AI Inquiry Analysis

This workbench identifies email types, organizes buyer needs, customer-provided info, missing fields, and follow-up advice. It evaluates business fit, purchasing readiness, and recurring intent separately with source quotes. The English draft reply is a final reference step to aid sales rep communication and will not send emails automatically.

You can receive follow-up stars by pasting an inquiry alone; seller background is optional. Cited product requirements and a purchasing stage support an initial procurement-signal preview: clear recurring demand or a forecast generally earns four stars, while a specific request earns three. Business fit remains unconfirmed; three stars does not mean low customer value. Optional seller context refines fit. Five stars require strong fit, clear procurement steps, cited high priority and recurring demand. One or two stars require evidence of weak fit. Missing effective purchasing requirements or stage remains unrated. Account scale is separate from the stars: forecasts are not orders, and samples do not establish company size, order value or conversion probability.

A single inquiry cannot verify legal entities, business scale, payment capacity, or real purchasing authority. Statement excerpts in the report remain unverified independently, and analytical inferences require human confirmation. You can voluntarily provide public websites to expand accessible public materials; self-stated website content still does not prove identity, credit, or financial standing. Further verification is required before transaction decisions.

No. Public website research is turned off by default; the tool does not automatically visit URLs found in emails. Only when you actively enable research and enter a URL will the server fetch a limited scope of public pages. The report displays actual fetched sources and coverage; incomplete, partial, or unavailable sites will be clearly stated, rather than treating domain guesses as investigated facts.

Providing verified products, applications, supported terms, and target customer types makes fit analysis more grounded. Known constraints are ideal for certification, capacity, MOQ, or acceptable trade terms, but avoid entering unverified promises. Without sufficient seller context, the tool marks match fit as unknown and lists questions needing clarification.

Expand references beside each judgment to view sources from inquiries, seller background, or public websites. Original Statements indicate that provided materials state these points, which does not equal real-world verification; Analytical Inferences indicate judgments made by the model based on materials; Pending Confirmation indicates insufficient information. High, medium, and low confidence levels of the model are not calibrated probabilities and should still be checked alongside citations.

Such judgments cannot rely solely on email length, provider, or country. Short emails may come from repeat inquiries or genuine procurement requests; reports should differentiate based on purchasing requests, context, and specific evidence. Classifications like spam, marketing, or cold pitch also require justification. Keep uncertain items as pending verification; automatic deletion of emails is not recommended.

Drafts require verification by sales representatives. The tool generates responses based on verified facts; unconfirmed prices, certifications, model numbers, payment terms, and lead times should be phrased as 'to be confirmed', prioritizing inquiries about critical missing details. After copying, verify contact details, product specifications, tone, and commercial commitments before sending. This page does not connect to email services and will not automatically send replies to clients.

Inquiries and seller details are submitted to the AI model service only after clicking "Analyze with Gemini". Website research requires manual activation. The page does not automatically save drafts or access email accounts. Use local sanitized previews to replace common email addresses and phone numbers with + country codes. Sanitization cannot catch all personal details, company names, or trade secrets; manually review text before submitting.

The page will clearly display failures or quota limits without using template results to fake completed AI analysis. You can retry or download the analysis prompt composed of your current inputs for model processing on your end. Prompt downloads are generated locally, contain your entered or anonymized materials, and do not imply an AI API call has been made.

You can copy requirements, customer information, qualification questions, or English drafts, and also download JSON, Markdown, and TXT reports. Reports retain the model name, generation time, cited sources, and inspection boundaries for easy review by colleagues. Files will contain analysis materials and customer information; please save and share according to team permissions. This tool will not automatically share or send data to others.