Getty Images/iStockphoto

With Tribble, RFP writers get their own shot of generative AI

Some large enterprises' sales organizations respond to a hundred or more requests for proposals daily. A new GenAI tool from ex-Salesforce execs could ease the workload.

Tribble Thursday formally released a generative AI RFP copilot. It promises to slash the time salespeople take to answer requests for proposals -- the tedious information-gathering part of sales that comes after the prospecting and before the champagne toast when the deal is closed.

The company, which launched about a year ago, is up against formidable competition racing for mindshare in this potentially lucrative niche of CRM users. Twilio released RFP Genie, which generates RFP responses. Google launched a similar tool internally. Last year, an IBM Watsonx RFP generator won an internal hackathon -- which sometimes leads to product releases. Response management tech vendors including Responsive (formerly RFPIO) are also integrating GenAI into their platforms.

Tribble hopes it will hit the right notes with prospective customers in part because the company understands users of Salesforce, the dominant CRM. Tribble co-founders Ray Shipley and Sunil Rao, respectively, came from Salesforce's sales operations and Salesforce Industries' vertical-specific clouds team. As such, Salesforce customers make up a good chunk of Tribble's early customers, even though Rao said the tool is CRM-agnostic and the company hopes to grow its share of other platforms' users.

Understanding the differences among vertical industries, too, may also give Tribble a leg up in eventually customizing offerings for customers in life sciences, financial services and other areas.

Tribble customer Own Company backs up cloud CRM data, offering customers daily snapshots of their data and disaster recovery services. While many Salesforce customers imagine that Salesforce keeps long-tail backups of their data, it doesn't.

"When you blow away 3,000 of your customer names, there's no getting them back -- so that's on you," said Rich Kline, vice president of solution engineering at Own, who is also a Salesforce veteran.

Many sales cycles begin with RFPs; it is how a customer gathers information from competing vendors to choose which one will win the business to fill a need. Solution engineers at Own, one of Tribble's earliest customers, used to write RFPs themselves and have their managers review their work.

Own's solution engineers now use Tribble to generate their RFPs, trained on 16 complex documents that contain data compliance and security data. While solution engineers remain the humans in the loop to edit the RFPs, Tribble has cut RFP response time in half, Kline said. The time savings enables them to get back to their main job of backing up sales reps with deep technical knowledge of Own's products and services.

"The answers come out where it passes the Turing test," Kline said. "Nobody can really tell if it was theirs. We're just not seeing the hallucinations, because we're keeping the amount of [training] information so tight."

Screenshot showing RFP training.
Tribble provides attribution of the sources it uses to train its RFP response model.

Tactical GenAI deployment on the rise

Tribble also released "digital sales engineer," a virtual assistant that answers technical questions for sales reps.

Tactical AI deployment -- as opposed to general-purpose AI designed to answer everything or solve a host of problems -- is also ripping a page from the Salesforce playbook. For almost a decade, Salesforce has deployed AI to solve one business process at a time, such as sorting sales leads to determine which is most likely to result in a closed deal or creating the most efficient truck route for a field service agent.

What's different about Tribble, said Alan Pelz-Sharpe, founder of Deep Analysis, is the tactical deployment of generative AI (GenAI). It's the start of a trend where many tech vendors are getting granular with GenAI.

"We have two parallel enterprise AI paths at the moment," Pelz-Sharpe said. "The first is just folks playing around with things like Microsoft Copilot and ChatGPT -- that's good, insomuch as they are getting familiar with the concept. But few are doing much more than playing around. The results are often disappointing, and the novelty wears off pretty quickly.

It's not about AI transforming an entire operation; rather, it's about AI automating a task or set of tasks that is currently costly and error-ridden.
Alan Pelz-SharpeFounder, Deep Analysis

"The other path is longer but more valuable, where organizations are figuring out how to deploy AI very tactically to meet specific but critical business needs. So, it's not about AI transforming an entire operation; rather, it's about AI automating a task or set of tasks that is currently costly and error-ridden."

Automating RFP responses, which Pelz-Sharpe characterized as "typically a long-winded and costly activity that often fails to produce good results," could be a great fit for GenAI.

Not only do humans currently spend a lot of time crafting RFP responses, but it can also take weeks to figure out which RFPs should be responded to and which ones shouldn't. Generative AI could save companies that time winnowing the field.

The big challenge for Tribble? Getting enterprise customers to rethink their RFP processes, which typically have been deeply entrenched, codified and proven over the years.

"This would be a radical overhaul of a long-established practice," Pelz-Sharpe said. "Any AI tool will need a fair amount of services spend to ensure it accesses the right data and generates a near 100% [appropriate] output."

As a startup, it might seem tempting for Tribble to tackle adjacent sales processes with generative AI. But cofounder Rao said the company will remain focused on the RFP copilot and the chatbot -- and customizing it for users in different industries to help automate the dissemination of technical information to salespeople.

"There is so much depth to this process across the different verticals, there is value in becoming more precise in each," Rao said. "There are so many different variations of what can happen during the [sales] process."

Don Fluckinger is a senior news writer for TechTarget Editorial. He covers customer experience, digital experience management and end-user computing. Got a tip? Email him.

Dig Deeper on CRM tools and strategy

Content Management
Unified Communications
Data Management
Enterprise AI