Cecilia

DISCO launched Cecilia in February 2026, calling it the "first scaled agentic AI in legal tech." Bold claim. But Cecilia does something the other platforms don't: it's included at no extra cost for existing DISCO customers. In a market where Harvey charges enterprise rates and CoCounsel bundles with Westlaw subscriptions, DISCO just handed its entire user base an AI agent for free.

Cecilia is purpose-built for large-scale document review — the kind of discovery work where firms review millions of documents across litigation and investigations. It doesn't try to be a general-purpose legal AI. It does one thing and goes deep.

What DISCO Cecilia actually does for document review

Cecilia operates as an agentic layer on top of DISCO's e-discovery platform. Instead of keyword searches and manual document-by-document review, Cecilia reasons through document sets the way a senior associate would — identifying relevant documents, categorizing them by issue, flagging privileged material, and surfacing the documents that actually matter.

The "agentic" part is the key distinction from DISCO's prior AI features. Previous AI-assisted review tools used predictive coding — essentially pattern matching. Cecilia goes further: it understands context, makes judgment calls about relevance, adapts its approach based on what it finds, and chains multiple analysis steps together autonomously.

For a production review of 2 million documents, the difference is massive. Traditional review requires a team of contract attorneys spending weeks (or months) clicking through documents. Predictive coding cut that by maybe 50-60%. Cecilia aims to reduce human review to the documents that genuinely require attorney judgment — potentially cutting the review population by 80-90%.

The practical impact: faster time to production, lower review costs, and more consistent quality than armies of contract reviewers working under deadline pressure.

Why 'first scaled agentic AI' is a meaningful claim

DISCO's "first scaled agentic AI" claim is worth unpacking. Harvey processes 700,000 tasks daily, so why does DISCO get to claim "first"?

The answer is scale of individual tasks. Harvey's 700K daily tasks are typically discrete — review this contract, draft this memo, analyze this filing. Each task involves one or a handful of documents. Cecilia tackles tasks involving millions of documents in a single workflow. That's a different kind of scale.

E-discovery is uniquely suited to agentic AI because the tasks are well-defined (relevance, privilege, issue coding), the data volumes are enormous, and the cost of human review is astronomical. A single large litigation can generate $5-20 million in document review costs. Cutting that by 50-80% with AI agents isn't incremental improvement — it's a structural change in how litigation economics work.

The ABA flagged that most "agentic" claims are marketing hype. DISCO's claim holds up better than most because document review is exactly the kind of high-volume, multi-step, judgment-intensive task where agentic AI delivers measurable value.

No extra cost: DISCO's pricing strategy

Including Cecilia at no additional cost for existing DISCO customers is a strategic move, not charity. Here's what DISCO is doing:

First, retention. E-discovery platforms compete fiercely for market share. Giving customers AI agents at no extra cost makes switching to a competitor painful — you'd lose Cecilia's capabilities and have to pay for an alternative.

Second, usage data. Every document Cecilia processes improves the system. DISCO gets millions of attorney-validated relevance decisions to train better models. The customers provide the training data; DISCO provides the AI.

Third, upsell surface. Cecilia drives more usage of DISCO's platform. More matters on DISCO means more hosting fees, more processing fees, and more per-GB revenue. The AI agent is the hook; the platform is the revenue.

For law firms, the economics are straightforward: if you're already on DISCO, Cecilia is pure upside. No budget approval needed, no new vendor evaluation, no procurement process. Just turn it on. That's a significantly lower adoption barrier than Harvey's enterprise sales cycle or CoCounsel's Westlaw bundling.

Cecilia vs Harvey vs CoCounsel for discovery work

For large-scale document review specifically, Cecilia has structural advantages over Harvey and CoCounsel:

Native e-discovery integration. Cecilia operates inside DISCO's review platform. Documents don't need to be exported, processed, or re-imported. The AI agent works on the same document set your review team sees, with the same coding panels, the same production workflows, and the same quality control tools.

Purpose-built for review volumes. Harvey's agents are powerful but designed for smaller document sets — contracts, agreements, filings. CoCounsel's strength is research, not bulk document classification. Cecilia is built from the ground up for million-document datasets.

Review workflow integration. Cecilia's output feeds directly into DISCO's review workflows — batch assignments, quality control sampling, privilege logs, production sets. There's no manual step between "AI identified relevant documents" and "team reviews AI's selections."

Where Cecilia falls short: it doesn't do legal research, drafting, or general-purpose analysis. It's a discovery tool. If you need contract review agents (Harvey) or research agents (CoCounsel), Cecilia doesn't replace those. Different tool, different job.

Implementation and governance considerations

Deploying Cecilia raises the same governance questions as any agentic AI in legal, plus some discovery-specific ones:

Defensibility. Courts have accepted predictive coding since *Da Silva Moore v. Publicis Groupe* (2012) and *Rio Tinto v. Vale* (2015). Agentic AI review is a step beyond predictive coding. Be prepared to explain and defend your methodology in meet-and-confer negotiations and to courts. Document your validation process.

Privilege protection. AI agents reviewing millions of documents will encounter privileged material. Cecilia's privilege detection capabilities need to be validated against your firm's privilege criteria — not just generic attorney-client tests. WilmerHale's March 2026 analysis warned that agentic AI creates "new hidden routes to privilege waiver". In discovery, that risk is amplified.

Quality control. Don't eliminate human review — redirect it. Use Cecilia to triage the document population, then have attorneys review a statistically significant sample of the AI's relevance decisions. Document your recall and precision rates. These numbers matter when opposing counsel challenges your production.

The firms getting this right are treating Cecilia as a force multiplier for their review teams, not a replacement. The AI handles volume; the lawyers handle judgment.

The Bottom Line: DISCO Cecilia is the strongest agentic AI play for large-scale document review — and including it at no extra cost makes it the easiest entry point for firms already on the DISCO platform.