Decision engine · API-first
The decision engine for stateful entities
Register your entities, connect your signals, and let Stateline keep the state of every account and decide which action to fire — with memory, suppression, and a full audit trail.
In active development · early design partners welcome
01
Signals
anything that happens
02
State
remembered per entity
03
Decision
pure, simulatable rules
04
Action
suppressed, retried, audited
The problem
Your automations don't remember
IF-THEN tools evaluate every event in isolation. No memory, no context, no explanation.
They alert your sales team about the same “hot account” three times in a week. They keep nudging customers who already churned. And when someone asks why an action fired, nobody can answer. Without memory there is no context — and without context, automation is just noise.
Stateline models every account, contact, or deal as a real state machine. Every decision is made with the entity's full history in mind — including the things that didn't happen.
How it works
From raw signals to the right action
Four stages, one durable event log. Every step is recorded, replayable, and explainable.
-
01
Send signals
POST any business event to a single API: product activity, CRM changes, billing events, parsed emails. Each signal is validated, deduplicated, and stored durably — in order, per entity.
-
02
State machines remember
Every entity runs versioned state machines for the dimensions you care about: lifecycle stage, risk, engagement. Timers detect absence — “no activity in 60 days” becomes a signal, just like any event.
-
03
Pure rules decide
Rules see the signal plus a snapshot of the entity's full state. They are pure functions, which means you can simulate any rule against your real history before switching it on.
-
04
Actions fire safely
The dispatcher applies frequency caps, cooldowns, and retries, then calls your webhook, CRM, Slack — or an AI agent. Every action carries a trace that explains exactly why it happened.
Key features
Infrastructure, not another dashboard
The primitives your team would otherwise spend months building — exposed as one coherent API.
-
Real state machines
Declarative, versioned machines per entity dimension. Memory is the data model, not a workaround built on lookup tables.
-
Absence detection
Timers emit synthetic signals when nothing happens: a trial about to expire, an account gone quiet, an invoice unpaid for 15 days.
-
Simulation before activation
Replay a candidate rule against your event history — “this would have fired 4,300 actions last month” — before it touches production.
-
Explainability built in
Every action answers four questions: which signal arrived, what state the entity was in, which rule version decided, and why.
-
Smart suppression
Frequency caps and cooldowns at dispatch — plus semantic suppression: when an entity's state changes, stale alerts simply stop matching.
-
Config as code
Entities, machines, rules, and destinations are versioned definitions. Review them in a PR, activate them explicitly, roll them back safely.
-
AI with guardrails
AI extracts signals from calls and emails, drafts rules from natural language, and can execute actions as an agent — while the decision path stays deterministic and auditable.
For your business
Turn scattered signals into revenue moments
Stateline's first vertical is go-to-market: the moments where reacting in time — once, and with context — is worth real money.
Catch buying intent in time
A trial account with heavy usage visits your pricing page. Stateline knows its stage, checks suppression, and alerts the right AE within seconds — exactly once.
See churn before it happens
Sixty days of silence is invisible to event-based tools. Stateline's timers move the account to at-risk and open a save play automatically.
Qualify with context, not scores
A PQL built from lifecycle state means something: trial, high usage, and a pricing visit together tell a story no isolated score can.
Trust every automation
When sales asks “why did I get this alert?”, there is a real answer: the signal, the state, the rule, and the version that decided.
Why Stateline
Not another scoring app
vs. workflow automation
Zapier-style tools process events and forget them. Stateline maintains state: memory, timers, and suppression are the core of the model — not duct tape around it.
vs. GTM scoring tools
Common Room or Pocus give you an opinionated app. Stateline is API-first infrastructure: your entities, your state machines, your rules — versioned and explainable.
vs. building it in-house
An event log, a state machine runtime, a rule engine, suppression, and an audit trail add up to months of plumbing. Get it as one API and spend the time on your playbooks.
FAQ
Frequently asked questions
What is a decision engine?
A decision engine ingests signals (business events), maintains the state of each entity — accounts, contacts, deals — and decides which actions to trigger based on explicit rules. Unlike simple automation, every decision accounts for the entity's full history, including events that never happened.
How is Stateline different from Zapier, Make, or n8n?
Workflow tools run stateless pipelines: an event comes in, actions go out, nothing is remembered. Stateline keeps persistent state per entity with real state machines, detects absence with timers, suppresses duplicate actions, and lets you simulate rules against history before activating them.
Is Stateline a CDP like Segment?
No. A CDP collects, unifies, and routes customer data. Stateline sits downstream: it expects resolved entity IDs and focuses on deciding what to do. They are complementary — you can pipe CDP events into Stateline as signals.
What is the difference between a signal, a state, and an attribute?
A signal is something that happened (“visited pricing”). A state is a condition the system derives and governs through a state machine (“stage = at_risk”). An attribute is a declared fact used as context (“industry = retail”). Stateline keeps all three consistent on one event log.
Can I test rules before they go live?
Yes. Because rules are pure functions over a replayable event log, Stateline can simulate any candidate rule against your real history and tell you exactly which actions it would have fired — before you activate it.
Where does AI fit in?
AI extracts signals from unstructured data (calls, emails, tickets), drafts rules from natural language, can rank candidate actions, and can act as an execution agent. The decision path itself stays deterministic: the answer to “why did this happen?” is never “the model decided.”
When can I use Stateline?
Stateline is in active development: the design is closed and implementation is underway. If you want to shape the product as an early design partner, talk to us — we are onboarding a small group.
Building your GTM engine?
We are looking for a small group of design partners — teams that want signal-driven, explainable automation on top of their accounts and pipelines. Tell us about your use case.
Talk to ushello@stateline.dev