rfpXer – responding to RFPs at scale

We at Forter, receive multiple RFPs from prospective customers that require our attention. If you don’t know what an RFP is, then you are a lucky bastard.

An RFP (literally a Request for Proposal) is a document written by a prospective customer, sent to the sales team, and forwarded to marketing team, then reforwarded to Engineering. A typical RFP consists of hundreds, sometimes thousands, of questions covering organizational procedures’ best practices. A typical question would be “What is your remote access policy?”

RFPs are known to have a very strong damping effect on Sales, Marketing, but mostly on the Engineering teams (the sales team just forwarded it, remember?):

Knowing the RFP throughput would keep growing, we came up with RFPXer. A scalable approach for caching, routing and answerXing RFPs.

The interface is pretty simple, and is based on Slack APIs. The rfpXer slackbot searches for previous answers from other RFPs. Internally we use Elasticsearch “More Like This” query, to find answers given in the past to a similar question.

In case past answers are not available, the bot can ask an employee to answer this question. We use the FastText clustering algorithm to categorize the question, and route it to the right person in the organization.

If the employee fails to respond, we use a proprietary machine learning algorithm to answer the question.

Forter will soon release the rfpXer source code under the AGPLv5 license. Stay Tuned !

Happy April’s Fools Day !

About This Author

Itai has a lot of experience with and affection for distributed data architectures, cloud computing, open source software and security standards. Itai achieved internet fame in 2015 in Forter's "Software Architecture Addict" video which he scripted, starred in and produced. He really isn't quite as obsessed as the addict in the video, but he has been known to say that he “has never left his keyboard since he was 6 years old.”