Architecture & stack fit

Reads what you have. Routes what you decide.

No migration. No rip and replace. Add the tools you already use. Bicycle reads them, explains drift, and routes action where work happens.

Architecture

The whole system, from your stack to the answer.

Bicycle reads the stack you already run, models it once, then detects, explains, and acts on what changes, inside your guardrails.

Surfaceswhat business teams touchAI layer
Executes a use case. Example use cases: Search conversion · Bookability · Approval monitoring.
Alerts & ActionsProactive notifications, incident routingData StoriesMBRs, on-demand narratives and reportsChatContext-aware conversational UIDashboardsPre-packaged and custom viewsActionsTrigger workflows, automate
Self-serve & governancethe safety railGovernance layer
accessapproved definitionsevidenceauditrollback
Cause EngineAI + MLML / analytical layer
Technical CausesBusiness Causes
e.g. a payment routing misconfiguration drove interchange downgrades across 4 merchant categories, an estimated $3.2M in excess processing fees.
Pattern Enginedetects what mattersML / analytical layer
AnomaliesTrendsForecastsImpact Ranking
e.g. authorization rate dropped 7.4% year on year, led by card-present (−11.2%) and cross-border (−8.6%) declines after new issuer routing rules.
Business modelthe semantic layerBusiness model layer
EventsTransactions, interactions, automationsDimensionsMerchants, products, regions, channelsKPIsRevenue, auth rate, NPS, AOV, fraud rateCauses / DriversPricing, latency, promotions, outagesActionsAlerts, tickets, workflows, SOPs, outreachPoliciesThresholds, SLAs, guardrails, resolution stepsPlaybooksRunbooks, workflows, case resolutionCohortsCustomer segments, user and behavioral groupsJourneysOnboarding, signups, support, fulfillment
OnboardingAI layerAI layer
Source onboardingDiscover sources and connectMappingSchema and semantic mappingData qualityValidate, monitor and scoreCatalogingMetadata, lineage and ownership

Curated, trusted, modeled data lands in the Business model layer.

ConnectorsBicycle-owned, in and outData layer
SignalCauseAction
Your stacksystem of recordData layer
Reads from core data
Warehouses, lakes, event streams, CDPs. The source of signals and KPI math.
SnowflakeDatabricksBigQueryRedshiftS3Kafka
Reads operational systems
Payment, order, support, catalog, pricing, CRM. The systems your business runs on.
ShopifyStripeSalesforceOracleSAPServiceNow
Integrates as evidence
Observability, logs, errors, session replay, deploys, docs. Evidence that explains what happened.
DatadogSplunkSentryNew RelicCloudWatchPagerDuty
Routes action
Messaging, ticketing, ITSM, lifecycle, automation. Where the next step gets routed.
SlackEmailJiraServiceNowWebhooksZapier
AIML / analyticalBusiness modelDataGovernanceData flowOnboarding flowContext feeds
Stack fit

Add your tools. See where Bicycle plugs in.

Bicycle reads from the systems you already run and routes action back into them. Add yours below to see what it connects to, what it explains, and where it acts. No migration, no rip and replace.

Build your stack

Your stack:
Start with a sample:
Or browse by layer
Add at least one evidence tool and one action tool to play with your stack.

00:00:00

Click a step, or press play.

Governed by your team

Your stack stays yours. So does the truth that runs on it.

Bicycle reads what you already have and recommends the next step where work happens. Your data team controls the definitions, what becomes trusted, and which action the business can take. The business self-serves on a governed model, not around it.

app.bicycle.ai published answergoverned
The answer
Conversion recovered ▲ +0.7pp
read from your warehouse & payments · this week
Definition & lineage
conversion = orders / sessions
4 sources on your stack · validated ▸ view source
Act
Restore route · your ops tool  approval
scoped by the data team · reversible · logged
01

Definitions you govern

One governed definition feeds every alert, story, dashboard, and chat answer. The number is the same everywhere.

02

Evidence, one click away

Answer to source query to evidence. Lineage is always attached, on the sources you already trust.

03

Act inside guardrails

Every action is scoped, previewed, approved, reversible, and logged, on the tools where work happens.

Internal deployment

A team stood up Bicycle on the warehouse, streams, and tools they already ran. No rebuild, no BI rip and replace. The first reviewed agent was live in days, not months, and the dashboards they trusted kept doing their job.

Vibe Analytics

Your first agent, on the stack you built above.

Vibe Analytics is the fastest path into Bicycle. Bring a prompt and your data, and it generates your first working agent on the stack you picked. Your team reviews what it built, the model, KPIs, and causes, before it goes governed. No long implementation project.

No credit card required

app.bicycle.ai vibeprompt → first agent
Bicycle generates your first working agent on your stack, for review
From a prompt to a reviewed first agent, on the stack you built above.
Get started

See Bicycle on your stack.

No migration. No rip and replace. Start with one focused use case and get a useful first pass on the stack you already run.

No credit card required