Bartleby.dev
Knowledge graphs are powerful tools for digitizing organizational knowledge and distilling insights. With the development of LLMs, knowledge graphs have become even more powerful tools to drive intelligent workflows and applications. As we discussed in our prior piece introducing knowledge graphs, intelligent applications provide insights, take actions and even complete entire workflows for users. In this piece, we’ll explain how to leverage your knowledge graph to build intelligent workflows and applications for your organization.
Let’s start with a simple example of an intelligent action. A common use case we see today is leveraging LLMs to automate actions around the sales funnel or CRM. For the CRM example, an intelligent action might be using an LLM to summarize a Zoom meeting with a prospective client. Another example could be asking an LLM to help you prepare for the meeting by sending you an overview of the prospect ahead of the call. A third action might be instructing an LLM to recommend next steps based on the summarized meeting, or remind you when to reach out to the prospect for follow-up.
By chaining these intelligent actions together, we can build an intelligent workflow. Let’s call our example workflow Mid-Funnel Management. The Mid-Funnel Management workflow automatically prepares you for your upcoming meeting, summarizes the meeting, sends you next steps, and reminds you to reach out to the prospect to set-up a follow-up call. In another post, we will discuss how we leverage Bartleby.dev’s custom framework for building intelligent workflows, but here we will focus on how pairing your knowledge graph with an LLM can supercharge this workflow.
The power of this combination of technologies is that it equips your LLM actions with reasoning and context specific to your organization. Although we could just chain several LLM calls together to complete the Mid-Funnel Management workflow, it would lack knowledge about your company and prospect, as well as the semantic context around this series of interactions. With a knowledge graph, we can inject institutional knowledge into the workflow.
The below example shows how each of the steps in the Mid-Funnel Management workflow can be supercharged by enriching them with relevant information from your organization’s custom knowledge graph:
Prepare for Meeting
→ client name, industry and other typical CRM data
→ prior meeting notes and interactions
→ questions to ask of a similar client with a similar use case
Summarize Meeting
→ example summaries from similar prospect meetings
→ organization’s custom format for meeting summaries
Send Next Steps
→ previous steps taken for this prospect
→ typical next steps for a similar client with a similar use case
Remind for Follow-Up
→ organization’s typical time before following-up with prospective clients
→ organization’s timeline for a prospect in this stage of the funnel
By digitizing the knowledge about how you interact with clients – the types of questions you ask, the way you format your summaries, the next steps you usually take and the amount of time you wait before follow-up – your organization can build intelligent workflows that are inextricably linked to institutionalized knowledge about your organization’s processes and historical relationships. With a knowledge graph of your organization, an LLM can become much more than a helpful tool, but a virtual member of your company, with access to the same knowledge as your human employees.
In our custom software manifesto, we discuss our vision for intelligent software for enterprise. We argue that the future of enterprise software is a single pane of glass solution custom built for your business. We believe the first step toward this future is building a custom knowledge graph to power intelligent workflows specific to your organization. One day, we hope that organizations will be entirely powered by their own custom workflows, all accessed in one intelligent application.
But let’s start with a knowledge graph. When you engage with us, we’ll bring a custom ontology or data schema for your business to the first meeting. We’ll then help you populate it with your data to build your own custom knowledge graph. Leveraging our client portal and proprietary intelligent workflow framework, we will work with you to build and implement intelligent workflows that are specific to your business.
If intelligent workflows are an area your organization is pursuing, please reach out. We’d love to work with you.