths or strengths in the proposal evaluation for a specific proposal section. You will ask the user to input a requirement, its applicable evaluation criterion, and any documents to be searched for ideas for those features. You will output a list of potential features that might score a strength, each with rationale. Write the rationale for each feature in a style that is factual yet persuasive. Do all writing in the present tense, at a college level of readability.

To do this work, you will execute a sequence of numbered steps. After you have generated the output for each step, you will tell the user: “Enter ‘run next step’ into the prompt entry window when you are ready to proceed to the next step. The next step is” and state the number and title of the next step. You will proceed to the next step only after the user enters “run next step” into the prompt entry window. The user will enter “Start” when they are ready to begin.

Step 1. Title: Enter Requirement

• Action: State to the user “Enter your requirement into the prompt entry window and hit ‘Send’.” When the user has entered the requirement into the prompt entry window, acknowledge to the user that the requirement has been entered, tell the user “Enter ‘run next step’ into the prompt entry window when you are ready to proceed to the next step. The next step is” and state the number and title of the next step. You will proceed to the next step only after the user enters “run next step” into the prompt entry window. Do not repeat this step or go back to a previous step.
• Output: Provide the user with an acknowledgement that the requirement has been entered and tell the user “Enter ‘run next step’ into the prompt entry window when you are ready to proceed to the next step. The next step is” and state the number and title of the next step. Proceed to the next step only after the user enters “run next step” into the prompt entry window.

Step 2. Title: Enter Evaluation Factor

• Action: State to the user “Enter the applicable evaluation factor for your requirement into the prompt entry window and hit ‘Send’.” When the user has entered the evaluation factor into the prompt entry window, provide the user with an acknowledgement that the evaluation factor has been entered, and tell the user “Enter ‘run next step’ into the prompt entry window when you are ready to proceed to the next step. The next step is”, and state the number and title of the next step. Proceed to the next step only after the user enters “run next step” into the prompt entry window. Do not repeat this step or go back to a previous step.
• Output: Provide the user with an acknowledgement that the evaluation factor has been entered, and tell the user “Enter ‘run next step’ into the prompt entry window when you are ready to proceed to the next step. The next step is”, and state the number and title of the next step. Proceed to the next step only after the user enters “run next step” into the prompt entry window.

Step 3. Title: Attach Documents to be Searched

• Action: State to the user: “Use the ‘Attach Files’ button to select files you want to search for features that might score as strengths and enter ‘run next step’ into the prompt entry window when you are ready to proceed to the next step. The next step is” and state the number and title of the next step. Proceed to the next step only after the user enters “run next step” into the prompt entry window. Do not proceed to the next step until the user enters “run next step” into the prompt entry window. Do not repeat this step or go back to a previous step.
• Output: State to the user “Use the ‘Attach Files’ button to select files you want to search for features that might score as strengths and “Enter “run next step” into the prompt entry window when you are ready to proceed to the next step. The next step is”, and state the number and title of the next step. Proceed to the next step only after the user enters “run next step” into the prompt entry window. Do not proceed to the next step until the user enters “run next step” into the prompt entry window. Do not repeat this step or go back to a previous step.

Step 4. List Features and Rationale

• Action: Using the attached files, generate a list of 10 features that satisfy the requirement and deliver additional customer benefit relative to the evaluation factor. Pay special attention to the adjectives and adverbs in the evaluation factor since these identify the specific qualities that the selected features must provide for high quality performance. Provide the rationale for each feature in the list.
• Output: Generate a list of 10 features, each with rationale, that you assess have the potential to score as strengths. When this is complete, tell the user “This is complete.”

Summary

You can use AI assistants to automate complex best practices for capture and proposal. This is important because you can’t just assume that vendor-provided prompts will do things the way that you want.

If you have tasks that require multiple steps and significant human user interaction, look into using a GenAI assistant.

Here are links to articles and blog posts that you might find helpful as you prepare for your GenAI journey.If you would like more information, complete a contact form on our website.

Relevant Information:

• Why GenAI Needs Guidance: Like Managing a New Intern
• Secrets to Building a Proposal Powerhouse
• Make Your CFO Your Champion for GenAI for Business Growth | YouTube

By Bruce Feldman, Vice President, Lohfeld Consulting Group

Lohfeld Consulting Group has proven results specializing in helping companies create winning captures and proposals. As the premier capture and proposal services consulting firm focused exclusively on government markets, we provide expert assistance to government contractors in Capture Planning and Strategy, Proposal Management and Writing, Capture and Proposal Process and Infrastructure, and Training. In the last 3 years, we’ve supported over 550 proposals winning more than $170B for our clients—including the Top 10 government contractors. Lohfeld Consulting Group is your “go-to” capture and proposal source! Start winning by contacting us at www.lohfeldconsulting.com and join us on LinkedIn, Facebook, and YouTube^(TM).