Generational change is reshaping today’s legal sector, making it challenging for firms to prosper. To stay competitive, legal providers are rapidly adopting AI-powered solutions firmwide.
AI, once considered advanced technology, is now as integral to law firms' operations as their case and practice management software. Today, professionals discuss agentic AI most often, as it plans and executes multistep processes with minimal input.
However, the rapid rise of AI doesn’t mean that all AI-powered solutions are equal. A law firm must determine the system that best suits its needs and avoid those that may create more problems than solutions. Legal professionals must investigate how the solution functions, identify its strengths, and assess its potential weaknesses. To achieve this, they need to ask the AI system provider detailed and precise questions — and demand equally detailed and precise answers.
Because selecting an AI provider is perhaps the most critical choice a legal firm will make this decade, a poor decision wouldn’t be merely a strategic misstep — it could be catastrophic.
Considering AI systems: Initial steps
Using AI in a legal practice means the system must handle and safeguard sensitive client data, while also ensuring the law firm stays in compliance with professional ethical standards and applicable state and federal laws. Firms must implement these systems while maintaining high levels of speed, accuracy, and reliability for clients.
There’s a great deal that could go astray with an inferior system. If an AI system produces assessments plagued with data hallucinations, it will wreak havoc on client services. An AI provider that’s lax about privacy and data protection could expose the law firm to data breaches, which may lead to lawsuits and lost business.
What should law firms look for in their AI technology provider? It boils down to a few essential requirements:
- Access to professional-level content and extensive, verified databases
- Having top security and professional expertise at hand
- Being committed to continual innovation, pushing to consistently improve the AI system
- Having a partner ecosystem that integrates the solution seamlessly into existing workflows
These aren’t qualities you can always verify through a provider’s website. Your firm’s decision will require having in-depth conversations with prospective AI providers. Finding the right AI solution may require time and effort. But make no mistake — every day you wait to adopt professional-grade AI will put you exponentially further behind.
Define use cases
Let’s start with the fundamental questions: how and where does your law firm want to use AI? Which of your firm’s strategic initiatives would most benefit from AI?
Here are two use cases where AI could be the most productive and transformative for your firm.
Legal research and analysis
Firms commonly use AI to enhance their legal research and analysis capabilities. The goal is to boost efficiency and speed by using a solution’s agentic capabilities to suggest and then execute the best steps for a given project at a lawyer’s request.
Determine whether the prospective AI system can summarize case law and identify relevant precedents. A defining question is “What type of legal expertise database does the system draw upon?”
Contract drafting
The AI system should also be proficient in drafting contracts, from initial drafts to the revision stage. It should be capable of catching anything amiss, whether it’s a grammatical mistake or any murky language that opens the client to potential risks. A quality AI solution will, among other things:
- Automatically propose changes to standard contract terms to better align with an organization’s objectives
- Analyze a motion for summary judgment from opposing counsel and research authoritative content to support your firm’s response
- Help a client expand into a new state by automatically updating an LLC operating agreement, and similarly ease the way for an international expansion
- Search and analyze documents to expedite due diligence and mitigate risk
The savings potential
According to a recent Future of Professionals Report, AI could save professionals about 12 hours per week over the next five years. That’s 200 hours a year — essentially, adding a virtual colleague for every 10 team members on staff.
The potential return on investment (ROI) is equally substantial, with an estimated $100,000 in additional billable hours for a U.S. lawyer. Furthermore, it leads to intangible improvements for the lives of legal professionals, such as a better work-life balance and reduced junior lawyer burnout.
Address ethical concerns and potential biases
A law firm should choose an AI system that strengthens its ability to detect and avoid any potential ethical violations, as well as detect and eliminate any biases in its operations.
In the age of AI, the issue of bias is pertinent, often produced by systemic, computational, statistical, and human cognitive factors. Such bias might not be readily perceivable when you review any given AI output. But its negative impacts — particularly when an AI system makes predictions or recommendations about individuals or groups based on erroneous assumptions — could affect client relations and even attract government regulators.
A law firm should know which measures the AI provider takes to prevent bias in its system. The provider might devise a code of ethics, listing the steps it will take to reduce bias in AI output. For instance, at Thomson Reuters, we employ data and AI ethics principles “to maintain meaningful human involvement, and design, develop, and deploy AI products and services that are reliable and consistent, empower socially responsible decisions, and use data in a manner that treats people fairly.”
As the legal profession has strict ethical guidelines, if AI tools do not align with these standards, law firms may unwittingly commit ethical violations. A law firm should ask its AI provider how it ensures compliance with ethical rules and client confidentiality requirements. The provider should have all the necessary resources and infrastructure to keep users in compliance with current guidelines and be ready to adapt to future regulatory changes.
Assessing potential providers
More than comparing features or pricing, scrutinizing how each provider will protect your clients' information is paramount. Robust security and confidentiality practices aren't optional — they're fundamental criteria.
Ensure data security and confidentiality
Legal professionals regularly handle sensitive information, such as clients’ personal and financial records or companies’ projected sales and expense budgets. Misusing or accessing such data without authorization can have significant legal and ethical consequences. Inadequate security measures that result in data breaches could lead to malpractice claims.
Law firms should thoroughly assess a provider’s data security levels and confidentiality practices before entrusting it with any discretionary information. These practices include:
- Client confidentiality procedures. Lawyers have ethical and legal obligations to protect client information. AI systems lacking adequate security measures risk exposing this data to hostile actors and putting the law firm in violation of its obligations.
- Compliance levels. Law firms typically adhere to a number of data protection regulations, such as the General Data Protection Regulation (GDPR) and the California Consumer Privacy Act (CCPA). AI providers should meet all these standards and, even better, comply with additional regional or international guidelines.
Potential red-flag data security scenarios
The following are common data security and confidentiality scenarios, along with ideal or poor responses from AI providers that are sufficient to trigger a “red flag” warning.
Scenario 1
Your firm is concerned about where data generated by an AI system will be stored and who has access to it, within your organization as well as at the AI provider.
Red flag answer
The provider avoids sharing data storage details and instead offers vague references to “industry-standard security” without specifics.
Scenario 2
Your firm wants to know whether data created from communications and transactions is used for training the AI model. If so, who retains ownership and rights to information that law firm professionals enter into the AI solution? Who owns the material the AI system generates?
Red flag answer
The provider gives vague, evasive answers about client data usage and ownership, refers to industry-standard protections, and does not allow users to opt out of training.
Scenario 3
Your firm wants its lawyers and clients to be able to delete any data upon request. What is the AI vendor’s data retention policy?
Red flag answer
The provider does not offer a clear deletion process and retains data for unnecessarily long periods.
Scenario 4
Your firm wants to know whether an AI system will share data with third-party vendors or service providers — and if so, how this data is shared, who these parties are, and what type of data they receive.
Red flag answer
The provider is reluctant to disclose details about third-party involvement and may even show hostility when asked, claiming the information is privileged.
Determine accuracy and reliability
The use of inaccurate and unreliable AI tools can lead to flawed legal advice, error-ridden document drafting, and potential ethical violations. A law firm must assess whether its prospective AI provider has an acceptable, if not superior, level of system accuracy.
To make this judgment, your firm should examine the AI system’s past performance and understand the steps the provider takes to enhance future performance.
Step 1: Evaluate AI and determine error rate
An AI provider’s error rate is a vital statistic. The higher the rate, the more concerned a potential system user should be, and a vendor that won’t share benchmarking data at all is one to avoid.
Here’s an example of how a single error could reverberate across a law firm. Let’s say an AI system drafts a contract including some wrongly inserted and inapplicable clauses. For one thing, this error renders the AI system pointless, as a legal professional could have to spend as much time correcting the document as they would have spent drafting it themselves.
But there are wider implications. Say the contract’s errors are missed by lawyer review, and the contract gets disseminated among the parties in an acquisition. Then the error is discovered weeks or months later, which could make the contract void and necessitate its re-drafting and re-signing by all parties. If this flawed contract delays or disrupts a transaction, or if the unexpected delay adds expenses to the deal process, the law firm risks losing its client or even facing litigation.
The law firm needs specific examples of AI performance history in the tasks most relevant to its practice areas, like contract review or legal research.
Step 2: Ensure regular system updates and improvements
The legal field is dynamic, with laws and regulations changing frequently. An out-of-date AI model may generate documents using obsolete and inaccurate information.
A law firm should have confidence in its provider and know the following:
- How it updates schedules and methods
- How it incorporates legislative and regulatory developments with its AI model
- Which resources it draws upon to keep databases current
- How frequently and consistently it offers training or user support for necessary system upgrades
Step 3: Consistently improve the quality and breadth of training data
One thing is clear: AI is only as good as the data it’s trained on. If the provider trains its AI on limited, outdated, biased, or inaccurate information, the output will reflect those flaws.
Law firms should inquire into the sources of the provider’s training data and determine:
- Does the system draw upon a wide range of legal cases and precedents?
- Does the provider have a dedicated trust team to assess the system’s output?
- Is there a “human in the loop” — which refers to human involvement in machine-learning workflows to reduce errors and improve model performance — quality assurance process implemented for responsible AI use?
- What is the accuracy of the information, and where does it come from — a proprietary and well-regarded information source, or scraped information from various websites, some of questionable legitimacy?
More questions to ask potential AI providers
How do you ensure the accuracy of data used by the AI system?
There should also be measures in place to comply with ethical and legal obligations regarding data privacy.
Can you explain the AI’s decision-making process?
It’s vital for legal professionals to see how the AI arrived at a particular conclusion to ensure its validity. They need to be confident in defending its use in court if necessary.
Do you provide proof of concept and offer live testing?
Prospective clients should see live demonstrations of the AI system and access it before purchasing. The provider ideally would offer a trial usage period. The AI provider should provide customer reviews and case studies upon request, as well as lists of their direct competitors, possibly with benchmark data that compares the AI provider’s system to those of its biggest rivals.
Confirm legal expertise and training
Legal AI providers often claim their systems have expertise. However, what truly matters are the specifics of this expertise.
For example, CoCounsel provides results grounded in content from Westlaw and Practical Law — that is, legal content enhanced by more than 1,200 expert, bar-admitted, full-time lawyers. Our AI experts apply patented techniques for data processing, test the latest large-language models (LLMs), and deploy vetted technology to ensure law firms have the right tools.
What should a law firm ask its potential AI provider to ensure an acceptable level of quality?
- Find out what specific legal processes the AI solution supports. Lawyers do not want to log into a new AI system designed to reduce their workload only to discover it cannot support their current drafting technology.
- Ask for examples of legal firms and departments currently using the solution.
Other things to determine:
- What training does the provider give clients, and how much ongoing support should a client expect from them?
- How does user feedback affect operations? What’s the process for updating and improving the AI solution if a substantial number of users request this?
- What assurances does the provider give for data privacy and confidentiality concerns?
- If the AI system produces errors that negatively impact a law firm’s output, what responsibility does the provider bear?
Making the decision
After a lengthy search, your law firm has found an ideal provider of an ideal AI solution. Now it’s time to determine the cost of implementing the system, along with its scalability.
Evaluate pricing plans and scalability
The AI provider will typically offer multiple payment options. They may use a subscription-based service, where users pay a monthly or annual fee. If they have a per-use plan, they charge a specific amount each time a legal professional makes a query to the AI system.
Should your law firm choose a subscription model, you’ll need to fine-tune how the subscription works. Will each lawyer have their own subscription, or could individual subscriptions encompass a set number of users at the firm? Are there usage limits for each given subscription, such as a ceiling on requests per day or at specific times of day?
There are also pricing tiers. For example, a 100-person law firm may put its 40 partners and senior associates on a costlier, all-access tier of system usage, while the rest of the firm is confined to a lower tier that limits usage.
Find out what complementary services a subscription provides. Are there additional fees for user onboarding and training times, or for technical support? The AI provider may have different pricing models for full-service AI and for its bespoke usage agreements. The latter includes API access pricing, where the AI provider charges users for integrating the AI solution into existing systems, and value-based pricing, in which fees are tied to cost savings achieved via the AI solution.
If the firm chooses the value-based pricing option, it puts even greater importance on assessing ROI. That is, how can the law firm determine what measurable savings it derives from using the AI system? The AI provider should provide case studies that assess the ROI of past clients, particularly for law firms of a similar size to the prospective client or with the same practice focus.
AI solutions should deliver time and cost savings that translate into measurable and shareable data, such as hours spent on a particular use case, which can affect pricing. If the AI system saves work hours for each associate, a value-added subscription price could subsequently reflect those savings.
Other pricing models include:
- Token-based pricing. This variation of usage-based pricing centers on the number of tokens — that is, units of text — processed in a specified period.
- Project-based pricing. In this model, plans have a flat fee for a specific project or use case, often with an agreed-upon end date or maximum number of hours.
- Enterprise agreements. These consist of customized pricing arranged for large-scale AI implementations, such as a law firm with multiple offices.
- “Freemium.” This is a cut-rate agreement in which basic features are free, but users pay for upgrades and advanced functionality.
- Hybrid models. Combinations of all the above are possible, typically centered on a base subscription that carries additional usage fees.
Scalability is key. Can the AI solution easily accommodate future increases in caseloads and data volumes? Is it possible to add or remove users easily if the law firm’s needs change? How will pricing change for notable usage increases or greater demands for data storage?
Ensure integration with existing workflows
Your firm has chosen an AI program, its provider, and a pricing plan. You’re ready to achieve substantial efficiencies and reduced costs. That is, until integrating the AI system proves to be such a headache that it adds massive time and expense to the rollout, delaying adoption and making some lawyers wary of using the system.
That’s why having a seamless integration with your existing legal tech stack is crucial. Even the most appealing AI programs lose value if they clash with your firm’s existing technologies. When you negotiate with an AI provider, determine how the system will interact with your current tech stack and what steps the provider will take to avoid a lengthy, frustrating integration.
The AI provider should explain:
- What’s involved in integrating the AI solution with a firm’s existing systems and workflows, along with an implementation timeline and any potential risks
- How the provider will work with the client’s internal technology team to ensure a smooth integration
- What potential additional costs could arise from implementing and customizing the AI solution
Further, many types of integrations will come into play.
Document and contract management
When an AI system links with such platforms as SharePoint or iManage, it can automatically generate, organize, and search legal documents. Integration with platforms like HighQ is essential for automating contract creation and review.
CoCounsel, for example, integrates research, analysis, and drafting in one seamless workstream, uniting its own AI legal assistant capabilities with Westlaw Precision, Practical Law Dynamic Tool Set, Microsoft 365, and document management system (DMS) platforms. Customers also report that CoCounsel reduces the time needed for document review and contract drafting by 63% (CoCounsel Core customer ROI survey, June 2024; CoCounsel Drafting customer survey, July 2024).
Research efforts
By aligning the AI system with comprehensive legal research tools like Westlaw, users gain access to a database of case law, statutes, and legal opinions. This capability allows a lawyer to work with confidence, knowing the AI system’s queries and research pull from authoritative content and top-tier legal expertise.
Data analysis and discovery
Connecting with data analytics tools enables the AI system to perform predictive analyses. E-discovery platform integration enhances the AI system’s ability to process and analyze all documents in litigation, negotiations, and investigations.
Relationship and communication management
When linked with customer relationship management (CRM) tools, the AI solution can streamline client interactions and enhance recordkeeping. Aligning with internal communication tools, including email, messaging platforms, and video conferencing, enhances internal operations.
Internal operations
Integrating AI with enterprise resource planning (ERP) software streamlines legal billing, financials, and other operational aspects within a law firm. Merging AI with project management tools such as Asana or Trello enables workflows to move efficiently.
Each integration should align with the law firm’s specific requests and meet all data security and compliance requirements.
Getting AI right the first time
Selecting an AI provider can be an overwhelming decision for a law firm to make. After all, there are so many aspects of AI usage to keep track of and so much information to absorb.
By asking the right questions of your potential AI provider and knowing what responses you need to hear, your firm will make its decision more easily and with greater confidence.
Finding the right AI solution is crucial for your firm. With decades of AI expertise here at Thomson Reuters, we ensure seamless integration into your workflows. CoCounsel, the professional-grade AI assistant connecting all that CoCounsel Legal comprises, is already being used by over 20,000 U.S. law firms and legal departments.
Enhance your firm's efficiency and effectiveness today with our comprehensive worksheet on choosing the right legal AI technology provider.
