From email to assignment: AI for consistent surveying requests
In Brief
AI that simplifies high-volume request processing
AllPoints Surveying is a land surveying company that receives a high volume of customer requests by email. These requests often arrive in inconsistent formats with missing or unclear details, requiring coordinators to review and interpret each one before routing and assigning work to regional teams.
Resource Data developed and tested a proof-of-concept AI system that reads incoming emails, extracts key information, and matches it to internal records. The system reached 94% accuracy in builder identification and reduced the time required to review and route requests, showing it can support higher volumes without adding manual workload.
Key Takeaways
Standardizing how incoming work is translated into action
-
Faster routing with 94% accurate builder identification
The system identifies builders from unstructured emails and matches them to internal records, so requests go to the right regional coordinator with less manual checking.
-
Structured request data replaces manual interpretation
Coordinators review pre-processed request data and confirm details instead of reading and interpreting each email from scratch.
-
Consistent outputs for faster review
Requests are formatted the same way every time, so coordinators can scan and validate them quickly without piecing together information from raw emails.
-
Handles incomplete and inconsistent requests
The system extracts builder names, addresses, and service types even when emails are partial or informal. This cuts follow-up emails and reduces delays.
-
Translates client language into internal task names
Client descriptions are mapped to standardized task types, so teams can assign the right survey without interpreting each request.
-
Support to higher volume without adding staff
By handling extraction and routing, the system lets the team take on more requests across regions as the company grows, without increasing headcount.
Meet Our Client
AllPoints Surveying
AllPoints Surveying provides land surveying services for residential and commercial development projects across multiple states, including Texas and Florida. The company handles a high volume of incoming land surveying requests each day, with coordinators processing approximately 50–60 emails per hour.
Incoming requests are reviewed by AllPoints teams to identify key details, confirm project information, and assign work to the appropriate regional team. This centralized intake process supports operations across multiple markets and keeps projects moving quickly from request to scheduling.
The Challenge
Manual email review created delays and inconsistent results
AllPoints receives customer requests by email, often with inconsistent formats, additional attachments and incomplete details. Some requests include full project information while others contain only a short description or partial address. Clients often describe services using their own terminology, which does not always match internal task names requiring coordinators to interpret and translate each request before assignment. As a result, coordinators must manually review each message to identify the client, confirm the location, and determine the type of survey requested.
As the company grew, request volume increased. The request intake process took more time and became harder to manage. Incomplete or unclear information led to follow-up emails or incorrect assignments. With each request requiring manual interpretation before routing, limits were reached for how quickly new work could be accurately scheduled.
The Solution
Extracting, matching, and standardizing into usable data
Resource Data developed an AI enabled solution that reads incoming customer emails and extracts the information needed to assign work. The solution interprets unstructured content and identifies key details such as builder name, project address, and requested service.
Extracted addresses are standardized and matched against internal records, including MasterJob data. Builder names are compared to existing customer data to support routing to the correct regional coordinator. To bridge differences in terminology, the system uses a reference dictionary that maps how clients describe services to standard internal task names. The system outputs structured data in a consistent format giving coordinators a clear starting point for review and assignment.
Features
Turning guesswork into consistent request handling
-
Email parsing for clean and usable input
Processes full email threads and removes signatures, disclaimers, and repeated content so extracted data comes from the actual request and is not confused by “noise”.
-
Context extraction for identifying request details
Interprets unstructured messages to identify builders, addresses, and service requests, even when information is incomplete or written informally.
-
Address normalization for reliable record matching
Standardizes address formats and applies exact and approximate matching to connect requests with existing records and reduce duplicates.
-
Builder matching for consistent request routing
Compares extracted builder information to internal customer records to support accurate assignment and reduce manual lookup.
-
Task translation for standardized job classification
Converts client terminology into defined internal task names using a reference dictionary, reducing interpretation work for coordinators.
-
Clear outputs for efficient human review
Presents extracted and matched information in a consistent format so coordinators can review and confirm requests without interpreting each email from scratch.
If every request need interpretation, scaling becomes difficult. Making that step consistent changes the equation.
- Jan Issabekov, Sr. Data Engineer, Resource Data
Results
Fewer back-and-forth emails, faster scheduling
The system identified builders with 94% accuracy, improving how requests are routed to the correct regional coordinator and reducing the need for manual verification. Task identification reached about 68% accuracy, giving coordinators a consistent starting point to confirm requests instead of interpreting each email from scratch.
By handling data extraction and matching, the system reduced the time spent reviewing emails and helped teams process incoming requests more efficiently. This made request handling more consistent across regions and enabled the team to manage higher volumes without increasing staffing levels.
What's Next
Expanding the system for broader automation and integration
The next phase focuses on improving task identification accuracy and integrating the system into production workflows. Future features would process attachments such as site plans and supporting more complex request scenarios.
