Salesforce and HubSpot may have placed a spotlight on generative AI agents, but they weren't the first to introduce them. In May, we shared how Tribble.ai was helping companies streamline their RFP processes with its Tribble agent. Now, Tribble is expanding the use cases it supports, and we spoke with one customer reaping the benefits.
Ironclad is a digital contracting platform company that helps businesses automate and streamline the contracting process. They first used Tribble.ai to improve the RFP process (Tribble's first use case). Dave Bloch, Ironclad's Solution Engineering Leader, shared how the company has expanded into use cases that support GTM teams.
Ironclad receives a high volume of RFP requests each month. Bloch said they went looking for a solution to improve their process, one that leveraged generative AI capabilities (Ironclad is an AI-first company, so they knew the benefits gen AI offers). There were a few reasons they selected Tribble, including its user-friendly UI and its level of partnership.
I'm not going to talk much about Ironclad's RFP use case except to say that with Tribble, the RFP process is now 30% faster. You can also read Clari's story to understand how Tribble supports the RFP process.
Streamlining RFP processes is a great use case, but Ironclad wanted to go further. As they thought about their data strategy and looked at the use cases they wanted to tackle, they saw that product capabilities were a natural fit. Bloch said Ironclad has a lot of public-facing support documentation that tells you everything you need to know about Ironclad, so they pointed Tribble at that first.
We surveyed our internal teams that manage knowledge, and in close collaboration with them, we have landed on data sources that are evergreen. And I think that's a big part of Tribble's technology strategy as well. Because of their integrations with Google Drive, for example, we can select content that will be updated automatically by the teams that are closest to it, and that gives us a lot of confidence that we're going to have the right answers when we get the responses from the bot.
The Tribble Brain sits on the backend, storing this information, which includes RFP responses and templates, FAQs, and PDFs generated by product enablement teams. It also integrates with Notion and Salesforce. In addition, Bloch said they do some tagging to ensure these sources are being pulled for the right questions.
He said they still need to be very intentional about the data sources and continually ensure they have the right information. They believe in the concept of "trust but verify." With Tribble responses, every source is listed, so the user can drill into the information to ensure they are getting an accurate response.
Ironclad rolled out an agent in Slack that allows the GTM team to get answers on anything about their technology, from its capabilities to how it stacks up against competitors, potential issues when helping troubleshoot with specific customers and questions on their selling motion. They started with a pilot of GTM influencers who were early adopters of generative AI technology and wanted to be a part of something innovative.
The pilot ran for a couple of months, and Bloch said they got an incredible response. They worked through some challenges, refining the model and how the data was tagged to ensure high fidelity of accuracy. When they got to a high level of confidence in the responses, they rolled the agent out to the rest of the GTM team and then to the entire company.
The Slack integration also enabled Ironclad to surface the tribal knowledge tucked away in product-related Slack channels. Tribble worked with them extensively, and now it can ingest these channels. Bloch said they can set the frequency and cut-off dates for pulling information, calling it a 'game changer.'
Imagine you're a seller, and you've just gotten off, you know, a bunch of customer calls for the day, and you're exhausted, and you've got a list of follow-up items because maybe you didn't know the answer to the customer's questions, and especially for a newer or ramping employee, it takes a long time to gain that tribal knowledge. And so in the past, what you would do is you would search through your email, you would search through old slack threads, you would ping your most knowledgeable solution engineer, and you might have to wait to get an answer. And then you'll have to take that time to then concoct a response to your customer. And we've essentially taken that time and condensed it to a matter of seconds because anybody from our company can go in and ask a question that might have taken them 30 minutes to an hour to think through, find the answer to, and then compose a response, and the agent just does that for you.
Moving forward, Bloch said they are moving towards helping GTM teams prepare for customer meetings. This includes providing responses such as what use cases the seller can mention and who are the marquee customers who have been good past references. Bloch said Ironclad wants to equip its sales reps (CSMs -- customer relationship managers) with the right stories to tell about the problems it helps customers solve.
I asked Block how they trained their people to use Tribble and learn how to write good prompts. He said what surprised him about rolling out Tribble was that they didn't have to do much in terms of change management.
When I was running the pilot, people were reaching out to me, saying, how do we get access to this thing? I heard it's very cool. And so I think it was the way we rolled it out, with that initial pilot group, and the group of people that we had, early adopters, influencers in their, you know, smaller, Ironclad communities, and it was just a lot of evangelism, and we have implemented in such a way that in certain Slack channels, folks can see what other people are, what the prompts that other people are using are, and they can learn from that. So, we started with the use cases that we supported. We did a really cool training for everybody initially, and then I think the adoption has really been an, 'I see these questions being asked. I see the way you're prompting; that's a really cool addition. I'm going to try to extrapolate that out.' And we're also doing continued enablement with teams, or that's what we're focused on so that they can get more out of the tool.
They call it "Ask SE with a buddy." Part of this is the trust but verify idea, where they want their people to check sources before sending something to customers. But they also proactively monitor the Slack channels where the agent lives and coach users when they see they haven't phrased a prompt correctly or asked the right question. Continual refinement is part of their ongoing process.
In a separate conversation with Sunil Rao, CEO and Founder of Tribble, he said it was interesting to see how people interact with Tribble and what they can learn from those interactions to improve the product and the experience.
When Tribble started working with Ironclad late last year, they talked about the new digital sales engineer product Tribble was building. Ironclad became both a design partner and the first customer to scale it out across a large percentage of the company (relative to other customers).
Bloch said that when selecting vendors to work with, technology is only one layer. Yes, you want to understand its value and ease and quickness with which you can roll it out. But for Ironclad, it's about partnership. He said Tribble takes the time to listen to its customers.
And I know that seems like an obvious thing, but they've been such great partners with us. And I think if you're a fast-growing company like ours, and you know, we've got big customers like Salesforce, OpenAI, and HubSpot, I think we value partners that are committed to our mission and committed to listening and working with us proactively to develop things like roadmap and help us be successful, and they've done that in spades.
With Tribble, Ironclad answers 4000-6000 questions a month for GTM teams, and that doesn't count the RFP teams using Tribble. Bloch said Tribble has helped double productivity. He also said that Tribble is helping improve the time required to ramp up new team members as they scale out their team, although he hasn't started quantifying the impact in this area.
Rao said that for Tribble it was important to meet the salesperson where they are, just in time, providing the right assets or information. What if, he said, instead of having a salesperson do a bunch of work preparing for a customer meeting and modifying generic assets to support a specific customer, that salesperson could generate exactly what they need right before the meeting? He said they are now figuring out how to teach that skill to Tribble.
But they are also going broader, looking at all the jobs a Sales Engineer does, and deciding what skills Tribble can learn to support the SE. Tribble has chosen to go deep on the SE persona, and Rao said they want companies to see Tribble as something you purchase or hire to perform a specific task in the best possible way.
We need to recognize that our work environments are changing, and AI is a huge reason for those changes. If we choose to look at AI as a partner and implement agents like Tribble as digital teammates, we will be able to spend more time and attention on the more strategic and creative aspects of our work -- whether we are sales engineers or marketers.
Sales teams may be better equipped to make this change because it's easier to envision a digital teammate doing a lot of the background work to gather information on products and services, past sales, and customer insights and bring it all together to support new customers. They already have interactions with customers, so it's natural to want more time to talk to them and think about what they need. Marketing may have a harder time making this shift, but I'll address that in my next article and explain why.