Contact person: Business Client

Phone:Show

Email:Show

Location: Palangkaraya, Indonesia

Budget: Recommended by industry experts

Time to start: As soon as possible

Project description:
"I already have a small library of “known” general-inquiry prompts—right now it’s just one example (“Can you say it aloud?”) paired with its approved answer. Whenever an incoming user message is substantially similar to one of these stored prompts, I want the system to reply with the matching response automatically. If the message is about a document the user has just uploaded, the system must instead summarize the document rather than refusing. Every other message should flow through to the normal reply logic I already have.

Here’s what I need from you:
• A lightweight Python module (open to another language if you feel strongly) that receives an incoming string, checks it against my growing list of general information-request prompts, and returns the corresponding canned answer when similarity crosses an adjustable threshold.
• Reliable similarity matching—semantic search or embeddings via libraries such as spaCy, Sentence-Transformers, or OpenAI embeddings are all fine as long as you justify the choice and keep the setup simple.
• A separate check that detects references to “the document I just uploaded,” “that file,” etc., and routes those messages to a placeholder summarize() function instead of the refusal text.
• Clear, commented code plus a short README explaining how to extend the prompt list and tweak the similarity threshold.

I’ll test by passing a set of close variants along with a few document-related queries; acceptance is meeting ≥95 % correct routing in that sample. Let me know if any part of the spec needs clarification." (client-provided description)