Metrics, logs, and traces — one agent, one lean pipeline. You pay nothing for telemetry until a dashboard, an alert, or a custom rule of yours actually needs it, and only for as long as it needs it.
Free tier with generous limits. No credit card, no time limit.
Why LeanSignal
Legacy observability vendors all give the same advice: store everything now, worry about the bill later. Follow it and you end up in The Pain Cycle.
LeanSignal applies lean principles to observability. Demands — not defaults tuned for the vendor's profit — decide what gets collected, and the loop starts running in your favor: The Value Cycle.
So what do the two cycles cost over time? Applying lean principles flips the cost dynamics completely — away from favoring your vendor, toward favoring you.
The result: an order-of-magnitude cost advantage when you start, and an expected 2× advantage over the best legacy observability setups.*
And one more thing: Lean saves trees 🌳
Agentic observability
Meet LeanBuddy — Your AI Teammate
Ask in plain language — "why is p99 up since 14:00?" — and LeanBuddy investigates for real: it queries your live metrics, logs, and traces, explains what changed, and hands you deep links straight to the evidence. It even knows what you are NOT collecting, and says so instead of guessing.
Connect Claude Code, Cursor, or VS Code to your tenant over MCP with one personal token. Ask about production while you write the fix — the metrics, logs, and traces behind an incident are one question away from the code that caused it. These questions run on your regular AI subscription, not on your LeanBuddy budget.
Opt a token into write access and the same agent that instruments your code declares the demand, builds the dashboard, wires the alert, and test-fires it against live data — before the pull request merges. The change and its observability ship as one motion, not a follow-up ticket.
Because the pipeline stores only demanded telemetry, the AI reasons over a small, relevant dataset instead of digging through a haystack. That means sharper answers, fewer tokens, and lower cost per question.
LeanBuddy shows the queries it ran and what each answer cost, and it works within a spending budget you set. You can always see where an answer came from and what you paid for it.
Tokens are read-only by default; writes are an explicit opt-in, deletes a separate one. Every AI-made change runs under your role, passes the same validation as the app, and lands in the audit log marked as agent work. One click revokes it all.
Features
You declare the signals you need. Demand — not supply — decides what gets collected, so noise is eliminated from day one instead of pruned after the bill.
Every signal is tied to the need it serves. You can finally see which spend buys which value — and cut the noise with confidence.
Telemetry is indexed and filtered at the edge, before it ever reaches storage — no more choosing between storing everything and whitelisting blindly.
Escape the store-everything, bill-shock, reactive-pruning loop. Start at zero and grow cost and value together as your needs change.
One agent, three signals
The same demand model across all three signals, each defined and explored in the query language you already use.
PromQL Every metric your agents collect, explored with the Prometheus query language you already know. Graph it before it is ever a demand, then pin what matters to a dashboard or an alert.
LogQL Search structured logs with LogQL. Agents keep everything locally and forward only the streams you demand — grep the edge live, store just what earns its place.
TraceQL Follow a request across services with TraceQL. Demand is service-scoped: name the services you care about, and their traces are stored in full, ready to inspect span by span.
Capabilities
Alert on any PromQL condition and route firing and recovery to email or Slack. A rule lives inside the demand it belongs to, so alerting on a metric turns its collection on automatically.
Scheduled HTTP checks watch your endpoints from the outside. After a set number of failures a check flips to Failed and notifies the same channels as your alerts — recovery included.
Go beyond dashboards: author your own metric, log, and trace rules in each signal’s query language to declare exactly what to forward — no dashboard required.
Build Perses dashboards that double as demand: the panels you save decide what gets collected. Every change is versioned, exportable, and portable between tenants.
Add an unlimited number of users to your tenant, a standalone instance of the observability stack that is yours alone. Admin, editor, and viewer roles keep everyone working with the right permissions.
Create multiple tenants and isolate telemetry by environment, stage, or any criteria you see fit. Each tenant is its own standalone stack, so the workloads never mix.
Manage your tenant from the terminal with the open-source leanctl. Demands, dashboards, alerts, and more work as commands, ready for scripting and CI.
Sign in with email and password, Google, GitHub, or your own identity provider. Custom SSO integration is available even on the free plan.
Curated bundles for PostgreSQL, MySQL, AWS CloudWatch, host metrics, and the major language runtimes. One import brings the dashboards and their demands together, so a new integration works in minutes.
The console ships with polished dark and light themes. Pick what suits your eyes or your room — your choice sticks across sessions.
OpenTelemetry
Prometheus
Perses
VictoriaMetrics
Loki
Tempo
Security, isolation & privacy
Your tenant is its own isolated stack — its own storage, its own compute. No noisy neighbours, so performance stays predictable, and your data never shares a table with anyone else’s. Isolation is the default, not an upsell.
Every critical change — demands, agents, checks, retention, users — is captured in a per-tenant audit log. Who changed what, and when, is always answerable, so changes stay accountable and compliant.
The LeanSignal agent is a fully open-source distribution of the OpenTelemetry Collector. It includes the components that establish connectivity with our backend and enable Demand-Driven Observability, and you can read, audit, and build every line of it. No proprietary black box and no hidden data paths.
Your telemetry is processed and stored in line with the GDPR and strict data-protection rules — you always know where your data lives and who can touch it.
Free service
Generous limits, no credit card, no time limit.
Your Usage page shows exactly where you stand against every limit. Reach one and ingestion simply pauses — no overage, no surprise bill. Monthly budgets reset on the 1st.
Outgrowing the free tier? Get in touch and we’ll set you up with a plan that fits.
Sign up, install our OpenTelemetry-based LeanSignal agent, and start collecting the telemetry that actually matters.
Free tier with generous limits. No credit card, no time limit.
* Based on our own observability setup running on top of LeanSignal, compared with the estimated cost of the same setup with legacy vendors.