Collaborative RFP platform
Software to streamline the way organizations handle Requests for Proposals. The software must offer the users a better and a fast way to collaborate and work on proposals by letting them create, edit, review, validate, and approve valuable knowledge all in one place.
- The major challenge was to upgrade the information search and retrieval to a stage where the software can generate required and matching information.
- The identification of sections/subsections from the document, extraction of questions and suggestion of answers for approval.
- The platform has different user roles with varied permissions and privileges who can collaborate and manage the RFP process better. The AI-powered one platform-multiple user solution helps to respond to RFPs seamlessly and faster by automating the RFP process. The platform auto-analyses documents to understand and extract the requirements using NLP and ML techniques. A number of questions are automatically answered by the software that learns through the responses of different users and the users can also customize their responses to each requirement. The Knowledge Manager that continuously curates, learns, and shares new knowledge is the key to this platform. As new questions are answered, a crowd-sourced knowledge base is automatically formed, accessible by all employees.
- For response processes that require specialist expertise, a number of additional capabilities are available to connect colleagues within the platform. This helps to quickly assign content tasks to SMEs, match experts to topics, track progress with ease and operate a real-time response process to deliver RFPs on time. A dynamic document generator helps the users create a draft proposal by inserting questions and answers with auto-formatting and smart organization capabilities. A bulk import feature integrated into the software allows it to import past RFP documents of an organization to speed the RFP responses.
- A Machine Learning model built using the Random Forest classifier does the process of section and subsection identification in the document. A Deep Learning model created using BERT performs the job of question extraction and another Deep Learning model created using LASER does the job of answer recommendation.
- Dropwizard, a Java framework, was used to create the whole platform. Deep learning models(BERT and LASER) were written in python and were exposed using flask servers. The flask servers were containerized using Docker.
- A unified solution to manage the RFP process better was built using AI and ML techniques.
- The platform helps centralize all the RFP answers in a knowledge base with quick access to expert responses.
- It also offers the users across organizations or within one organization to share a common workspace by assigning tasks, collaborating and approving responses, and managing deadlines which are key in a sales process.