Reveal combines AI investigation with visual evidence analysis

Reveal combines legal evidence review, AI analysis and visual exploration of a document collection. Its portfolio includes several established discovery brands, while its current offer spans hosted software and a private deployment option. The useful question is how a team moves from patterns in the data to checked conclusions and review decisions, with the deployment and product packaging made explicit.

01 / ProductOne company spans several discovery tools and methods

Reveal’s company history records acquisitions including Brainspace, NexLP, Logikcull, IPRO and Onna. These names should be understood within the current company portfolio rather than counted as independent new companies. A buyer still needs to identify the exact product and deployment in the proposal; shared ownership does not establish identical features across every inherited offering.

The Reveal platform covers processing and document review, including text extraction, transcription, translation, search and AI models. Preparing usable evidence is part of the proposition. The quality of that preparation matters: an analysis system working from incomplete text may miss the point even when its later reasoning sounds coherent.

Reveal AI presents investigation, case strategy and review capabilities. The page describes an approved plan for case-strategy work, source-linked outputs and review methods combining generative AI with technology-assisted review, or TAR. These are distinct stages: exploring what happened is not the same as applying a settled document-coding protocol.

The analytics overview adds concept clusters, communication views and other visual ways to explore a collection. These can help a reviewer decide where to look next. A dense cluster or unusual communication pattern is a lead for investigation, however, not proof that the people or documents involved have the legal significance a team suspects.

02 / AudienceUnfamiliar evidence is a good reason to start visually

A plausible user is an investigation team receiving a mixed collection of email, chat and attachments without a reliable map of its themes. Visual exploration can help the team find terminology, groups of related documents and communications worth reading. AI questions can then focus on a narrower issue whose sources a reviewer is able to inspect.

Another fit is a review team with a recurring document category and a maintained classification process. Here the attraction is combining models and reviewer feedback, rather than treating each new matter as a blank slate. Reuse still needs validation: terminology, participants and the meaning of a relevant document can change substantially between collections.

OpenText is a useful comparison for broader enterprise information and discovery requirements. Intapp provides context for professional-services data and operational workflows. For Reveal, compare the complete journey from unfamiliar evidence to reviewed decisions, including the people who prepare data, interpret visual patterns and approve the final work.

03 / WorkflowA proposed investigation moves from patterns to a tested issue

Choose an authorised historical investigation with a known outcome and a mixed evidence set. Include routine communications using similar vocabulary to the issue of interest, as well as a small set of important documents that use unusual language. This is a proposed evaluation, not a report of testing. Keep the established findings separate from the team performing the initial exploration.

Start with the collection’s preparation and coverage. Confirm the intended custodians, time period, attachment handling and extraction status. Examine the conceptual and communication views for candidate themes, then open actual documents before naming a theme as significant. A visual grouping may reflect common formatting, a shared project or a repeated phrase rather than the disputed conduct.

Turn one promising theme into a bounded factual question. Ask for the relevant sequence of events, supporting documents and unresolved alternatives. Review any proposed investigation plan before it runs, with an explicit scope and stopping point. The purpose is to help the team inspect the record, not to delegate responsibility for determining which allegations are justified.

Compare the resulting account with the source documents and deliberately search for counterexamples. Distinguish a message’s author, recipient and mentioned person; these roles should not collapse into a single list of participants. Similarly, separate the time of the underlying event from the time someone later described it. Those distinctions can materially change the narrative.

If the team then wants a broader document review, define a separate coding protocol. Use known positive and negative examples, retain ambiguous cases for an expert and validate on material not used to refine the criteria. A promising investigative answer does not establish the recall or precision of a classifier applied to the entire collection.

Finish by giving a second reviewer the evidence trail and proposed conclusion without the exploratory conversation. Ask them to identify what is directly supported, what is inferred and what remains unknown. Record missed documents, correction effort and the cost of preparing the data. The desired outcome is a repeatable route from an unfamiliar collection to a defensible human judgement.

04 / PricingDeployment choices change what the commercial comparison includes

The Private Deployment page describes options across customer cloud and on-premises environments, with AI and analytics in the package. That is a different commercial and operational proposition from simply purchasing a hosted account. A broad inclusion statement on one deployment page should not be applied automatically to every hosted product or inherited brand.

Ask for two comparable scenarios if deployment flexibility is important. Use the same evidence volume, retention period, user roles and required capabilities. Include infrastructure, implementation, upgrades, support and the staff who will operate the environment. A licence model that avoids a particular data meter can still have substantial costs elsewhere in the service.

05 / DistinctionVisual exploration and model reuse can support different decisions

The combination of visual exploration and focused AI analysis gives a team several entry points into an unfamiliar record. A reviewer can begin with a concept, a communication pattern or a direct question. The value is in moving between those views while checking actual documents, rather than allowing a visually striking pattern to become an unsupported conclusion.

The AI product page also describes reusable models and hybrid review approaches. Reuse can preserve work invested in a prior classification task, but a model’s usefulness should be demonstrated on the new matter. The team should document what stayed the same, what changed and which examples were used to determine whether the transferred method still fits.

Reveal’s AI pledge says it will not use client data for model training without explicit permission. Preserve that qualification. It is a commitment about permission, not a universal statement that customer data is never used for any training purpose. Procurement should connect the wording to the actual service, configuration and agreement being considered.

06 / QuestionsConfirm the product boundary before interpreting a platform promise

The first question is exactly which products the proposal includes. The company’s acquisition history and broad website can make a portfolio look more uniform than the account a particular team will use. Ask for the intended workflow to be demonstrated in the proposed deployment, with the same modules, roles and support arrangements that would apply after purchase.

The second concerns infrastructure responsibility. For a private deployment, establish who handles updates, capacity, incident response and model-service configuration. Customer control is useful only when someone can operate it reliably. A team evaluating deployment options should compare the operating model as carefully as it compares data location or the initial licence quotation.

The third is evidence quality across formats and languages. Include poor scans, short messages and attachments in a representative test, and inspect what the platform actually extracted before assessing the generated answer. This article reviewed public sources without access to a live deployment. It does not validate vendor performance claims or establish review accuracy, feature parity or contractual limits for a particular customer.

07 / DecisionUse exploration to define the review you can justify

Reveal merits evaluation when a team needs to understand a complex evidence collection and then apply a controlled review process. Start with a known investigation and require the narrative to survive independent source checking. Choose the deployment only after its product scope, operating responsibilities and total cost are clear.