I have a working system: Telegram bot saves customer data to Google Sheets (email, payment type, zelle) and context/rules to Qdrant. Email Assistant receives emails, finds customer in Sheets, searches Qdrant for relevant knowledge, and creates Gmail draft replies.
I have fixed scripts for different situations (new order prepay/postpay, tracking questions, payment questions).
Should I: store scripts in a separate Google Sheets table and substitute without LLM generation, or is there a better approach?
Can share n8n workflow JSON if helpful.
I’m looking for someone here who can assess whether I’ve chosen the right implementation path. Maybe there’s someone here I can tell more about my business with so we can discuss the architecture together!
Hey, if you’re interested, I offer a 30 min free strategy call: Book a Call - SmoothWork, we can definitely look at a high level overview of your system and see what suggestions I would have.
Thank you very much for your comment! I was counting on the community’s help and am not ready to hire you as an employee or your team, if I understand correctly how this community works. Sorry.
you’re asking exactly the right kind of question for this community.
Short answer: I wouldn’t choose between “templates only” or “LLM only”. The most robust approach in e-commerce is usually hybrid.
!- Keep your fixed scripts/templates as the source of truth for each scenario (prepay, postpay, tracking, payment issues, etc.). Use the LLM only as a language layer, not as the decision maker — it should adapt tone, reference the incoming email, and stitch together data from Sheets + Qdrant, but it should not invent policies, dates, or rules.
2 -Classify the email (using rules or a lightweight LLM), select the correct template, fill in structured data (customer, order, payment), pull supporting context from Qdrant (policies, FAQs), let the LLM rewrite the final reply in natural language, then create the Gmail draft.
This gives you consistency, lower hallucination risk, easier maintenance, and safer scaling.
3- Google Sheets is fine early on (fast iteration, easy edits). As complexity grows, teams usually move templates to versioned files (repo) or a DB with explicit versioning. Sheets isn’t “wrong” — just be aware of governance later
Thank you very much for your detailed answer! It helped me decide how to develop the automation! I chose Google Sheets because they’re easy to edit and I don’t have a lot of scripts!
I’ll try doing just that. I’ll move all the scripts to Google Sheets and specify all the triggers for them, and let LLM decide which script is best.
