About Discovarc
Discovarc is a Chicago-based seed-stage company building the workflow layer that makes predictive document review a standard practice at litigation support firms — not a specialty capability reserved for the largest cases. We are in early revenue, working with litigation support firms and e-discovery managed service providers that handle 5–50 active review matters per month.
Seed-stage, early revenue, working directly with litigation support firms
Discovarc is at the seed stage, with early-revenue traction from litigation support firms and e-discovery managed service providers across the US. Our current customers run between 5 and 50 active review matters per month and are using Discovarc to standardize their TAR deployment across their project manager staff — reducing setup time per matter and generating consistent, defensible QC documentation.
We are focused on the US litigation support market in this phase, working with firms where predictive review is a known capability but where no repeatable workflow has been established. The product is in active use and we are expanding the customer base with a small number of new firm partnerships per quarter. We are not in a broad sales motion at this stage — each onboarding is hands-on to validate the protocol against the firm’s existing matter workflow.
Our current integration coverage spans Relativity and RelativityOne, DISCO, Everlaw, and Reveal, covering the review platforms used by the majority of US mid-market litigation support firms. Nuix and Logikcull integrations are on the near-term roadmap, driven by customer demand from the firms we are currently onboarding.
Make predictive document review the default protocol at every litigation support firm, not the exception.
Most litigation support firms already have access to active learning technology through the review platforms they subscribe to. The gap is not access — it is the repeatable, documented workflow for seeding the model, validating recall, and producing a TAR protocol that holds up under opposing counsel scrutiny. That is the problem Discovarc was built to close.
Every litigation support firm that deploys TAR consistently reduces first-pass review cost by 35 to 45 percent per matter. The technology that makes that possible exists in Relativity, DISCO, Everlaw, and Reveal today. Discovarc provides the structured workflow layer that turns that technology into a repeatable, defensible standard practice — so every project manager can deploy it on every eligible matter, not just the largest ones.
The story behind Discovarc
Naomi Ashford spent eight years managing document review at a Chicago litigation support firm before founding Discovarc. The moment that shaped Discovarc came in 2023, when her team completed a 400,000-document first-pass review for a commercial contract dispute. Twelve contract attorney reviewers worked for six weeks at a total cost of $290,000.
After production, the team ran a retrospective analysis and discovered that 80% of the responsive documents were concentrated in the first 15% of the collection—when sorted by the relevance model they built in the last week of review as a post-mortem exercise. The model they needed existed in Relativity’s Analytics module the entire time. It was not used because the firm had no repeatable protocol for seeding it, validating recall, and generating a defensible TAR protocol for the producing party’s records.
Naomi co-founded Discovarc with Patrick Oduya, who had spent five years building document processing and ingestion pipelines for a leading document forensics software company. Their first prototype was a structured active learning workflow layer: a step-by-step protocol guide for seeding a 500-document training set, running iteration batches, and validating recall at each stage. Piloted at two Chicago litigation support firms, the protocol reduced first-pass review hours by 35 to 40 percent on the first matter where it was applied.
Discovarc today is a platform-agnostic active learning layer that sits above Relativity, DISCO, Everlaw, and Reveal. It handles the seeding, iteration management, QC validation, and protocol documentation that firms need to run defensible TAR consistently—without requiring each project manager to build the workflow from scratch on every new matter.
The principles behind Discovarc
- Recall defensibility before speed — Every matter Discovarc runs produces a documented QC sample at 95% confidence and ±2% margin. Defensibility is not optional; it is the reason firms can trust predictive review in production.
- Platform-agnostic by design — Discovarc works with Relativity, DISCO, Everlaw, and Reveal. Firms should not be forced to change their review platform to run better TAR protocols. We integrate with the tools you already use.
- Project manager workflow first — The people who run document review at litigation support firms are project managers, not data scientists. Every Discovarc workflow is designed for the operator who already has a full matter workload, not for an AI specialist.
- Every disposition explained — When Discovarc assigns a predictive coding disposition to a document, the reasoning is documented. No black boxes. Every disposition that goes into a production record is traceable to the model state and the QC sample that validated it.