We’re introducing Examples – a feature that lets Instafill.ai replicate your voice and tone when filling out forms.
Many forms contain fields that require generated text – descriptions, narratives, summaries – based on your source data (call transcripts, documents, reports). Our AI fills out these fields accurately, but the writing style is generic. Examples changes this: you provide samples of how you’d write those fields, and our AI replicates your voice automatically.
The problem
Complex forms – healthcare documentation, construction permits, legal contracts – often contain fields that need more than simple data entry. These fields require generated text: behavioral observations from therapy sessions, project descriptions from site reports, incident narratives from call transcripts.
Even when our AI fills out these fields correctly with accurate information, the text often looks unnatural. Common telltale signs include:
- Overly formal or stilted phrasing that doesn’t match how humans actually write
- Excessive use of em dashes (—) where commas or periods would be more natural
- Generic vocabulary that sounds like it came from a template
- Repetitive sentence structures that lack variation
- Unnatural transitions between ideas
For professionals who need documents to sound authentic and human-written – whether for client-facing materials, legal submissions, or medical documentation – this means extra time editing after each form is completed. The information is correct, but the voice is wrong.
Examples solves this fundamental problem.
What Examples does
Instead of writing detailed instructions about tone and style, you provide real examples of your writing for key fields. Our AI analyzes your tone, vocabulary, sentence structure, and formatting – then replicates your voice when generating text in every form.
The result is text that sounds like you wrote it, not an AI. Your specific phrasing patterns, your preferred terminology, your writing rhythm – all preserved automatically.
How Examples works
From a technical perspective, Examples uses few-shot prompting. When filling out a field with examples, our AI processes it with higher attention, taking into account your reference text to understand exactly what output you expect.
What the AI analyzes
When you provide examples, our AI studies multiple dimensions of your writing:
- Tone and formality level: Professional vs conversational, direct vs diplomatic
- Vocabulary choices and phrasing patterns: Technical terms, industry jargon, preferred synonyms
- Sentence structure: Length, complexity, use of compound sentences
- Formatting preferences: Punctuation style, use of lists or paragraphs
This analysis happens automatically. You don’t need to explain your style – the AI learns it from your examples.
Step-by-step: Adding examples to fields
Step 1: Open your form in the Field Editor. You can access the Field Editor from the form details page after fine-tuning is complete.
Step 2: Click on any field you want to customize. The “Edit field” modal will open.
Step 3: Navigate to the Examples tab. You’ll see three tabs: General, Dependencies, and Examples.
Step 4: Add your example text. Paste or type examples of how you’d write this specific field. Each example should represent well-written content that matches your voice.
Step 5: Add multiple examples if needed. Click “Add example” to include additional reference text. There’s no limit – the more examples you provide, the better the AI understands your voice and tone.
Step 6: Save your changes. Click “Save” to apply the examples to the field.
Step 7: Test the results. Fill the form with new data and review how the AI replicates your style when generating text for that field.
Fields with vs without examples: Fields without examples still fill accurately using our core AI. Fields with examples fill with your specific voice, making the generated text indistinguishable from text you wrote yourself.
When to use Examples
Examples is valuable whenever authentic, human-written text matters in generated fields. Here are detailed scenarios where Examples makes a measurable difference:
ABA therapy and medical documentation
The challenge: Clinical reports contain numerous fields requiring generated text – behavioral observations, progress notes, assessment summaries. When filling these from session transcripts or raw data, AI-generated clinical language often sounds generic and templated, lacking the natural voice that comes from experienced practitioners.
How Examples helps: Add examples of well-written behavioral observations, assessment summaries, and progress reports from your actual documentation. The AI learns your clinical writing voice – how you describe behaviors, reference treatment goals, and document interventions. When generating text from new session transcripts, every field sounds natural and authentic, matching the voice your team uses in all documentation.
Result: Reports are ready for submission immediately, with text that sounds like it was written by your clinical team, not automated software.
Construction permits and compliance forms
The challenge: Permit applications contain fields requiring generated technical descriptions – scope of work, project specifications, material descriptions. When generating these from project data or site reports, generic AI text lacks the specific terminology and authoritative phrasing patterns that your company uses.
How Examples helps: Provide examples of approved permit descriptions, scope statements, and specifications from your previous successful submissions. The AI learns how you structure technical descriptions, reference building codes, and describe construction methods. When generating text for new permits, each field maintains your company’s specific voice and terminology.
Result: Permits read like they were written by your team, using the exact language and format that your jurisdiction expects and approves.
Legal documents and contracts
The challenge: Legal documents contain fields requiring generated narrative text – incident descriptions, factual summaries, procedural explanations. AI-generated legal text often exhibits telltale signs: excessive formality, repetitive phrasing, or unnatural sentence structure that experienced attorneys immediately recognize.
How Examples helps: Add examples of key narrative sections, factual summaries, and explanatory text from your firm’s approved documents. The AI learns your firm’s writing conventions, preferred phrasing patterns, and formatting standards. When generating text for new documents, output maintains consistent voice and avoids the artificial quality of generic AI.
Result: Documents reflect your firm’s standards automatically, with language that sounds like it came from your attorneys, not automated software.
Additional use cases
Any workflow where authentic, human-written generated text matters benefits from Examples:
- Insurance claims: Generate incident descriptions and damage assessments from adjuster notes while maintaining your company’s voice
- Grant applications: Generate narrative sections from research data while preserving your organization’s writing style
- Real estate contracts: Generate property descriptions and disclosures with your firm’s terminology
- HR documentation: Generate performance reviews and incident reports from notes while keeping your company’s consistent tone
- Financial reports: Generate analyses and summaries from data with uniform voice
Examples vs. field descriptions: When to use each
Both Examples and field descriptions are powerful tools for customizing form filling. They work best together, but serve different purposes:
| Scenario | Best approach |
|---|---|
| Specify a required format (dates, phone numbers) | Field description |
| Explain conditional logic or special rules | Field description |
| Define tone, writing style, or narrative structure | Examples |
| Show how to structure long-form generated text | Examples |
| Demonstrate specific phrasing or terminology | Examples |
| Maintain consistent voice across similar fields | Examples |
| Avoid AI-sounding generated text | Examples |
Using both together: For best results, combine field descriptions with Examples. Use descriptions to explain rules and constraints (“Use MM/DD/YYYY format” or “Skip if applicant is under 18”), and Examples to demonstrate voice and style for generated content. The AI considers both when filling fields.
Why Examples works better than instructions
You might wonder: why is showing examples more effective than writing detailed instructions about tone and style?
1. Humans struggle to describe their own voice
Most people can’t accurately describe how they write. They might say “professional but friendly” or “clear and concise,” but these descriptions are subjective and don’t capture the specific patterns in their actual writing. Examples let the AI analyze your real writing, not your description of it.
2. Examples provide concrete patterns
When you say “use a conversational tone,” different people interpret that differently. When you show three examples of your actual writing, the AI sees exactly what “conversational” means to you – your specific vocabulary choices, sentence lengths, and phrasing patterns.
3. Multiple dimensions analyzed simultaneously
A single example contains information about tone, vocabulary, structure, formatting, and more. The AI extracts all these dimensions at once, understanding how they work together in your voice. Instructions would require you to explicitly describe each dimension separately.
Technical insights: What makes Examples sophisticated
Examples might look like a simple UI feature – just text boxes where you paste content. But underneath, our AI is doing sophisticated analysis.
Individual field processing
Standard fields are filled in batches for efficiency. Fields with Examples are processed individually when generating text, allowing the AI to focus exclusively on matching your reference style. This higher attention produces more accurate voice replication.
Pattern recognition across examples
When you provide multiple examples, the AI identifies patterns that appear consistently across them. This helps distinguish between your general voice (which should be replicated) and specific content details (which are unique to each example).
Context-aware application
The AI doesn’t just copy your examples – it applies your voice to new content. It learns your style, then uses that style to express whatever information needs to go in the field. The vocabulary and phrasing match your voice, but the facts come from your input data.
Current availability and access
Examples is currently in beta. We’re setting it up manually for each user to ensure the best experience.
Field Editor access required: To use Examples, you need Field Editor access enabled for your account. It’s public and can be enabled for any account, but currently requires activation on our side. Contact us at [email protected], and we’ll activate it for your account and guide you through adding your first examples.
Getting started
To begin using Examples:
- Contact [email protected] to request Field Editor access
- We’ll enable the feature for your account
- Open any fine-tuned form
- Navigate to the “Form fields” tab and choose “Edit” in the needed fields
- Add your examples in the Examples tab
- Test with real data to verify results
If you have questions about setup or want guidance on which fields benefit most from Examples, our team is available to help.
Roadmap: What’s coming next
Automatic example extraction: Currently, you add examples manually by copying text into the Examples tab. We’re building automatic example extraction: upload previously completed forms (PDFs, Word documents, even handwritten documents), and Instafill.ai will automatically extract field values as examples. One click, and your entire form learns your voice. This feature is in development and will be released soon.
Best practices for using Examples
1. Choose representative examples
Pick examples that truly represent how you want fields filled. Don’t use examples that were rushed or don’t match your usual quality. The AI will replicate whatever you show it.
2. Use 2-4 examples per field
While there’s no limit, 2-4 well-chosen examples are usually sufficient. More examples help the AI identify consistent patterns, but beyond 4-5, you see diminishing returns.
3. Vary your examples slightly
If all your examples are nearly identical, the AI might replicate too closely. Show slight variation in how you express similar ideas – this helps the AI learn your voice rather than memorizing specific phrases.
4. Test and refine
After adding examples, test the form with real data. Review the generated output to ensure the AI is replicating your voice accurately. You can always add more examples or edit existing ones to improve results.
5. Focus on fields requiring generated text
Not every field needs examples. Name fields, dates, and addresses don’t benefit from voice replication – they’re simple data entry. Focus examples on narrative fields, descriptions, explanations, and any text where tone and phrasing matter.
Why this matters
Examples represents the difference between a tool that fills out forms and a platform that understands professional document workflows.
Examples might look like a simple UI feature, but underneath, our AI is doing sophisticated analysis of your writing. It identifies patterns in tone, vocabulary, sentence structure, and formatting – then applies those patterns consistently when generating text across forms. This level of voice replication isn’t possible with simple field instructions or templates.
When you work with AI form filler, you’re working with a platform that understands professional document workflows deeply enough to replicate human writing, not just fill in blanks. That’s the difference between a mature product and a weekend project.
Ready to use Examples?
Get Field Editor access and start teaching Instafill.ai your voice today. Contact us at [email protected] to get started.