Application Processing

Cell provides prompt builders for every stage of insurance application processing — from detecting application forms to filling completed PDFs.

Application detection

Classify whether a PDF is an insurance application form:

import { APPLICATION_CLASSIFY_PROMPT } from "@claritylabs-inc/cell";

// Use as a prompt with any model + the PDF
const response = await generateText({
  model: yourModel,
  messages: [{
    role: "user",
    content: [
      { type: "file", data: pdfBase64, mediaType: "application/pdf" },
      { type: "text", text: APPLICATION_CLASSIFY_PROMPT },
    ],
  }],
});

Field extraction

Extract all fillable fields from an application form:

import { buildFieldExtractionPrompt } from "@claritylabs-inc/cell";

const prompt = buildFieldExtractionPrompt();
// Returns a prompt that extracts fields as structured data:
// - text, numeric, currency, date, yes_no, table, declaration types

Auto-fill

Match extracted fields against known business context to pre-populate answers:

import { buildAutoFillPrompt } from "@claritylabs-inc/cell";

const prompt = buildAutoFillPrompt(fields, businessContext);

Question batching

Organize unfilled fields into topic-based batches for asking the user:

import { buildQuestionBatchPrompt } from "@claritylabs-inc/cell";

const prompt = buildQuestionBatchPrompt(unfilledFields);
// Groups fields by topic (e.g., "Company Information", "Loss History")

Answer parsing

Parse free-text replies back into structured field values:

import { buildAnswerParsingPrompt } from "@claritylabs-inc/cell";

const prompt = buildAnswerParsingPrompt(questions, userReply);

Email generation

Generate batch emails asking users for missing information:

import { buildBatchEmailGenerationPrompt } from "@claritylabs-inc/cell";

const prompt = buildBatchEmailGenerationPrompt(batch, context);

Reply intent classification

Determine what a user's reply to an application question means:

import { buildReplyIntentClassificationPrompt } from "@claritylabs-inc/cell";

const prompt = buildReplyIntentClassificationPrompt(userReply, currentBatch);

Field explanation

Generate explanations for complex insurance form fields:

import { buildFieldExplanationPrompt } from "@claritylabs-inc/cell";

const prompt = buildFieldExplanationPrompt(field);

Confirmation summary

Generate a summary of all filled fields for user confirmation before PDF generation:

import { buildConfirmationSummaryPrompt } from "@claritylabs-inc/cell";

const prompt = buildConfirmationSummaryPrompt(filledFields);

PDF mapping

Map filled field values back to PDF form coordinates for filling:

import { buildAcroFormMappingPrompt } from "@claritylabs-inc/cell";

const prompt = buildAcroFormMappingPrompt(filledFields, acroFormFields);
// Maps answers to AcroForm field names

import { buildFlatPdfMappingPrompt } from "@claritylabs-inc/cell";

const prompt = buildFlatPdfMappingPrompt(filledFields, pageInfo);
// Maps answers to percentage-based page coordinates

Lookup fill

Fill fields by looking up values from existing documents:

import { buildLookupFillPrompt } from "@claritylabs-inc/cell";

const prompt = buildLookupFillPrompt(fields, documents);