Legal aid and client intake have long depended on in-person attorney interviews to complete standardized government and organizational forms—a model that is slow, expensive, and inaccessible to clients without reliable access to counsel. For an AmLaw 100 firm like Davis Wright Tremaine, serving clients in sensitive matters such as domestic violence and immigration, the stakes of an inadequate intake process are high. Existing AI tools introduced new risks rather than solving the problem: general-purpose models are prone to hallucination, lack the emotional sensitivity required in trauma-adjacent contexts, and failed to meet the increasingly complex form-completion requirements of major clients.
DWT's internal R&D unit, De Novo, built Crescendo—a proprietary conversational AI engine trained on DWT's own language model, dwt.ai—to invert the standard AI paradigm. Rather than prompting an AI to generate answers, Crescendo reverse-engineers official government and organizational forms into structured, empathic chat-based interviews. The system uses conversational AI to ask questions, gather unstructured client responses, and convert them into fully populated, properly templated forms ready for attorney review. Designed with emotional intelligence to mirror the nuance of in-person interviews, Crescendo was developed rapidly by De Novo in direct response to a major client's request, deploying on DWT's own infrastructure without reliance on third-party AI vendors.
Crescendo eliminated the requirement for a lawyer to be present during initial form completion, making standardized legal forms accessible to anyone with a voice- and internet-enabled device at any time. The qualitative outcomes are significant:
Have a similar implementation?
Share your customer's AI results and link it to your vendor profile.
Submit a case study →