Features
Everything Pulse does to help you catch problems early — and, just as important, what it doesn't do. No fluff, no invented numbers.
Beat your customers to it
Scheduled HTTPS checks from outside your infrastructure, turned into incidents your team can act on.
Checks run on a schedule — every five minutes on the pilot plan — against public HTTPS URLs you verify. Availability is calculated from completed checks, with coverage gaps shown separately.
An incident opens after consecutive failures (three by default) and resolves after consecutive successes (two by default). Both are configurable per monitor.
Acknowledge, assign, comment on, resolve and reopen incidents. Every transition is recorded with who made it and when.
Schedule maintenance so planned downtime doesn't raise alerts or count against availability.
No invented numbers
What Pulse can show depends on your provider plan and the access you grant. When data isn't available, we say so.
Connect through the Vercel integration to see projects and deployment status, and line deployments up against incidents. Log and analytics access varies by Vercel plan and is enabled only where it's verified.
Authorize with Supabase OAuth to read project status and service health (database, auth, storage and more). Usage and log data depend on your Supabase plan and granted scopes. We don't ask for your database password or service-role key.
Each capability is shown as available, stale, disconnected, unauthorized or unsupported. None of those last four is ever treated as healthy.
If a provider doesn't expose a metric on your plan, you'll see “unavailable” and why — not an estimate dressed up as a measurement.
Evidence, not vibes
Metrics are calculated first; AI only summarizes the evidence it's given, and every actionable claim links back to it.
A summary of what happened across your projects, written for owners with the technical detail developers need.
Each finding cites the checks, incidents and provider data behind it, with the time period, confidence level and known limitations.
Confirmed observations are labeled separately from suspected causes. Pulse doesn't claim to know a root cause it can't prove.
The AI has no access to change your systems. If AI is unavailable, metrics and incidents keep working and you get a clear “report unavailable” state.
Affected: Customers paying at checkout
What happened
ConfirmedFor about 40 minutes on Tuesday afternoon, customers trying to pay waited 4 seconds or more for the checkout page. It recovered on its own at 2:50 PM.
Why it matters
Slow checkouts cost sales: some customers likely gave up. Nothing was lost or charged twice, but it will happen again under the same load unless it's fixed.
What we see
ConfirmedCheckout response times rose from ~300 ms to ~4 s, starting 2:10 PM, while a background import job was running.
Likely cause
SuspectedThe import job opens a new database connection for every item it processes, starving checkout of connections.
What we see
ConfirmedDatabase connections hit 117 / 117 at 2:12 PM, with many sessions sitting idle in transaction.
Likely cause
SuspectedTransactions left open by the job hold connections long after their work is done.
Evidence
Database fix
Effort basis: One configuration change, applied without downtime.
Impact: Frees stuck connections so checkout stays responsive while jobs run.
Pros
Cons
Who: Your database admin
Engineering fix
Effort basis: One job to refactor and test against a production-sized import.
Impact: Removes the root pressure on the database, so imports and checkout can run together.
Pros
Cons
Who: Your developer
Less 2 a.m. fire drills
One alert when an incident opens, one when it recovers. The right people hear about it — once.
Incident and recovery emails to the members and client contacts you choose, plus your weekly reports.
Text alerts for people who explicitly opt in and verify their number, with quiet hours and severity thresholds they control.
Repeated notifications for an unchanged incident are suppressed, and you're notified once on recovery.
Every client, one workspace
Direct businesses and agencies both get owner and developer views. Agencies can manage many clients from one workspace.
Invite teammates with the role they need. Removing someone takes effect right away — access is checked against live membership, not stale sessions.
Organize client organizations inside one workspace, with portfolio summaries and per-client filtering.
Client users see only the organization or projects they've been granted — never your other clients.
Export your authorized data, and request workspace deletion with a documented recovery period.
Coming to pilot workspaces
Two features rolling out to pilot customers now. They may not be in your workspace yet — we'll tell you when they are.
Stop answering “is it down?” one email at a time. Drop a copy-paste snippet on your site and customers see live health from your Pulse checks.
Acme Status
Powered by PulseMinor slowdown: API · checked 3 min ago · checks every 5 minutes
Last 30 days
99.94% uptime
“The site feels slow” becomes a number you can act on — before it becomes a complaint.
Customer portal
Fast
Admin portal
A little slow
Checkout
Fast
99.4%Last 30 days · about 259 min down
What's pulling it down: one checkout outage on the 12th (3 h 5 min) accounts for most of it.
Connections
37 / 117
Slowest query
SELECT … FROM orders WHERE customer_id = $1
avg 412 ms · worst 1,870 ms
Being upfront
We'd rather you know now than find out later. These aren't available today:
Start your pilot and find out about problems within minutes — from Pulse, not from your customers. Business email required.
Backup when you need it
Pulse is built by InnoStak, a software engineering and automation partner. When a fix is beyond your team, hand it over with the evidence attached. Optional, and quoted separately from your Pulse plan.
Talk to InnoStak (opens innostak.com in a new tab)