The Death of the CRM

One of the worst kept secrets in VC and startup circles is that LLMs have fundamentally changed how we interact with data. From product telemetry to application code to help center articles: all of this data can be analyzed and reported on at scale with LLMs. For GTM teams, this means breaking down and reimagining the holy grail of customer data: the CRM.

CRM History

First, a quick whirlwind history: The earliest CRMs commercialized relational databases by focusing on customer data. Siebel Systems popularized the tool in the 1990s (Thomas Siebel now runs C3.ai), and Salesforce brought it into the cloud. Now, nearly every business tracks their customer pipeline in a CRM. This is used to better understand customers at scale and predict what the customer will do: engage with your reps, buy your product, renew your contract, churn, etc.

However, the reality of modern CRMs is they are only as good as the data in the system. This is reliant on human input, manual processes, and unreliable data sources and integrations. No company has a true customer 360: just ask any sales leader running a deal review.

What’s Different Now

With LLMs, you can now reliably query unstructured data the same as your structured data. Compared to CRMs, there is at least an order of magnitude more unstructured customer data across customer conversations and knowledge across internal systems: calls, Slack, email, public documentation, public filings, help center documentation, internal sources of truth – the list goes on.

With LLMs, you can ask questions of the data flexibly and at scale:

• Which of my customers are the most likely to churn because the last QBRs didn’t go so well, and what is the total contract value across those customers?
• Which of my reps fully qualified their opportunities? Are they following MEDDICC/MEDDPICC?
• Which “need-to-have” product features were most requested by customers, for the next sprint planning cycle?
• How are my reps handling customer objections? What are the most common objections and what are the best responses?
• The list goes on...

Why is this Impactful? Where do things go from here?

Anything you’d ask your reps or CSMs, you can now ask an LLM. There is no manual data upkeep or ongoing data validation needed. Customer conversations are the source of truth, and LLMs read straight from the source – and at scale. You can get a full customer 360 with much of the nuance needed to make real revenue decisions straight from the data, all with no human translation or, god forbid, reading CRM notes (better and more accurate notes are being AI-generated from call transcripts on-demand).

At the present, the CRM is still the “single source of truth” of customer data. In the future, the CRM will sit alongside customer conversations and other unstructured data sources as one of many “sources of truth”. It will take a backseat to the aggregation layer of the various “sources of truth” that sits on top. More data will live in information-dense call transcripts, and the corresponding custom fields in Salesforce (e.g. your MEDDICC checklist) will disappear.

As a rep, manual data entry will be a thing of the past. Many new gen AI sales tools are already automating notes and next steps, but in a post-CRM world, there will be no need to track any of this. As a sales or success leader, you get direct insights and visibility into your pipeline, letting you make the best decisions. We are not saying goodbye to deal review just yet, but we certainly think it can be streamlined with more visibility into conversations. The ability to understand every aspect of your pipeline is at your fingertips, the right interface just needs to be built.

Pretty soon, someone is going to build the first general purpose query and reporting layer for all customer data: structured and unstructured. Curious where this fits in with Quilt’s roadmap? Reach out to dan@quilt.app to learn more.