Dust vs Locus Robotics

Dust's feature set is entirely software. You define an agent with instructions, attach knowledge and connected tools, pick a model per agent from 20+ options (OpenAI, Anthropic, Google, Mistral, DeepSeek), and run it on a schedule, on an event trigger, or from chat. Multi-agent orchestration, self-improving skills that encode team expertise (shipped July 2026), forwarding an email thread to an agent (May 2026), native and remote MCP servers, and Frames dashboards round out the platform. Its governance layer — Spaces for per-team data segmentation, a dual-layer permission model separating agent data access from user access, SSO via Okta/Entra ID/Jumpcloud, SCIM provisioning, and EU data residency — exists because buyers are rolling agents out across departments, not to individuals. Locus's feature set is physical plus orchestration. Its AMRs handle picking, putaway and point-to-point transport without fixed paths or permanent infrastructure; LocusONE rebalances work across associates and robots in real time as volume and labor change; Locus Array (launched Jan 2025) extends this to fully autonomous Robots-to-Goods picking, putaway, induction, drop-off, slotting and replenishment, with Locus claiming up to 90% labor reduction. Locus integrates with WMS platforms (SAP EWM, Manhattan, Blue Yonder, Oracle, NetSuite, Infor, Körber) — a different integration universe entirely. There is no overlap in capability: one automates tasks in software tools, the other automates movement of physical goods.

Dust uses a freemium model with credit-based usage that varies by which model and tools an agent invokes — credits don't roll over, so buyers who need flat-rate predictability should account for that. The friction is upfront effort, not the bill: the agent list, skills and permissions need real configuration before the platform pays off, which is why Dust's own 'not for' list excludes individuals wanting a single personal chatbot and small non-technical teams without an ops owner. Locus publishes no self-serve pricing at all; it's contact-sales with a Robots-as-a-Service subscription, meaning you commit to ongoing fees for robot fleet access rather than a one-time hardware purchase. Whether that pays back depends on order volume and labor cost at your site — Locus itself flags low-daily-volume warehouses as a bad fit because RaaS fees won't amortize. These two pricing structures aren't comparable in any meaningful sense: Dust's cost scales with AI model and tool usage across a knowledge-work team; Locus's cost scales with robots deployed across a square footage of warehouse floor. A budget holder evaluating one is not comparing line items against the other.