TL;DR: In this blog, legal tech experts provide a roadmap for implementing AI tools through a structured framework covering vendor evaluation, due diligence, adoption strategies, and ROI measurement. The panel from VSCO, Zeppelin, and Zuboot.ai outlines practical steps for assessing AI security requirements, conducting thorough vendor vetting, balancing risk with efficiency gains, encouraging team adoption, and developing metrics to quantify impact. The action plan includes specific implementation phases from needs assessment to organizational transformation.
As artificial intelligence transforms the legal profession, how can legal teams effectively evaluate, select, and implement AI tools? At our recent SpotDraft Summit 2025, experienced legal tech leaders shared practical insights on navigating the rapidly evolving AI landscape—from conducting proper due diligence to encouraging adoption and measuring ROI.
The AI revolution is well underway in the legal industry, with new tools emerging daily that promise to transform everything from contract negotiation to legal research. For in-house legal teams, the challenge isn't whether to adopt AI, but how to do so effectively, responsibly, and with appropriate risk management.
Our panel of legal technology leaders offered pragmatic guidance on approaching AI implementation, sharing both successes and cautionary tales from their experiences on the front lines of legal innovation.
Meet our expert panel
Our discussion featured insights from leaders representing various perspectives within the legal technology ecosystem:
- Florence Chan (Moderator), who advises companies on building smart, scalable legal teams and transforming legal departments into growth engines
- Jose Lopez, General Counsel at Zeppelin and former in-house counsel at Samsung and Salesforce
- Akaash Gupta, Deputy General Counsel at VSCO, where he manages privacy, product compliance, and commercial transactions
- Kyle Kelly, Founder and CEO of Zuboot.ai, who previously led e-discovery initiatives at Salesforce and built Coinbase's e-discovery department
The current state of AI adoption in legal
To start the conversation, our panelists shared their personal comfort levels with AI, revealing a generally optimistic but appropriately cautious approach.
"I'm at a 4.5 [out of 5]," said Florence Chan. "I'm the laziest person in the room. So if the Terminator can come in and mop my floors and answer my emails and my calls from my mom, fantastic, I'm on it."
Jose Lopez echoed similar enthusiasm: "I like to experiment with not just AI tools, but legal tech generally. I try to operate my workday and our team's workday by trying to avoid doing the same thing twice."
Akaash Gupta took a balanced approach: "I think I'd probably be at a four as well, simply because I've seen Terminator, and it does not end well. But the way I look at AI is that I can use it to actually enjoy my job more—by helping me get away from tasks that I don't need to be spending too much time on."
This cautious optimism set the stage for a practical discussion on how legal teams are currently leveraging AI technologies.
Current AI use cases in legal departments:
Our panelists highlighted several key applications of AI that are already delivering value in their organizations. Here's how you can implement similar solutions:
Legal research and regulatory monitoring
Action steps
- Identify regulatory areas requiring constant monitoring (privacy laws, financial regulations, etc.)
- Implement AI tools that can track legislative changes across multiple jurisdictions
- Set up automated digests that summarize regulatory developments in your industry
- Create templates for AI to transform complex regulations into actionable compliance checklists
Akaash Gupta shared: "As a product counsel, I've found using AI for legal research has been super helpful. It's really tough to stay on top of all the laws that are impacting our products and technology every day. There's a new state privacy law like every day, right? Using AI to help stay on top of those things and distill complex regulations into more digestible ways has been super helpful."
Contract analysis and management
Action steps:
- Audit your existing contract repository to identify metadata extraction opportunities
- Train AI systems on your standard agreements to improve recognition accuracy
- Develop standardized data extraction parameters (parties, termination dates, renewal terms, etc.)
- Create automated workflows that flag contracts requiring human review
- Set up regular reports that highlight upcoming contract expirations or renewals
Jose Lopez noted: "We use SpotDraft, and the AI in SpotDraft is something that we use every day. The smart data capture feature will take the metadata from contracts and make it something that's malleable and usable so you know what's in your corpus of agreements without having to read through everything one by one."
Legal document generation and analysis
Action steps:
- Inventory your most commonly used legal templates (NDAs, responses to subpoenas, etc.)
- Create AI prompts that extract key information from incoming legal requests
- Develop standard response templates that AI can customize based on specific facts
- Implement a human review checkpoint before document finalization
- Track template effectiveness and refine based on feedback
Kyle Kelly described how his company Zuboot uses AI: "If you receive a subpoena and you're sweating palms and mom's spaghetti, all that jazz, you're probably wondering how you figure that out. I built a tool that allows you to understand the facts of the matter inside of that subpoena to prepare yourself. And then if you also don't have drafts, we will generate templates for you based upon the facts of the matter."
Finding the Value Proposition: A Strategic Framework for AI Implementation
With hundreds of AI tools marketing to legal departments, how do you determine when and where AI can add the most value? Our panelists suggested a systematic approach to identifying high-impact opportunities.
Strategic AI implementation framework:
- Conduct a task audit
- Document all recurring tasks performed by your legal team
- Measure time spent on each activity category
- Identify patterns of repetitive work
- Flag tasks that cause bottlenecks or delays
- Match AI strengths to your needs
- Information distillation: Use for regulatory research, case law analysis, due diligence
- Pattern recognition: Deploy for contract analysis, risk identification, compliance monitoring
- Language processing: Utilize for drafting, summarization, translation, and communication
- Prioritize implementation targets
- Start with high-volume, low-complexity tasks
- Focus on areas with measurable outcomes
- Target processes that affect customer-facing teams like sales
- Begin where risk tolerance is higher
Jose Lopez recommended: "I start by asking what is AI good at. We know that it's really good at that distillation function—taking a big corpus of information, distilling it, giving you something that's more digestible. It's also really good at taking media from one form to another, like transcribing meeting recordings."
Evaluating AI vendors: A due diligence checklist
Once you've identified potential use cases, selecting the right AI tool requires systematic evaluation. Our panelists emphasized a structured approach to vendor assessment, with particular focus on data security and privacy.
AI vendor due diligence checklist:
- Data security and privacy assessment
- ✓ Request documentation on data handling practices
- ✓ Verify if vendor trains their models on your data
- ✓ Confirm data storage location and duration policies
- ✓ Determine who has access to your instance and data
- ✓ Ensure data separation between customers
- ✓ Verify encryption standards for data in transit and at rest
- Internal coordination
- ✓ Consult with IT about existing tools that could meet your needs
- ✓ Check if other departments have already vetted similar solutions
- ✓ Explore cost-sharing opportunities across departments
- ✓ Identify necessary integrations with current systems
- Contractual protections
- ✓ Secure appropriate indemnification provisions
- ✓ Establish clear SLAs for accuracy and performance
- ✓ Implement right to audit provisions
- ✓ Include data portability/export requirements
- Proof of concept testing
- ✓ Test with your own real-world examples
- ✓ Implement a structured evaluation methodology
- ✓ Compare results against human performance
- ✓ Set clear success metrics before beginning the POC
Akaash Gupta outlined his threshold questions: "There are just some kind of threshold questions that I do like to ask, even before the use case. I want to know what happens with my inputs. Like if I input something, are they training on it? And if so, are they storing it? Who has access to my instance? Are you able to see everything that I'm asking AI to look at? I mean, if you're a public company and you're using AI to help with an acquisition, for example, that can't leak."
Kyle Kelly emphasized the importance of understanding your own needs first: "You have to understand what you actually need, because everyone will sell something to you. I also encourage legal teams to talk to teams outside of legal, because you may already have a tool in the cabinet that you were not aware of."
How to encourage AI adoption
Even with the right tool in place, getting your team to embrace the technology requires a deliberate change management approach. Our panelists offered some tactical steps for driving adoption across legal teams and stakeholders.
AI adoption strategies:
- Create compelling use cases
- Develop personalized examples for different team members
- Document before/after metrics for similar teams
- Demonstrate time savings with concrete examples
- Calculate productivity gains in tangible terms
- Implement progressive training
- Start with low-risk, high-reward applications
- Create simple "quick start" guides for common tasks
- Schedule progressive learning sessions (basic → advanced)
- Pair AI-enthusiastic team members with hesitant ones
- Reduce barriers to experimentation
- Set up sandboxed environments for risk-free testing
- Create template prompts for common legal tasks
- Establish a "no judgment" policy for initial attempts
- Celebrate and share successful use cases
Akaash Gupta focused on demonstrating clear benefits: "I think it's just showing people how it'll make their jobs easier. If you can tell a salesperson that they're going to close their contract within half the time and get their commission sooner, they're not going to take much convincing. For people on the legal team that may be resistant, do you really want to spend all your time removing auto-renewals from an order form, or do you want to draft an interesting, complicated partnership agreement?"
Kyle Kelly advocated for a "just do it" approach: "You fall, you have some scabs on the way, but you find your balance, you make movement. If you look at what you have to do as your normal day-to-day job, and you're like, how can I do this better? I'd say that's probably 95% of the people that you would meet that want to just do better."
A balanced framework for measuring ROI and managing risk
Once you've implemented an AI tool, systematically measuring its impact while managing risk requires a structured approach. Our panelists emphasized practical metrics and risk calibration methodologies for legal AI implementation.
ROI measurement framework:
- Quantitative performance metrics
- Track contract processing time (from draft to signature)
- Measure volume of matters handled per attorney
- Calculate cost per legal transaction
- Monitor average response time to legal requests
- Analyze automation percentage (tasks automated vs. manual)
- Risk calibration approach
- Create a tiered risk classification system for legal matters
- Develop appropriate review protocols for each risk tier
- Set clear guidelines for when AI output requires human review
- Implement random quality checks against human-only benchmarks
- Document all instances of AI errors for continuous improvement
- Stakeholder satisfaction tracking
- Gather feedback from internal clients on legal service improvement
- Measure Net Promoter Score for legal department services
- Track references to legal as "business enablers" in company surveys
- Document specific business opportunities accelerated through AI
- Efficiency reinvestment strategy
- Plan how to reallocate time saved through automation
- Identify high-value activities for increased attorney focus
- Create metrics for tracking strategic vs. administrative work
- Develop a roadmap for expanding AI capabilities based on success
Akaash Gupta emphasized the importance of risk tolerance in determining ROI: "It ultimately comes down to how much risk you're willing to take and how much trust you're going to put into the tool. If you want to operate with 0% risk and every time you use AI to draft a contract you redraft it yourself, then your ROI is not going to be very good because you're not really going to be saving any time."
He continued: "If you are comfortable with just a certain level of risk and a certain level of imperfection, while still doing your job as a lawyer who is representing the client and being responsible, then you can get a really good ROI out of it and it can unlock you as a lawyer and help you help the business accomplish a bunch more things."
The future of AI in legal: Preparing your department for 2029
"I think AI is going to be like Google, it's going to be everywhere. It's going to be in the background. That you're like, 'oh, of course, now I don't have to look at my calendar, my calendar tells me what I'm doing.' It's going to feel a lot like that.” - Kyle Kelly
Our panelists concluded with strategic perspectives on how AI will transform legal practice in the coming years. Here's how forward-thinking legal departments can prepare today for tomorrow's AI-augmented reality.
Five-year AI readiness plan:
- Develop a legal tech evolution roadmap
- Audit current manual processes for future automation potential
- Map skill development needs for your team's evolving roles
- Plan phased implementation of increasingly sophisticated AI tools
- Evolve your department's operating model
- Redesign legal team structure around human-AI collaboration
- Develop new roles focused on AI oversight and optimization
- Establish governance frameworks for AI-human accountability
- Set new success metrics
- Shift from hours billed to value delivered metrics
- Track business impact rather than legal department efficiency
- Measure proactive risk mitigation rather than reactive problem-solving
Jose Lopez expressed confidence in the enduring role of human lawyers: "I'm not afraid of the Terminator scenario. With respect to the profession, I'm also not worried about being replaced. I think there's still just the goulash of the relationship, understanding the business super well, understanding the priorities, objectives, dreams, fears of your client."
Akaash Gupta offered an optimistic view: "I actually think five years from now our jobs are going to be way more fun. I love being a lawyer. I love parsing apart a complicated factual issue into all the legal components and coming up with a solution and advising. A lot of our jobs isn't that, and I think the AI can kind of take on those tasks that are necessary and keep the department running, that'll allow us to work on more interesting stuff."
Key takeaways for implementing AI in legal
Based on our panel's insights, here are practical guidelines for legal departments navigating AI implementation:
- Start with what AI does well — distillation, media conversion, metadata extraction — and map those strengths to your most time-consuming tasks
- Conduct thorough security and privacy due diligence on AI vendors, particularly regarding how they handle your data and whether they're training on it
- Understand your risk tolerance — it will determine both which tools you select and how you measure their ROI
- Look for existing tools in your organization before purchasing new ones; collaborate with other departments
- Encourage adoption by demonstrating clear, tangible benefits to both the legal team and their internal clients
- Measure success quantitatively where possible, such as through time-to-signature metrics
- Accept appropriate imperfection — requiring perfection undermines the efficiency benefits of AI
- View AI as an assistant, not a replacement — focus on how it can free you for higher-value work
As Akaash Gupta aptly summarized, the future of legal AI is not about replacing lawyers but enhancing their capabilities: "Human-centered, AI-assisted" represents the sweet spot where technology amplifies rather than diminishes the unique value that legal professionals bring to their organizations.
This blog post is based on the panel discussion "Practical Approaches in Selecting, Evaluating, and Managing AI Tools" from our recent SDSF event, featuring Kyle Kelly (Zuboot.ai), Jose Lopez (Zeppelin), and Akaash Gupta (VSCO), moderated by Florence Chan.
