CS Disco is an Austin legal-software vendor whose second-quarter print tests whether Cecilia, the in-house generative assistant, and AutoReview, the automated first-pass document review tool, can lift a usage-based e-discovery franchise off a mid-single-digit growth floor. The investment debate is not whether litigators want artificial intelligence. It is whether a still-unprofitable platform that bills mostly on matter volume can convert that demand into durable software growth without giving away ingest fees, the one-time charges for loading raw data, faster than platform subscriptions replace them. Management raised the full-year revenue range after a quarter in which total sales grew 13 percent, yet adjusted earnings before interest, taxes, depreciation, and amortization stayed negative. The market prices the equity as a cash-backed option on that conversion, not as a proven compounder.
The most important recent development is the January commercial reset branded DISCO Platform, a simplified per-gigabyte rate that bundles e-discovery, Case Builder, and Cecilia on every matter. Chief Executive Officer Eric Friedrichsen said the company reached its year-end run-rate goal for that model by June, half a year early, because law firms wanted straightforward pricing and larger matters followed the simpler quote. The mechanism matters. Simpler pricing lowers the sales friction that used to keep giant matters on Relativity or Everlaw, and bundling Cecilia on every matter turns the assistant from an add-on into default workflow. The shareholder consequence is mixed in the near term. Faster Platform adoption can cut ingest fees even as it lengthens matter duration and raises the odds of AutoReview attach. Finance chief Aaron Barfoot already flagged that if Platform keeps outrunning the plan, near-term revenue can sag from the lost ingest even while the installed base looks healthier.
The tension sits in the same print. Software still supplies most of the top line, and large accounts that spend more than $100,000 over a trailing year now contribute the bulk of sales, but usage-based contracts still generate the overwhelming majority of revenue. That mix means a quiet litigation calendar, or a single large matter that settles early, can erase a quarter of apparent AI momentum. Adjusted EBITDA deteriorated slightly versus the year-ago quarter even as cash burn from operations improved, which is the signature of a company still buying growth with research spend on Unified Litigation Solution, Advanced Research, and AutoReview. The strongest counterargument is that the AI revenue tripling is real product-market fit that the current multiple simply refuses to underwrite. The weaker read is that tripling off a small base, plus services strength tied to managed review on the largest matters, is still a cyclical e-discovery story wearing an agentic costume.
The next test is the fourth-quarter claim of adjusted EBITDA profitability, paired with whether AutoReview's record quarter repeats as Advanced Research, the multi-step reasoning layer, rolls out to the full customer base. If the Platform mix keeps growing without a visible ingest-fee hole in the third-quarter guide, the conversion thesis gains a second data point. If large-customer growth slows while research spend stays elevated, the cash pile becomes the entire equity story rather than a bridge. Those two variables, Platform mix versus ingest drag and the path through fourth-quarter breakeven, decide whether the present multiple is patient or merely indifferent.