Signal vs Pecan AI
Pecan is a predictive AI agent — it builds models for a fixed set of business questions like churn, LTV, and lead scoring. Signal builds the model AND the interface your team uses, with the agent that keeps both running, for whatever question your data has in it.
01
Predictions, or Machines
Pecan's unit of value is a prediction — written back to your CRM or warehouse. Signal's unit of value is a Machine — the model, the interface, and the agent that runs it, compiled together.
Pecan
Signal
What you ship
Predictions delivered to CRM/warehouse
Machine = model + interface + agent
Where the model lives
Pecan's platform
Portable, your tenant
Team interface
A column in your CRM
Custom interface generated for the use case
Ownership
Vendor-hosted, paid by batch
Yours, compiled, take it with you
02
Open use cases, or fixed catalog
Pecan ships 8 named solutions. If yours isn't on the list, you wait. Signal's agent plans whatever build the question demands.
Pecan
Signal
Use case scope
8 fixed (churn, LTV, lead scoring, demand, …)
Open — agent plans the build
New use case
Wait for Pecan to ship one
Describe it, get one
Custom logic
Limited to platform templates
Agent composes from 9 model skills
Cross-skill builds
Each use case is its own product
Compose freely (e.g. churn + auto-save-play)
03
Pricing and minimums
Pecan is sales-led, annual-commit, gated by prediction batches. Signal is self-serve, usage-based, with no annual.
Pecan
Signal
Entry tier
$760/mo Starter, annual commit
$5–20 ad hoc
Pricing axis
Prediction batches
Actual compute + tokens + storage
Self-serve
No — sales-led at every tier
Yes
Annual commit
Required
None