Open source · v1.18 ships today

Replay the truth
behind every bug.

Frontend session replay, backend errors, logs and traces — stitched on a single timeline. One open-source platform replaces the LogRocket + Sentry + log-tool stack you’re quietly outgrowing.

Start free Watch a 90-second demo 12.4M sessions replayed last 24h
No credit card · 10K free sessions/mo SOC 2 Type II · GDPR · HIPAA-ready Self-host with one Docker command
Trusted by engineering teams who’ve outgrown the legacy stack 2,400+ orgs
01 — Session replay

Replay every session.
Then jump straight to the evidence.

The bug report says “checkout is broken.” You have 142 sessions to choose from.

Highlight records every user session — DOM mutations, console output, network calls, redux state — at 60fps with privacy-by-default masking. When you open one, the playhead is already on the error frame. You see what the user saw, three keystrokes before it broke.

Pixel-perfect DOM replay

rrweb-based capture, <3% CPU overhead, masks every input by default.

Auto-jump to the failure frame

Errors, rage clicks, dead clicks and network failures are flagged on the scrub bar.

Privacy you can prove to legal

Server-side PII redaction, domain allowlists, GDPR-compliant retention controls.

Search by any signal

Filter by browser, version, A/B variant, custom user trait, or any error message in console.

app.acme.io/checkout
01:17.842
acme.
PlansCustomersSettings
Confirm your order
Review the items in your cart and complete checkout with a single click.
Subtotal
$248.00
Tax
$22.32
Card number
Pay $270.32
ERR · cart is undefined · checkout.ts:88
02 — Unified observability

Three pillars,
one truth.

Sessions on the front end. Errors and traces on the back end. Logs in the middle. Highlight is the only platform that ships all four with a shared trace ID — so the customer’s click and your server’s 500 are the same line on the same timeline.

F.01

Session replay

Watch real users interact with your app. 60fps DOM capture, network & console included, with sub-3% performance overhead and privacy-by-default masking.

F.02

Error monitoring

Auto-grouped exceptions across web, mobile and backend. Source-mapped stack traces, GitHub PR links, and Slack alerts that already know the offending session.

TypeError: cart is undefined412
RangeError: maximum call stack88
FetchError: ECONNRESET /api/me61
UnhandledPromise: stripe.charges23
ChunkLoadError: 4f3.chunk.js9
F.03

Logs & traces

OpenTelemetry-native distributed tracing and structured logging on ClickHouse. Sub-second queries across billions of rows, with automatic correlation to the originating session.

/checkout
842ms
db.cart
412ms
stripe
280ms
db.user
86ms
cache.set
12ms
render
8ms
03 — Open by default

Source-available, self-hostable,
yours.

github.com/highlight/highlight · AGPL-3.0

Read every line. Audit every byte. Or fork it.

Highlight is open source — not “source-available marketing.” The full SDK, ingest pipeline, ClickHouse schema and frontend live on GitHub under AGPL-3.0. Run the same code we run in our cloud.

8,420 stars · 412 contributors · 1,847 PRs merged ↗ github.com/highlight/highlight
# Install in any Node, Next.js, Remix, or SvelteKit app
$ npm install @highlight-run/next
# Wrap your root layout — that’s the whole setup
$ highlight init && highlight start
Average setup time, last 1,000 installs · 3 min 12 s
F.04 — Self-host

One Docker command. Your infrastructure.

Deploy the entire stack on your own VPC. Customer data never leaves your network. Used by HIPAA, FedRAMP and GDPR-bound teams.

SDKbrowser · server
IngestOTel · gRPC
StoreClickHouse
UINext.js
docker compose up — 90 s to first session
F.05 — Integrations

Plays nice with the tools you keep.

OpenTelemetry-native means Highlight slots into the OTel collector you already run. Native integrations across the rest of your stack.

Next.js Remix SvelteKit Go Python Ruby Java Slack Linear PagerDuty GitHub Vercel
27 official SDKs · 64 community plugins
F.06 — AI assist

Ask why — get the session, the trace and the PR.

Highlight’s AI assist surfaces the offending commit, suggests a fix, and pre-fills the GitHub PR description. Available on every session, every error, every trace.

# prompt
why is checkout failing for paid users?

# answer · sourced from 412 sessions, trace 9a3f…b2c1, PR #2241
computeTotal() on checkout.ts:88 reads cart.items before resolveCart() resolves. Fix ↓
F.07 — Security

SOC 2 Type II. HIPAA-ready.

Annual third-party audits. PII redaction at the SDK boundary. EU/US data residency. SSO, SCIM and audit logs on every plan above Free.

SOC 2 II GDPR HIPAA-ready SSO/SAML SCIM
Last audit · Mar 2026 · download report
04 — Inside the product

From the support ticket to the offending line of code, in four clicks.

A real session. The user reported they couldn’t check out. Highlight already grouped the error, attached the session, captured the slow query and surfaced the failing line. You’re looking at the same panel an on-call engineer sees at 3 a.m.

Sessions Errors 3 Logs Traces Dashboards Alerts
acme · production last 24h K
Affected sessions 412
JM
jordan@notion.soChrome · 02:14 · 1m 47s
err
EW
elena@vercel.comSafari · 02:11 · 2m 04s
err
RK
ravi@linear.appChrome · 02:08 · 0m 52s
err
SO
sofia@stripe.comFirefox · 02:04 · 3m 11s
slow
MG
marcus@hightouch.ioChrome · 01:58 · 1m 22s
err
AK
aiko@cal.comChrome · 01:52 · 0m 38s
err
PL
priya@knock.appEdge · 01:47 · 4m 02s
slow
DB
diego@trigger.devChrome · 01:41 · 1m 09s
err
acme.
PlansCustomersSettings
Confirm your order
Review the items in your cart and complete checkout with a single click.
Subtotal
$248.00
Tax
$22.32
Total
$270.32
Card number
Pay $270.32
ERR · 01:17.842 · cart is undefined
❚❚
01:17 / 02:48
412 affected · 6.4% of /checkout traffic first seen 02:14 today regression in PR #2241
Go to PR Assign Silence
05 — Honest comparison

One tool replaces
the three you’re paying for.

If you’re running LogRocket for replay, Sentry for errors and Datadog for logs, you’re burning roughly $2,400/mo at 50K MAUs and stitching context manually. Highlight is one timeline, one bill, one open-source codebase.

Highlight
LogRocket + Sentry
Sentry + Datadog + LR
Session replay DOM-accurate, 60fps
Native
LogRocket only
LogRocket only
Backend errors & traces OTel-native
Native
Sentry only
Sentry + Datadog
Logs at scale ClickHouse-backed
Native
Add another tool
Datadog
Replay ↔ backend stack on one timeline shared trace_id
One click
Manual stitching
Three dashboards
Open source AGPL-3.0, full repo
Yes
Closed
Closed
Self-host on your VPC Docker / k8s
Day 1
Enterprise add-on
Sentry only
PII redaction at SDK boundary never leaves the client
Default on
Configurable
Configurable
Estimated bill at 50K MAU based on Mar 2026 rate cards
$390 / mo
$1,290 / mo
$2,470 / mo
06 — Pricing

Pay for sessions,
not for seats.

Predictable usage-based pricing. Unlimited team members on every plan — including Free. Self-host the entire stack for $0, forever.

Free

Side projects, OSS, and the first ten thousand sessions of every startup.

$0
10K sessions · 1M errors · 1M logs / month
  • Full session replay, errors, logs, traces
  • Unlimited team seats
  • 7-day retention
  • Community Discord support
  • Self-host the full stack — forever
Most popular

Team

Growing product teams who’ve outgrown the free tier and the legacy stack.

$150/mo · usage from
200K sessions · 50M errors · 100M logs / month
  • Everything in Free
  • 30-day retention
  • SSO & SAML, audit logs
  • Slack & Linear & PagerDuty integrations
  • Email & Slack-Connect support, <4h SLA
  • AI assist on every session and error

Enterprise

Regulated industries, large teams, and anyone who needs the data inside their own VPC.

Custom
Volume discounts · annual contracts · dedicated CSM
  • Everything in Team
  • Self-host on your VPC, fully managed
  • Custom retention up to 13 months
  • HIPAA BAA, SOC 2 II report, DPA
  • SCIM provisioning, EU/US residency
  • 99.95% uptime SLA, 24/7 on-call

Unlimited team seats on every tier · cancel any time · pay-as-you-go beyond included usage

07 — Customers

Engineers who’ve fired
three tools to keep one.

We replaced LogRocket, Sentry and a half-broken in-house log pipeline with Highlight in a single afternoon. The first thing my staff engineer said was ‘why doesn’t every observability tool work like this?’. Our mean time to resolution dropped from 47 minutes to 9.
EW
Elena WhitfieldVP Engineering · Hightouch
The session replay alone justified the switch. Watching the customer’s exact click path next to the failing trace is the debugging UX I’ve wanted for a decade.
JM
Jordan MarshStaff Engineer · Knock
We’re HIPAA-bound and we self-host the whole stack on our own VPC. Highlight is the only tool in this space that made that path actually pleasant.
RK
Ravi KapoorCTO · Aurora Health
OpenTelemetry-native means I’m not locked in. My ingest pipeline already speaks OTel — Highlight slotted in behind it without changing a single line.
SO
Sofia OkaforPlatform Lead · Whop
The pricing page is the only one I’ve ever read top-to-bottom and not felt like I was being lied to. Usage-based, generous free tier, no per-seat surprise.
MG
Marcus GableFounder · Basedash
Our support team uses Highlight more than the engineers do. They paste a session URL in the ticket and the whole “repro it on my machine” theatre disappears.
PL
Priya LakhaniHead of Support · Knock
08 — FAQ

Twelve answers,
one commitment.

Everything an engineering lead asks before greenlighting a new observability vendor. If we missed one, our docs are public — every single line.

Browse the docs
The full repository — SDKs, ingest pipeline, ClickHouse schema, the Next.js frontend you’re looking at — lives at github.com/highlight/highlight under the AGPL-3.0 license. No “source-available” sleight of hand. You can read every line, fork it, and run the same code we run in our cloud.
DOM mutations at 60fps via rrweb, console output, network requests & responses (with body capture optional), unhandled errors, and any custom event you emit. On the backend, the OpenTelemetry-compatible SDK captures spans, logs and metrics. Average client-side overhead is below 3% CPU.
No, by default. Every input field, password and known-PII attribute is masked at the SDK boundary before the payload leaves the browser. You configure allowlists per domain and per element. We provide a HIPAA Business Associate Agreement and EU/US data residency on the Enterprise plan.
Run docker compose up on any Linux box. The stack consists of a Go ingest service, ClickHouse for storage, and our Next.js frontend. We publish a reference Helm chart for Kubernetes and a Terraform module for AWS. Most teams are recording their first session in under 90 seconds.
Yes — that’s the whole point. Highlight ingests standard OTLP over gRPC or HTTP. Point your collector at our endpoint (or your self-hosted one) and the traces, logs and metrics flow without changing your application code.
Roughly $920/mo at 1M sessions including 250M errors and 500M log lines, on the Team plan. Volume discounts kick in above 5M sessions. Compare to a LogRocket + Sentry + Datadog stack at the same volume, which typically lands between $4,800 and $7,200/mo depending on retention.
First-class SDKs for Next.js, Remix, SvelteKit, Nuxt, React, Vue, Angular, vanilla JS, React Native, Expo, iOS and Android on the client side. On the server: Node, Go, Python, Ruby, Java, Rust, .NET, PHP and Elixir. Anything else can ship via the OpenTelemetry SDK.
Every backend trace and log is automatically tagged with the session ID emitted by the frontend SDK. Open any error and the affected sessions appear inline. Open any session and the corresponding backend trace is one keystroke away.
Yes. Our Vercel integration auto-instruments preview deployments and groups errors by branch. Codespaces and Replit work via the standard SDK plus a single environment variable.
Most teams ship Highlight alongside the existing stack for a week, validate parity, then cut over. We provide a one-click Sentry import for historical issues and grouping rules. The median migration we’ve shepherded ran from kickoff to cutover in 11 days.
A team of seventeen engineers and designers headquartered in San Francisco, backed by Y Combinator, AIX Ventures and a roster of operator angels including the founders of Vercel, PostHog and Linear. We were built by people who got paged at 3 a.m. and decided the dashboards were the problem.
Open the live demo at app.highlight.io/demo — it’s a fully populated workspace with synthetic traffic, no email gate, no “book a demo with sales” popup. The whole product is also runnable locally via the public repo.

Ship the next bug fix
in minutes, not days.

Ten thousand sessions, one million errors, unlimited team seats — free, forever. Plug in a single line of code. Watch the next user’s click trip a backend exception. See it in one timeline. Fix it before they email support.

No credit card 3-min setup Cancel any time Open source, AGPL-3.0