A Cloud-based Agro-advisory Service Information System (CASIS) uses interactive operating mode where assignment of questions from farmers to experts is done manually. Questions as input to the system are received randomly in a day and experts are supposed to respond within a specified time. The system has 20 experts in its database who respond to farmers questions and it can receive more than 30 questions per day. If there is a significant delay in the responses to a question then the question is reassigned to another expert. Each expert behaves differently when responding to their assigned questions. In order to address the shortcomings, the experts’ assignment algorithm was developed utilizing the respondents’ response probabilities and time of responses. Assignment decision is based on using a model that trains ‘CASIS’ on the determination of best experts. CASIS training algorithm is developed to complement current weakness. The algorithm doesn’t omit experts who respond late but complements them with active ones. The decision boundary is homogeneity and numerical so as to give a single output quickly. The input (x1, x2) and output (y) variables are numeric. The main concept is that the output is generated using linear combination or weighted sum model The algorithm considers response time as best criteria to satisfy the farmers who send the questions. This algorithm provides a great chance of finding a quick answer within a short period of time. Automatic expert assignment is essential to achieve high adoption of the system that satisfies the on-time farmer advisories demand and promote efficiency as well as effective extension services for rural development.