Executive Summary
- At RelFest Chicago 2026, Relativity showed just how quickly legal AI is becoming part of everyday work, not just something being tested. Today, 19.5 petabytes of data are under management, and AI has touched more than 54% of reviewed documents.
- Relativity aiR now reaches across the matter lifecycle, from early case assessment and review through production and drafting with its Gavel/Word integration. Avalon helps clients bring these capabilities together into a connected, defensible workflow rather than relying on a collection of disconnected tools.
- Practical AI capabilities available today include natural-language matter exploration through Assist, AI-assisted review grounded in established TAR and proportionality standards, and Custom Analyses for less traditional evidence such as handwriting, images, and source code.
- Security and governance remain central to the platform, with permission-aware agents and no model training on user data. At the same time, questions around privilege and consumer AI tools make clear that organizations still need thoughtful, client-specific guardrails.
- The biggest takeaway: Successful AI adoption is about people and process as much as technology. Law firms, corporate legal teams, and service providers like Avalon all have a role to play. Looking ahead, claiR (early 2027) and reusable “skills” point toward a more conversational future built on the foundation that’s already in place.
At RelFest Chicago 2026, the keynote painted a picture of an industry moving beyond AI experimentation and into practical, governed execution. As a trusted Relativity Silver Provider partner, Avalon saw a clear message for legal teams: Meaningful value does not depend on waiting for a distant, fully autonomous future. Relativity is already embedding AI across the legal data lifecycle – from early case understanding and collection through review, production, and drafting. Avalon’s role is to help clients translate those capabilities into defensible eDiscovery and forensic collection workflows, then implement and operate them effectively on the ground.
Relativity reported 19.5 petabytes of data under management and said more than 54% of documents reviewed on the platform are now touched by AI technology. Those figures matter because they position AI as part of the operating model, not an isolated pilot. Relativity also pointed to measurable high-value outcomes in the billions, alongside faster case preparation, backlog reduction, and cost savings. The platform’s current direction is therefore less about adding novelty and more about making legal work easier to start, safer to scale, and more connected from evidence to outcome.
From AI Features to an End-to-End Legal Data Intelligence Platform
One of the keynote’s clearest themes was continuity. Relativity aiR is intended to support the full matter lifecycle rather than force teams to move between disconnected AI tools. Early case assessment can surface the shape of a matter sooner. Review capabilities can help prioritize, classify, and understand documents at scale. Production remains connected to the same governed data environment. Drafting can then build from the evidentiary record instead of beginning as an entirely separate activity.
For Avalon clients, that continuity is important. Every handoff between systems can introduce delay, duplicated effort, or uncertainty about the source of an answer. Avalon helps clients design the full workflow – from collection strategy and data handling through review, production, and downstream work product – so the platform supports the legal process rather than becoming another disconnected tool. A connected implementation creates a clearer chain from data to decision while keeping permissions, auditability, validation, and matter context consistent throughout the engagement.
What Teams Can Use Now
Natural-language access to matter knowledge
Assist brings natural-language interaction into Relativity, reducing the distance between a legal question and the documents that can answer it. Users can ask questions without first translating every issue into a highly technical search workflow. Early insights dashboards and document summaries help teams orient themselves quickly, identify themes, and decide where deeper analysis is warranted. Avalon can help clients identify the right questions, configure the supporting workspace and data, train users, and establish escalation and validation practices. The benefit is not that search expertise becomes irrelevant; it’s that more people can begin exploring a matter sooner while experienced specialists refine and validate the work.
AI-assisted review with defensibility in view
Relativity’s emphasis on explainable, auditable output stood out. The Judicial Panel reinforced that courts do not necessarily need an entirely new legal framework for every AI-assisted discovery tool: Established principles governing technology-assisted review (TAR), reasonableness, and proportionality remain highly relevant. In legal work, a fast answer is useful only when the team can understand how it was produced, test its quality, and defend the process. Built-in validation, metrics, traces, and audit mechanisms support that requirement. Human-in-the-loop controls remain essential, particularly when a system recommends or initiates consequential action. The operating principle is straightforward: Use AI to accelerate judgment, not to bypass it.
This is also where disciplined implementation matters. Teams should benchmark outputs against agreed quality standards, define approval points, and match the technology to the problem. Proportionality should guide those choices: The depth of validation, review, and documentation should reflect the stakes, volume, complexity, and burden of the matter.
Avalon helps clients make those decisions in the context of real matters, existing review protocols, staffing models, and risk tolerances. A single prompt may be enough for a straightforward summary. A repeatable workflow is better for a defined process. An adaptive agent may suit a more fluid, multi-step task. Selecting the simplest approach that reliably solves the problem can improve efficiency, control, and defensibility.
Custom analysis without a coding project
Custom Analyses gives teams a way to extend AI to matter-specific questions without building an application from scratch. Examples discussed at RelFest included image classification and transcription of handwritten notes, but the larger lesson is flexibility. Legal data rarely arrives in one predictable form. Avalon’s forensic collection and eDiscovery teams can help assess the data, preserve appropriate context, prepare it for analysis, and configure focused workflows for the unusual formats and fact patterns that often drive cost and risk.
This capability also opens practical paths for adjacent use cases. Relativity highlighted growing investment in data breach response and Freedom of Information Act matters, along with analysis of images, source code, and other complex content. Expanded collection integrations, including enterprise platforms and AI-generated content, help bring more of that evidence into a governed workflow. Avalon can bridge the gap between the source system and the Relativity workspace by advising on collection method, preservation, processing, data mapping, and downstream review design. For clients, the opportunity is to apply a familiar, defensible operating framework to new categories of legal and investigative work.
A more direct bridge from evidence to work product
The integration of Gavel’s AI-powered drafting tools within Microsoft Word addresses a persistent gap in legal workflows: The separation between evidence review and document creation. Keeping drafting connected to the underlying record can reduce friction while preserving citations and source relationships. Lawyers can move from identifying relevant facts to developing work product without losing the path back to the documents that support each proposition.
Security and Control Are Part of the Product
RelFest also reinforced that security cannot be bolted on after an AI workflow is designed. Relativity’s models operate in secure cloud environments, with user inputs not retained for model training. Agent harnesses limit access to the data and tools a user is permitted to reach. That permission-aware approach is especially important as AI evolves from answering questions to taking steps across systems. Avalon can help clients map those controls to collection procedures, workspace permissions, user roles, data-transfer practices, retention requirements, and matter-specific confidentiality obligations before the workflow goes live.
That message is timely. Consumer-grade AI tools may be easy to access, but their privacy, retention, training, and contractual practices can vary. The Judicial Panel also highlighted unsettled questions about whether prompts, AI-generated materials, and related work product will receive privilege protection in every circumstance. Sensitive legal data therefore demands clear boundaries, traceability, and control, as well as informed review of applicable terms of service. The value of an enterprise legal AI platform is not just model performance. It’s the surrounding architecture that governs what the model can see, what it can do, when it must ask for approval, how prompts and outputs are handled, and how its work can be reviewed later.
Adoption Is a People-and-Process Challenge
Technology alone does not create a successful legal AI program. The Legal Data Intelligence State of the Union showed why leadership cannot be assigned to one group. Law firms contribute cross-industry experience and competitive investment; corporate legal departments bring direct ownership of data, risk, and internal processes; technology providers supply platforms and tools; service providers translate capabilities into repeatable operations; and regulators shape the guardrails. Relativity’s community now spans more than 100,000 users in over 40 countries, across private practice, corporations, service providers, and the public sector. That breadth reinforces why deployment cannot rely on a single template. Each organization needs clear objectives, matter-appropriate controls, trained users, and an operating model that connects legal judgment, data ownership, technical execution, and governance.
For Avalon, the implementation lesson is to bring the full ecosystem into the workflow early. Attorneys can define the legal questions and risk thresholds. Corporate legal teams can establish business context, data ownership, and governance requirements. Avalon’s consultants, review professionals, forensic specialists, and Relativity implementation team can translate those requirements into defensible workflows, establish collection and integration requirements, configure the platform, and support users through launch and refinement. Regular measurement then turns adoption into a learning cycle: Compare results, investigate exceptions, refine instructions, and document what changed. This collaborative discipline is what converts a promising feature into a trusted part of everyday legal service delivery.
A Product Roadmap Grounded in Today’s Work
Governance will also vary by jurisdiction. RelFest contrasted the European Union’s risk-based regulatory model with Australia’s more pragmatic reliance on existing law. For global organizations, AI governance therefore cannot be reduced to one universal checklist.
Avalon can help clients document the data sources, jurisdictions, users, purposes, controls, and approval requirements associated with a workflow so that implementation decisions reflect both platform capability and the client’s legal environment. Relativity’s signature new product, claiR (in Advanced Access now, but slated for release early in 2027) points toward a simpler, lawyer-centered conversational experience in which users can ask questions, draft reports, and apply predefined or custom skills while administrators retain visibility into the work. As these capabilities mature, Avalon can help clients evaluate fit, prepare workflows and data, and incorporate new functionality without disrupting established controls. The direction is promising, but today’s available tools already provide a practical foundation for measurable improvement.
The roadmap also describes skills as repeatable units of expert work. In practical terms, skills can capture a proven process so that it can be used consistently, shared across a team, and improved over time. External connections through Model Context Protocol servers could eventually bring additional knowledge and systems into those workflows. Future capabilities mentioned included personally identifiable information (PII) detection and contract analysis. These developments merit attention, but they should be evaluated against actual use cases, quality requirements, and governance controls rather than treated as reasons to postpone current initiatives.
Relativity’s Advanced Access programs are an important part of this roadmap. They give customers and partners a structured way to test emerging functionality against real legal work and help shape the product before broader release. The broader release record – more than 130 major capabilities and 120 bug fixes year to date – also signals a product organization focused on continuous improvement, not periodic reinvention.
How Avalon Helps Legal Teams Move Forward
Organizations looking to translate RelFest’s message into action can start with three practical priorities:
- First, identify a bounded, high-value workflow where faster understanding would materially improve the matter, such as early issue identification, privilege-related analysis, or review prioritization.
- Second, define how accuracy will be measured and where human approval is required.
- Third, keep the work inside a secure, permission-aware environment with a complete audit trail.
Engaging Avalon early gives legal teams a partner that can connect these decisions to collection realities, eDiscovery strategy, Relativity configuration, and day-to-day execution.
Avalon helps clients make those choices with the full lifecycle in mind. As an experienced service provider, we combine strategic consulting with ground-level execution: Assessing use cases, designing forensic collection and eDiscovery workflows, preparing and moving data, configuring Relativity, validating results, documenting processes, and supporting adoption. We also help clients align those workflows with proportionality, privilege, confidentiality, security, and accountability requirements. The goal is not to deploy AI everywhere at once. It is to establish repeatable patterns for data preparation, workflow design, validation, oversight, and continuous improvement. A well-chosen first use case can build confidence and create the operational foundation for broader use.
RelFest made clear that the market is entering a new phase. Conversational interfaces, reusable skills, and increasingly capable agents will continue to evolve. Yet the most valuable insight for legal teams is that the foundation is already here: AI-assisted review, natural-language exploration, early insights, custom analysis, connected drafting, secure access controls, and defensibility mechanisms are available to improve real work today.
The organizations that benefit most will be those that move deliberately – pairing the technology with experienced people, clear processes, and measurable standards. The future roadmap is compelling, but it’s not a prerequisite for progress. Avalon’s takeaway from RelFest is to start with what works now, prove value in a governed way, and build from there. For teams ready to take that step, Avalon offers the Relativity experience, forensic and eDiscovery expertise, and hands-on implementation support needed to turn platform capability into dependable legal outcomes.
Ready to put legal AI to work in a secure, practical, and defensible way? Contact Avalon to discuss your goals and learn how our Relativity, forensic, and eDiscovery experts can help you identify the right use case, build a governed workflow, and turn AI capabilities into dependable legal outcomes.
This article was created by Martin Mayne, Avalon’s VP of eDiscovery.
