What Claira can do for your review

Claira helps teams run AI-assisted document review inside Nuix Discover. You write plain-language prompts; Claira reads document text (and the source image, audio, or video file when you switch the Scan as dropdown) and writes structured answers into Nuix fields you connect. Use this page as a map from “first prompt” to large-scale production runs.

Quickstart — prompts, generator, and templates

• Prompt box: Type what you want the model to do (for example, “Summarize in two bullets” or “Extract any agreement dates”). Save prompts you will reuse.
• Prompt Generator: When you need help wording or structuring a request, use the generator inside Claira to draft and refine instruction text before you save or run.
• Templates (Quickstart library): Start from proven patterns for relevance, privilege, PII, objective coding, and more, then adjust for your matter.
• Prompt History and PDF export: Every saved prompt keeps its version history — variations grouped into chains with word-level diffs and a case-wide Final mark — and any version downloads as a PDF, on its own or with the full revision history and its comments, for approvals and records.

See Prompting overview, Prompt templates and Prompt History.

Single scan (one document)

Single scan means: one open document + one saved prompt → one AI response written to your chosen field. Use this to test wording, validate output on tricky documents, and tune prompts before you scale.

See Single Review.

Bulk scan (many documents)

Bulk scan applies the same saved prompt across many documents at once (up to 999,999 documents per run). Results and task metadata land in connected fields; you monitor progress and outcomes in Bulk Tasks.

See Bulk Scan.

Multi-Code (one pass, several fields)

Multi-Code is for when a single instruction should fill multiple outputs in one pass—for example objective coding (date, title, type), or splitting and organizing extracted values across several fields during complex review. You configure multiple destination fields; Claira returns structured content for each.

See Objective coding (Multi-Code section).

Media scans (image, audio, video)

When productions include images, audio, or video, Claira supports media-aware scanning on every plan—useful for scans that must "see" the page, "hear" the recording, or "watch" the clip, not only extracted text. The Scan as dropdown next to every scan action has four options: Text (default, 1 token/doc), Image (5), Audio (10), Video (20).

See Media scans for the full list of accepted formats per mode.

Insights (one distilled, cited output)

Once a bulk scan has produced per-document results (or a field already holds them), Insights distills them into one output—a chronology, relevance summary, PII report, or investigation narrative. Every claim carries an inline citation back to its source document, you can revise the result iteratively, and the finished Insight downloads as a PDF listing all source documents.

See Insights.

Insight templates (one-click starting points)

Insight templates offer five common case outputs — Chronology, Early case assessment, Key documents, Issue analysis, and Privilege log — as a row at the top of the New insight card. Each knows which prompt template produces the data it needs: a green dot means the case already has it and the insight is one click away; otherwise Claira opens Bulk Tasks with the prompt filled in. A finished bulk task that ran a matching prompt offers the insight directly.

See Insight templates.

Agent Mode (case-scoped chat)

Agent Mode is secure, case-scoped chat: ask about this matter, scan and billing context, saved prompts, bulk task history, and more. You can also step through chat-native bulk scan with explicit confirmations when you want to run work from the conversation.

See Agent Mode.

Case Context

Case Context stores short, structured background your team agrees on (parties, issues, instructions). Grounding the assistant this way keeps prompts and answers consistent with how your team defines the matter. A case can hold several contexts — one per workstream — created from templates that match your world (litigation, regulatory investigation, internal investigation, law enforcement, or start blank), each with its own terminology and editable sections, and one marked as the case default.

See Case Context overview.

Field Migration (move values between fields)

Field Migration copies or moves values between Nuix Discover fields across a selection—with a dry-run Check selection pass and a real one-document test with undo before the full run. It involves no AI and costs no tokens.

See Field Migration.

Bring Your Own Model (a model your organization runs)

Bring Your Own Model points a case at an OpenAI-compatible endpoint your organization runs — including Azure OpenAI — instead of a Claira-hosted model. Set it up in Settings with the endpoint URL, model ID and API key; Claira tests the connection with named checks and, once saved, sends the case's single reviews, live bulk scans and Multi-Code passes to your endpoint. Per-document billing is unchanged.

See Bring Your Own Model.

Local Mode (a model on your own computer)

Local Mode goes one step further: Claira runs every review on a model on your own computer — Ollama, LM Studio, vLLM or llama.cpp — with your browser talking to it directly, so document content never leaves your machine. Single reviews, live bulk scans, Multi-Code, Agent Mode, Insights, Case Context, the Prompt Generator and Smart Revise all run on your local model; only background scans and image, audio, video and PDF scans are switched off while it is on.

See Local Mode.

Admin Portal (your whole organization)

The Admin Portal is where organization admins see every case Claira runs in, manage users and roles, read the partner invoice, route emails by role and connect Discover portals. It opens from the Claira menu in Discover (Manage organization) at app.claira.to; there is no separate sign-in.

See Admin Portal overview.

Where to go next

• What is Claira? — first-scan walkthrough
• First-time orientation
• Agent Mode limitations — what the assistant does not replace

---

Need help? Contact us at support@claira.to.