Claude and contract lifecycle management: how AI and CLM work together

Artificial intelligence has moved well beyond experimentation in legal teams.  

Tools like Claude are increasingly becoming part of the everyday toolkit, helping lawyers draft clauses summarize documents and accelerate routine work. The productivity gains are real, which is why conversations about AI are now shifting from whether legal teams should use these tools to how they fit into the broader contract management process.

That’s exactly the question we put to Summize Chief Technology Officer Rich Somerfield and Product Marketing Manager Bailey Gifford. Their view? The future isn’t Claude versus CLM – it’s Claude and CLM.

How is Claude transforming the way lawyers work?

According to Bailey, one of the biggest misconceptions in the market is assuming that AI productivity tools and contract lifecycle management platforms solve the same problems.

“Claude is incredibly good at helping lawyers work faster on individual tasks. Whether that’s reviewing a clause, drafting a first pass of a contract or summarizing an agreement for stakeholders, it acts as a highly capable assistant.”
- Bailey Gifford

For legal professionals juggling large workloads, that capability is definitely valuable. But the key is that productivity and management aren’t the same thing.

“What really improves with Claude is individual productivity. A CLM is doing something very different. It’s managing intake, negotiation, approvals, execution, storage, obligation tracking and reporting. Claude can help lawyers review and work with an individual contract, but it doesn’t manage your contract portfolio.”
– Bailey Gifford

It’s also important to recognize the limits of general-purpose AI. While tools like Claude can be highly effectively for drafting, summarizing and supporting contract review, they aren’t purpose-built legal systems and shouldn’t be relied on as the sole source of contract understanding or large-scale contract data extraction.

That’s an important distinction, because legal teams aren’t just responsible for reviewing contracts. They’re responsible for managing processes, collaborating with stakeholders and maintaining visibility across thousands of agreements.

Is AI alone enough for contract management?

In short, no. For Rich, the difference comes down to something much simpler: process. Businesses need consistency as well as the right answers at the right time.

“When you’re managing contracts at scale, you need repeatability. You need approvals to happen in the right order, you need obligations tracked, you need audit trails, you need version control.”
- Rich Somerfield

While AI models have become dramatically more capable over the past few years, Rich argues they’re still not designed to provide the structured workflows organizations rely on.

“AI is fantastic at helping people perform individual tasks. But businesses want process. They want reliability. That’s not really what AI models are tuned for.”
- Rich Somerfield

Instead, Rich sees the real breakthrough happening when AI is combined with systems that already provide those controls.

“What we’ve seen over the last year is that the real power comes from combining the model with a harness. By that I mean something that provides rules, structure and orchestration around the AI. In many ways, that’s exactly what a CLM is doing.”
– Rich Somerfield

The hidden challenges of building legal workflows around AI

As tools like Claude introduce custom workflows and advanced integrations, some organizations are beginning to explore building their own legal processes directly on AI platforms. Rich says it’s technically possible, but the challenge comes after the demo.

“AI demos really well. I could spend 15 seconds showing you an incredible contract review workflow. But taking that demo and making it repeatable, consistent and reliable for a business takes an incredible amount of effort."
– Rich Somerfield

That distinction is often overlooked when evaluating AI technology. Rich compares it to software development itself: building a proof of concept can happen quickly, but turning it into a reliable, maintainable system is where the real work is.

“There’s a huge difference between showing that something is possible and turning it into a business-critical process that hundreds of people depend on.”
- Rich Somerfield

Bailey sees another challenge emerging as organizations start building their own AI-powered workflows: knowledge silos. The result is often a collection of individual workflows rather than a standardized process.

“If one lawyer uses Claude to review a contract, that review lives primarily in their chat environment. You can share outputs, sure, but that’s not the same thing as actual collaboration. It’s not the most efficient way to share information. Full CLMs are designed for multiple people to be working in a document at one time on a contract with version control, approvals and governance all built in. You can share information with Claude, but it’s really an efficiency tool, not technology that’s built for the reality of how legal teams work.”
– Bailey Gifford

What happens when you move beyond single contracts?

The conversation becomes even more important when organizations start looking beyond individual contracts and towards portfolio-level insights. Reviewing one contract is one thing, but repeatedly querying thousands of agreements is another entirely.

“If you want to understand average payment terms across an entire portfolio that could be hundreds of thousands of contracts, you don’t want an AI model reading every contract from scratch every time.”
– Rich Somerfield

At that point, the challenge isn’t just accuracy, but efficiency as well. Standalone AI assistants rely on tokens for every interaction, so repeatedly analyzing large contract portfolios can quickly become slow, expensive and subject to usage limits – if you run out of tokens, your workflow can simply stop.

Purpose-built CLM platforms take a different approach. Rather than sending every query back to the AI model, they extract and structure key contract data upfront. AI can then work with that structured information instead of repeatedly processing the same documents.  

“The extraction phase is critical. Once you’ve structure that information properly, you’re using the right technology for the problem. You aren’t spending tokens every time you need an answer.”
– Rich Somerfield

At this point, it’d be reasonable to ask whether CLM platforms like Summize are using tokens as well, and they are. The difference isn’t whether tokens are used, but how they’re orchestrated behind the scenes to make AI more efficient, predictable and scalable.

“In Claude, every interaction consumes tokens. In Summize, AI capabilities are still powered by tokens, but they’re orchestrated behind the scenes. We optimize model selection, prompt engineering and context for each task, so AI is only used where it adds value. That means customers aren’t managing token usage themselves, and their contract workflows and structured data remain accessible regardless of how AI is being used.”
– Bailey Gifford

Rather than treating AI as the entire platform, Summize treats it as one part of a broader contract management system. The result is a more scalable approach: AI delivers productivity where it’s most valuable, while the platform provides the structure, workflows and governance that organizations rely on as they grow.

Governance, privacy and why legal data can’t be generic

Both Rich and Bailey emphasize that one of the most overlooked risks in AI-first legal workflows is data provenance. Bailey’s clear that legal outputs are only as reliable as the data they’re grounded in.

“If you’re getting outputs trained on general data, they’re not going to reflect how your organization actually negotiates. They’ll reflect how the general market or model thinks contracts should be handled.”
– Bailey Gifford

In practice, that difference matters for obvious reasons – legal negotiation styles, risk tolerance and fallback positions vary significantly across organizations.  

“You need your organization’s data, not only from a security perspective, but from a consistency perspective. Legal work needs to behave like legal work in your business, not generic legal advice from a model.”
– Rich Somerfield

This is where purpose-built systems matter: keeping data structured and controlled within a governed environment, rather than spread across ad hoc AI tools.

Is the future Claude vs CLM?

For both Rich and Bailey, the most interesting development is AI connecting to CLM, rather than AI replacing it.

Through technologies like MCP, legal teams can increasingly access contract data and workflows through conversational interfaces, while still benefiting from the structured governance and reliability of a dedicated CLM platform.  

“MCP allows us to plug Summize directly into Claude. You get the dynamic chat experience people enjoy, but it’s connected to structured contract data, established workflows and all the governance that’s already built into the platform.”
– Rich Somerfield

In practice, it means legal teams don’t need to choose between productivity and process – they can have both.

“The conversation shouldn’t be Claude or CLM. The productivity gains from AI are real and legal teams should absolutely be exploring them. But AI assistants don’t replace real contract lifecycle management.”
– Bailey Gifford

Instead, the strongest legal teams will combine the speed and intelligence of AI with the visibility, collaboration and governance that CLM provides – and as legal technology continues to evolve, that balance might prove to be the most valuable capability of all.

Want to go deeper on where AI fits into contract lifecycle management? Read our guide to AI and CLM for a closer look at how the two work together in practice.