From enquiry to an approved quote
Combine AI interpretation, established pricing logic and human judgment.
What is difficult today?
A service operator reads an enquiry, extracts requirements, checks the price and rewrites the same information into another system. Missing details lead to back-and-forth messages. The goal is not for AI to invent the business terms; it is to prepare consistent work for a permitted decision.
The delivered module extracts a proposed request, flags missing data and calls the approved pricing function. A person reviews exceptions and the exact quote version. The application creates the record and sends it only under the defined approval policy, using the customer’s own AI connection.
A sequence with accountable decisions.
The owner defines the target. The engineering work keeps the relevant business rules and returns evidence at each acceptance boundary.
Receive
Keep the enquiry and allowed documents.
Understand
Extract fields and flag ambiguity without inventing values.
Calculate
Apply the approved price rules and version.
Review
Approve exceptions, capacity and the quote revision.
Complete
Create the record and send once when permitted.
Data, rules and proof—not only screens.
Data involved
- Minimized enquiry content and allowed attachments.
- Customer identity and a versioned price catalog.
Rules to preserve
- AI does not invent the price or promised deadline.
- An approval applies only to its quote revision.
- Customer runtime AI is paid directly to the provider.
Evidence to inspect
- Complete and incomplete enquiries.
- Unsupported service or price missing.
- Stale approval after quote edit.
- Duplicate incoming event and attempted double send.
Owner decisions
- May standard quotes be sent automatically?
- Which exceptions always need a person?
AI interprets. Business rules calculate. People decide.
The first input is an enquiry, not a perfect database record. Documents may be missing or a delivery date may be ambiguous. The AI module extracts candidate fields and records what still needs clarification. Existing pricing functions calculate an amount from approved inputs. Do not ask the model to invent a price because the source information looks plausible. An unusual deadline or exception can require a named human approver.
Define the boundary between a draft and a sent offer. The application creates a versioned draft, lets a person review it and sends only after the configured permission is present. Retries should not create duplicate offers or send the same message twice. Tests include missing attachments, contradictory details, provider failure, duplicate events and an unauthorized send. The deployed application stores its own provider keys server-side and continues to own the workflow. Yepsy engineers the capability; it does not become the company’s quotation department.
Follow a related example from start to release.
A worked example showing decisions, implementation and verification.
The video could not be loaded. Try again later.
Prepare the next decision.
An API makes software reachable. A workflow makes AI useful
Separate technical connectivity from a controlled business process that reaches a defined result.
Read the guide →From enquiry to draft quote: a bounded AI workflow
Use AI to interpret the request without letting it invent prices, capacity or permission to send.
Read the guide →Classify documents without turning uncertainty into a wrong record
Design an extraction workflow with source evidence, clear categories and a real review queue.
Read the guide →Start with a product.
Keep building a business.
Bring an idea or the software you already own. Give the next improvement a clear path from intent to evidence.