Is Trajectory a direct competitor to Moda?

Only loosely. Both companies use the phrase continual learning, but Trajectory is a post-training data plane and Moda is a production-conversation analytics product. The buyers differ — ML team vs agent product team — and the outputs differ.

Yes, and many teams will. Moda surfaces what to change in production (prompt, tool, workflow, context, memory, model); Trajectory turns the underlying traces into training data for the model side of that. They sit at adjacent layers.

Does Trajectory detect hallucinations or tool misuse?

There is no public claim of behavioral failure detection on Trajectory's site or docs. The product is positioned as a data primitive for post-training, not as agent observability.

Does Trajectory cluster user intents?

No. Their primitive is the Trajectory itself — a normalized telemetry record — not a user-intent taxonomy.

Who is Trajectory built for?

Early customers include Clay, Decagon, and Harvey — teams shipping their own agentic models that need a clean data plane for continual post-training. Investors include Conviction, Bessemer, Radical, BoxGroup, and angels Jeff Dean, Fei-Fei Li, plus founders of Notion, Dropbox, Braintrust, and Hugging Face.

Moda vs Trajectory

Trajectory (Conviction-led $15M seed, May 2026) is a continual learning data platform: an SDK that turns traces and telemetry into a standardized Trajectory primitive, then makes that data available for post-training and steering agentic models. Early customers include Clay, Decagon, and Harvey. Trajectory shares the continual learning framing but sits at a different layer — it is the data plane for teams post-training their own models. Moda is self-improvement on the harness layer: production-conversation analytics that surfaces what users want, where the agent fails, and which harness component (prompt, tool, workflow, context, memory, eval, or model) needs to change next. Moda's learnings live outside the model weights; Trajectory packages traces for teams that will update the weights. The two layers are complementary.

When the question is what users actually want, where the agent is breaking down in production, and which layer of the stack to change next.

Capability	Moda	Trajectory
Layer of the stack	Production analytics — intent map, emergent intent, behavioral cohorts, frustration root cause routed to a layer of the stack.	Continual learning data plane — Trajectory SDK converts traces / telemetry into a standardized primitive for post-training.
Primary unit	Conversation segment + intent cluster.	Trajectory (the named primitive) + TelemetryEvent.
Intent clustering	Automatic live intent map with emergent intent detection.	Not in scope.
Behavioral failure detection	Tool call failures, schema drift, workflow loops, agent path issues, model behavior shifts surfaced as part of the learning loop.	Not claimed; failure-mode language is absent from the docs.

Book a call Contact