How AI-Assisted Evidence Review Is Changing California Employment Litigation Strategy
What an Email Search, an Arbitration Record, and a Corporate Mediation Reveal About the Opportunities and Limits of AI-Assisted Litigation
By Joshua Milon | Managing Partner, Workers’ Rights Legal Group
Did an employee complain to an outside agency about unlawful workplace conduct? And if so, when?
For an employment attorney evaluating a potential retaliation claim, those questions can be central to determining what the evidence supports. In one matter I handled at Workers’ Rights Legal Group, the answer was buried somewhere among approximately 10,000 emails supplied by a prospective client.
I began with a smaller collection of approximately 370 emails, using an artificial intelligence platform to search for evidence of an outside complaint. Within minutes, the software identified a relevant email, provided its date, and linked directly to the original document.
I could then examine the email, verify its contents, and assess its significance to the potential claim. Instead of reviewing hundreds of messages one by one, I had a specific piece of evidence to investigate.
That experience illustrates how artificial intelligence is changing evidence review in employment litigation.
Electronic discovery can involve years of emails, personnel files, internal communications, and deposition testimony. A document buried in those records may contradict an employer’s explanation for termination, establish when management learned of a complaint, or expose an inconsistency in a witness’s account.
AI-assisted analysis can make those documents easier to locate and examine. But its value extends beyond speed.
The real opportunity is to use technology to investigate the evidence more thoroughly, test competing explanations, identify weaknesses, and develop stronger legal arguments.
There is also an important limitation. Finding a document is not the same as determining what that document proves.
At Workers’ Rights Legal Group, the California plaintiff-side employment firm I founded in 2013, we have incorporated AI-assisted analysis into case evaluation, discovery, arbitration preparation, and settlement negotiations.
Our experience, also documented in Knool’s September 2026 case study of Workers’ Rights Legal Group, has shown how useful these tools can be when working with extensive evidentiary records. It has also reinforced why professional judgment, source verification, and an understanding of the governing law remain essential.
The technology can help us find information that might otherwise require substantial time to locate. Our responsibility is to determine whether that information supports the legal conclusions we are considering.
Evaluating a Case Means Looking for Evidence That Could Hurt It
In many wrongful termination cases, the fact that an employee was fired is undisputed. The disagreement concerns why.
An employee may contend that termination followed a discrimination complaint, a request for disability accommodation, or a report of unlawful conduct. The employer may point to performance problems, attendance, misconduct, or restructuring.
California generally recognizes at-will employment, but employers cannot lawfully terminate employees for reasons prohibited by law.
The challenge is determining what actually motivated the employer’s decision and whether the available evidence supports a claim.
That requires more than reviewing documents that favor the employee’s account. A thorough evaluation must also address evidence that could undermine it.
When assessing a prospective case, I want to know what the strongest argument against our position might be. If the employer claims an employee was terminated for performance problems, I want to examine the records supporting that explanation as carefully as the evidence suggesting retaliation or discrimination.
AI-assisted review helps us conduct that investigation. We can search communications, reconstruct timelines, compare accounts, and direct attention to inconsistencies that warrant further examination.
In one whistleblower matter, the documentary record established that an employee sent a particular email and was terminated approximately 60 days later.
That timing was relevant to the retaliation allegations, but it did not answer every question. We still needed to evaluate what the employee reported, who knew about the complaint, what happened afterward, and whether the employer’s explanation was consistent with the surrounding evidence.
The same approach can reveal weaknesses in an employee’s account.
Early in my career, I was more inclined to accept a client’s recollection of events before testing it against the available documents. Experience taught me that a client’s sincerity and the accuracy of every detail are not necessarily the same thing.
I’ve encountered situations in which a client’s recollection did not match the documentary record. In one matter, AI-assisted review identified a discrepancy involving an important date before we incorporated that information into a legal complaint.
The inconsistency did not automatically defeat the claim. It gave us a reason to investigate further.
That distinction matters. A discrepancy may have a reasonable explanation, or it may reveal a significant weakness in the proposed legal theory. Its importance depends on the evidence and the circumstances.
Identifying these issues early allows attorneys to address them before they become central to discovery, deposition testimony, or motion practice.
It can also influence whether a firm accepts a case.
When the evidence does not support a proposed claim, that is information we need to know before committing the client’s time and the firm’s resources to litigation.
“One of the most valuable ways I use AI in California employment litigation is to challenge my own assumptions. Finding evidence that supports a client’s claim is important, but identifying evidence that could undermine it is just as critical to making sound legal decisions.”
For me, one of the most useful applications of AI is not finding more evidence that supports what I already believe. It is helping me question those beliefs and determine whether the record supports them.
Why Evidence and Chronology Matter Under California Retaliation Law
The importance of locating a particular email or establishing a precise chronology becomes clearer when examining California’s legal standards for retaliation.
Different statutes protect different forms of employee conduct, and the applicable evidentiary requirements depend on the cause of action.
For example, California Labor Code section 1102.5 protects employees from retaliation for certain disclosures involving suspected violations of law.
Under Labor Code section 1102.6, an employee pursuing a qualifying whistleblower retaliation claim must demonstrate by a preponderance of the evidence that protected whistleblowing was a contributing factor in the adverse employment action.
Once that showing is made, the employer bears the burden of proving by clear and convincing evidence that it would have taken the same action for legitimate, independent reasons.
The California Supreme Court confirmed this framework in Lawson v. PPG Architectural Finishes, Inc. (2022) 12 Cal.5th 703.
Other retaliation claims arise under the California Fair Employment and Housing Act (FEHA).
Government Code section 12940(h), for example, prohibits retaliation against employees for opposing practices forbidden by FEHA or participating in protected proceedings.
These causes of action have distinct legal requirements. The burden-shifting framework applicable to a Labor Code whistleblower claim should not automatically be applied to a FEHA retaliation claim.
But both illustrate why the timing of an employee’s complaint, management’s knowledge, and subsequent employment decisions can be important to the analysis.
A single email may establish that a complaint occurred on a particular date. Additional evidence may be needed to determine whether the communication constituted protected activity, whether the relevant decision-makers knew about it, and whether retaliation contributed to the adverse action.
A close sequence of events may support an inference of causation, but timing alone does not resolve every question.
Attorneys must examine the chronology alongside the employer’s explanation, the available testimony, and the applicable legal standards.
AI-assisted review can help assemble that chronology and locate relevant communications across extensive records.
But the legal analysis does not end when the timeline is complete.
The attorney must determine which facts matter, whether they satisfy the elements of the claim, and how the evidence should be presented.
The value of technology depends, in large part, on the quality of the legal questions guiding the search.
Reexamining the Evidence During Arbitration
One of my most significant experiences using AI came during an arbitration that I initially assessed as having a 95 percent likelihood of an unfavorable outcome.
That was my professional assessment of the case at the time, not a statistical prediction generated by software.
The opposing party had made no monetary settlement offer, and the matter proceeded to arbitration.
After the hearing, we faced the task of preparing a closing brief from an extensive evidentiary record. The challenge was to identify which testimony, admissions, and documents best supported our arguments.
AI-assisted review gave me another way to examine the record.
Rather than relying exclusively on a sequential review of transcripts and exhibits, I could search for particular issues and identify connections among different sources of evidence.
During that process, the platform highlighted an admission by an HR representative that warranted closer examination.
I returned to the underlying testimony, reviewed it in context, and considered how it related to the issues before the arbitrator.
The admission deserved greater emphasis than I had initially given it. We incorporated that evidence into the arguments developed for our closing brief.
The arbitration ultimately concluded with a favorable award.
Preparing the closing brief required approximately 40 hours, compared with my estimate of 80 hours using a more traditional process.
The time savings were meaningful, although those figures reflect my experience and estimate in that particular matter rather than a controlled comparison.
What mattered more to me was the opportunity to revisit the evidence and improve the quality of our analysis before submitting the brief.
A deposition transcript may contain hundreds of pages, and the significance of a particular answer may depend on documents introduced elsewhere or testimony from another witness.
When reviewing extensive records, it is possible to understand the case well and still overlook a connection worth exploring.
AI-assisted analysis provides another way to find those connections.
That does not mean the technology develops the legal argument. It identifies material for the attorney to evaluate.
I still had to determine what the HR representative’s admission established, why it mattered under the applicable law, and how it should be incorporated into our closing argument.
The distinction is central to how I use AI in litigation.
I am not asking the technology to decide the case for me. I am using it to examine the evidence more thoroughly so that I can make better-informed strategic decisions.
Testing a Corporate Defendant’s Position During Mediation
Settlement negotiations present a different challenge.
During mediation, an attorney may need to evaluate an opposing party’s characterization of the evidence while discussions are underway.
In one matter, our firm faced a corporation valued in the trillions of dollars.
When the mediator relayed the company’s position concerning the underlying facts, I used AI-assisted review to search the evidentiary record and determine whether that position matched the available documents.
Having the ability to locate relevant information during negotiations gave me another way to test the corporation’s assertions against the record.
That capability is particularly important when litigating against defendants supported by extensive legal departments and outside law firms.
Large corporate defense teams may have substantially more personnel available to organize documents, analyze testimony, and prepare litigation strategy.
Our firm operates with five attorneys.
We cannot always devote the same staffing resources to document analysis as a major corporate defense team. AI-assisted review allows us to examine extensive records more efficiently while concentrating our attorneys’ time on the legal questions those records present.
I do not view that as eliminating the resource differences between a smaller plaintiff-side firm and a large corporate defendant.
It does, however, give us another way to investigate factual disputes and respond to opposing arguments without requiring the same number of people to perform every review task.
A settlement position supported by a specific document deserves to be evaluated in light of that evidence. A position that appears inconsistent with the record warrants further investigation.
The technology helps make the relevant information accessible.
The attorney must still determine how it affects liability, damages, witness credibility, litigation risk, and settlement strategy.
That is where experience becomes essential.
A Real Citation Can Still Support the Wrong Conclusion
One of the most important lessons I’ve learned from using AI in litigation concerns the difference between finding an authentic source and accurately interpreting it.
Much of the discussion about legal AI has focused on hallucinations, including fabricated cases and nonexistent citations.
Those are serious problems, but they are not the only risks.
An AI system can provide a real citation and still reach the wrong conclusion about what the cited material proves.
I’ve encountered situations in which AI-generated analysis cited actual deposition testimony but attributed greater significance to that testimony than the witness’s answers justified.
In one example, the software characterized testimony as supporting an allegation that a supervisor had harassed an employee.
When I examined the underlying exchange, the answers established a supervisory relationship and workplace disagreements. They did not establish the alleged harassment.
The software had directed me to genuine evidence. Its interpretation of that evidence was the problem.
“AI can identify a genuine document and still reach the wrong conclusion about what it proves. In employment litigation, I have to examine the underlying testimony, consider the surrounding evidence, and determine whether the facts actually support the legal argument. A real citation is not a substitute for legal judgment.”
That distinction can be difficult to recognize if the attorney relies on the generated summary without returning to the source.
A deposition answer may appear significant when presented in isolation. Read alongside the preceding questions, subsequent testimony, and other evidence, it may support a narrower conclusion or a different interpretation.
I have encountered similar problems when an AI response directs me to a lengthy transcript or document that is generally relevant to the issue but does not establish the particular proposition attributed to it.
For that reason, I do not treat a citation as confirmation that an AI-generated conclusion is accurate.
I examine the cited material, read the surrounding context, and compare it with other evidence before deciding how it should be used.
This is particularly important in employment litigation, where the legal significance of testimony may depend on who made a statement, what that person knew, when an event occurred, and how the evidence relates to the elements of the claim.
A system may identify testimony related to workplace conflict without establishing unlawful harassment. It may locate a complaint that preceded termination without establishing retaliatory causation.
These are legal distinctions that require careful analysis.
AI can reduce thousands of documents to a smaller group warranting further investigation. That is a substantial advantage.
But an attorney cannot assume that the software’s explanation of those documents is accurate simply because the documents exist.
For me, verification means more than confirming a citation. It means determining whether the underlying evidence actually supports the conclusion I intend to draw from it.
That responsibility becomes especially consequential when the evidence will be used in a complaint, motion, arbitration brief, or other formal legal submission.
California’s Guidance on Responsible AI Use
California’s professional responsibility guidance reinforces the importance of maintaining independent legal judgment when using AI.
In May 2026, the State Bar of California updated its Practical Guidance for the Use of Generative Artificial Intelligence in the Practice of Law.
The guidance addresses how existing professional obligations apply to the use of generative AI, including competence, confidentiality, communication, supervision, and candor to the court.
It emphasizes that attorneys must understand the capabilities and limitations of the systems they use and exercise independent judgment when reviewing, verifying, and correcting AI-generated work.
Those responsibilities extend beyond detecting nonexistent legal authorities.
Attorneys must also evaluate whether generated arguments accurately represent the evidence, whether cited documents support the propositions attributed to them, and whether the analysis reflects the governing law.
Confidentiality requires similar attention.
Employment litigation may involve sensitive client communications, personnel files, medical information, and other confidential records. Before using AI tools to process that material, attorneys must evaluate how the systems collect, store, use, and disclose information.
A platform designed for legal work may offer features relevant to confidentiality and document management, but those features do not eliminate the attorney’s responsibility to assess the protections in place.
The State Bar’s updated guidance also addresses emerging agentic AI systems, which can perform multistep tasks with greater autonomy.
These capabilities may expand the ways law firms use AI for document review, discovery preparation, and client intake.
But greater automation does not reduce the need for meaningful attorney supervision.
The guidance explains existing professional obligations rather than replacing the California Rules of Professional Conduct.
In my practice, the principle is straightforward: AI can assist the work, but the attorney remains responsible for the work product.
From AI-Assisted Drafting to Persuasive Legal Arguments
Locating evidence is only one part of litigation preparation.
Developing that evidence into a persuasive legal argument requires another level of analysis.
At Workers’ Rights Legal Group, we also use AI to help organize and prepare initial drafts of litigation materials.
Those drafts are starting points, not finished legal work.
An initial draft may accurately summarize testimony but fail to explain why it matters under California employment law.
It may emphasize one legal theory while overlooking another that is supported by the record. It may also advance an argument without adequately anticipating the opposing party’s response.
These are issues an attorney must identify and address.
Our responsibility is to determine which legal theories are supported by the evidence, how the applicable standards should be applied, and which arguments can withstand scrutiny.
That may require additional research, further examination of the record, or a different approach to presenting the facts.
My experience with discovery, depositions, motion practice, arbitration, and trial informs those decisions.
An attorney familiar with the complete record can recognize when a draft places too much weight on a particular statement, overlooks unfavorable testimony, or fails to account for an important legal distinction.
The ability to produce a coherent draft quickly is useful. It should not be confused with the ability to produce a persuasive, legally supported argument.
I see AI-assisted drafting as a way to reduce some of the time spent organizing information and developing an initial structure.
That creates more opportunity to focus on the reasoning, evidence, and strategy that ultimately determine the strength of the work product.
The goal is not simply to produce legal documents faster. It is to use the available time and information more effectively.
How Practicing Attorneys Can Influence Legal AI Development
Our experience has also shown me the importance of involving practicing attorneys in the development of legal technology.
We use Knool, an AI platform designed for litigation, and provide feedback based on our work handling California employment cases.
Representatives from the company meet periodically with our attorneys and staff to discuss evidence-review workflows, case evaluation, document organization, and opportunities to improve the software.
Our feedback has helped inform product improvements.
More recently, those discussions have included consultation and client-intake capabilities that we are exploring for potential future use.
I believe this kind of collaboration is valuable because the demands of actual litigation are not always apparent from a technical demonstration.
An employment case may involve years of communications, conflicting witness accounts, multiple legal theories, and an evidentiary record that changes throughout discovery.
Attorneys need tools that can help them navigate those materials while preserving access to the original sources.
They also need to recognize when a generated response appears convincing but does not accurately reflect the evidence.
Those limitations become apparent through practical use.
When attorneys can explain where a system helps, where it creates problems, and what additional capabilities would improve their work, that feedback can contribute to more useful technology.
For me, the relationship between legal practice and AI development is not simply about adopting new software.
It is about ensuring that the technology addresses the needs of practicing litigators while respecting the professional responsibilities that govern our work.
The Broader Implications for California Employment Litigation
Artificial intelligence is changing the practical economics of evidence review.
For years, one of the clearest advantages held by large corporate defense firms has been the ability to assign substantial personnel and financial resources to extensive document productions.
A smaller plaintiff-side firm may have access to the same evidence but fewer people available to examine it.
AI-assisted review can reduce some of that disparity.
A five-attorney firm can now search extensive records, reconstruct timelines, identify potentially relevant testimony, and investigate opposing factual assertions more efficiently than traditional manual review alone may allow.
The implications extend beyond staffing and speed.
Employment litigation requires attorneys to distinguish between allegations and evidence, evaluate competing explanations, anticipate defenses, and determine whether the record satisfies the applicable legal standards.
Faster access to information can support each of those activities, provided the evidence is examined critically.
Our experience has also shown why greater access to information does not eliminate the need for experienced counsel.
AI may locate an email, identify an admission, or suggest an argument. It may also overlook context, overstate the significance of testimony, or reach a conclusion the underlying documents do not support.
The attorney must determine whether the evidence is reliable, what it establishes, and how it should influence the case.
The central question in California employment litigation remains unchanged: Does the available evidence establish the facts necessary to support the claim under the governing law?
What is changing is the speed and efficiency with which attorneys can investigate that question.
I expect the firms that benefit most from AI will not necessarily be those that automate the greatest number of tasks.
They will be the firms that use the technology to examine evidence more thoroughly while maintaining rigorous verification, sound legal judgment, and control over litigation strategy.
The most consequential change I have seen is not that AI can perform the work of an employment attorney.
It is that AI can help an attorney examine a larger and more complex evidentiary record, revisit issues that deserve closer attention, and devote more time to determining what the evidence actually proves.
That is where I see its greatest value.
Key Takeaways
- AI-assisted review can make extensive evidence collections more manageable. Targeted searches allow attorneys to locate potentially relevant communications, records, and testimony without relying exclusively on sequential manual review.
- Early case evaluation should include unfavorable evidence. Identifying contradictions and documentary weaknesses before litigation can lead to more informed decisions about whether and how to proceed.
- Chronology is central to many employment retaliation claims. The timing of complaints, management knowledge, disciplinary actions, and termination must be evaluated under the legal standards governing the particular claim.
- AI-assisted analysis can support arbitration and mediation strategy. The ability to revisit testimony and retrieve relevant documents can help attorneys refine arguments and evaluate opposing factual positions.
- An authentic citation does not guarantee an accurate interpretation. Attorneys must examine the underlying evidence to determine whether it supports the conclusion attributed to it.
- AI-assisted drafting requires substantive attorney review. Organizing evidence and generating an initial argument are different from developing a legally supported and persuasive final work product.
- Professional responsibility remains with the attorney. California’s guidance reinforces the importance of competence, confidentiality, verification, and supervision when using AI.
- Technology can help smaller plaintiff-side firms examine complex records more efficiently. These capabilities may reduce some staffing demands without eliminating the resource differences between smaller firms and large corporate defendants.
- Practicing attorneys can help shape legal AI development. Feedback drawn from actual litigation experience can improve the usefulness and reliability of tools designed for legal practice.
Related Analysis
The Hidden Role of Insurance in California Employment Litigation: How Policy Structure Shapes Settlement Strategy
An examination of how insurance structure, defense costs, deposition testimony, and trial readiness can influence settlement timing and litigation strategy.
California Employee Rights in Wrongful Termination Cases: 2025 Legal Update
An analysis of wrongful termination, retaliation, and evolving legal protections affecting California employees.
California Arbitration Ruling Signals Tougher Scrutiny of Language Access and Electronic Signatures
An examination of a California employment arbitration ruling involving language access, electronic signatures, and the enforceability of arbitration agreements.
This article is for informational purposes only and does not constitute legal advice. The use of artificial intelligence does not guarantee any particular legal outcome, and prior results do not guarantee a similar outcome. Each employment case depends on its specific facts, evidence, and applicable law.
