By feeding AI your strongest past proposals, you teach it the structure, tone, and logic that actually closes deals in your niche. The next proposal it generates mirrors your proven formula rather than defaulting to generic templates.
Few-shot prompting means showing an AI a few examples of the desired output format before asking it to generate new output. Instead of describing what you want ("Write a professional proposal"), you show examples of proposals you've written and ask the AI to follow that pattern. The AI learns from examples rather than instructions alone.
This is different from zero-shot prompting (no examples, just instructions) and one-shot (one example). Few-shot typically means 2–5 examples, though more can help for complex tasks.
Every freelancer has a voice, a structure, and a style. Your past proposals embody your positioning, your level of formality, your use of case studies, your pricing presentation approach. When you ask a generic AI to "write a proposal," it doesn't know these nuances. It generates something generic and professional but not distinctly you.
With few-shot prompting, you paste 2–3 of your best previous proposals and say: "I'm writing a proposal for a new prospect. Here are three examples of my work. Follow this pattern: structure, tone, level of detail, and how I position my value." The AI reverse-engineers your style from the examples and applies it to the new work.
Start by identifying 3–5 proposals you're genuinely proud of. These should be diverse in some ways (different industries, different service offerings) but consistent in your core approach. Save these as templates in a document or folder.
For each template, add a brief label: "Template A: 15-person B2B SaaS proposal with strong ROI focus," "Template B: Solo creator service proposal with emphasis on case studies," etc. This helps you select the right template for new scenarios.
When you have a new prospect, select the most relevant 2–3 templates. Paste them into your prompt with explicit instruction: "The examples below show my proposal style, structure, and tone. Write a proposal for [new prospect] that matches this approach."
Few-shot prompts sometimes need tweaking. If the AI's output is 80% right but misses a specific element (e.g., you always include a timeline in your proposals, but the AI didn't), add a note to your examples or explicitly highlight that element.
After you get solid output from a few-shot prompt, consider saving that prompt as your new template. This becomes your "golden prompt" for that type of work. You now have both style examples and a working prompt structure.
Some freelancers create a "meta-example"—a annotated proposal that explains why each section exists. "This opening paragraph names the specific problem mentioned in our discovery call. This middle section uses a case study from a similar company. This closing includes our standard timeline and next steps." When you show this annotated example alongside plain examples, the AI understands both the what and the why.
Few-shot prompting pairs powerfully with prompt chaining. You do research in step 1, then in step 2–3, you feed those insights into a few-shot prompt: "Here are three examples of my proposals. Here's the research I've done on [prospect]. Write a proposal matching my style."
It also pairs with context windows. Load all three examples plus the prospect research into one conversation, then run the few-shot prompt. The AI has total context to produce highly customized work that feels personally written.
Try this: Pull out three proposals you've actually written and landed. Paste them into a document. Write one sentence describing what makes each one your style (tone, structure, unique angles, emphasis). Then take a prospect brief for a current opportunity. Write two prompts: one zero-shot ("Write a proposal for this prospect") and one few-shot ("Here are three examples of my proposals. Write one for this prospect that matches this style"). Generate output with both. Compare. The few-shot version will feel far more like you.
Peri can explain this concept, give practical examples, help you decide whether it applies to your situation, or recommend a journey if appropriate.
Explore related journeys or tell Peri what you're working through.