ReplayGuard by Toln

Know what breaks
before you ship.

Replay sanitized production API traffic against your next release. Find behavioral, schema, and performance regressions before your customers do.

Request early access ↗ ▶ See how it works
Customer-hostedAWS-nativeFail-open capture
REPLAY COMPLETE
release-4.28
97/100
Ready to deploy

Behavior matches production

Requests tested842,921
Coverage94.8%
Critical0
Affected traffic0.03%
P95 LATENCY

BUILT FOR TEAMS SHIPPING CRITICAL APIS

FINTECHCOMMERCEHEALTHPLATFORMSINFRASTRUCTURE
01 / THE GAP

Your tests know what you expect.
Production knows what is real.

Edge cases hide in legacy clients, strange payloads, forgotten endpoints, and undocumented integrations. ReplayGuard turns that real behavior into a reusable regression suite.

01

Capture reality

Sample real API requests and responses without interrupting production traffic.

02

Remove the risk

Redact sensitive fields before anything reaches storage, logs, or queues.

03

Validate change

Replay safely against candidate releases and surface meaningful differences.

02 / THE WORKFLOW

Real traffic. Clear signal.

One continuous path from observed production behavior to an evidence-backed release decision.

01↙CaptureProduction traffic
02✦SanitizeSecrets removed
03↻ReplayCandidate release
04≠CompareSemantic diff
05✓ShipWith confidence
03 / CORE CAPABILITIES

Everything between traffic
and a confident release.

Production observability, safe replay, and regression analysis in one customer-controlled system.

↙
CAPTURE

Production traffic recorder

Capture REST and JSON requests through a fail-open reverse proxy with configurable sampling.

  • Path, method, status, and header rules
  • Endpoint discovery and inventory
  • Request, response, and latency metadata
✦
PRIVACY

Boundary-first sanitization

Remove sensitive fields before captured data enters storage, queues, logs, or traces.

  • Credential-header masking
  • Configurable JSONPath redaction
  • Metadata-only unsafe-body handling
↻
REPLAY

Controlled request replay

Transform traffic for candidate environments while enforcing explicit safety boundaries.

  • Host and CIDR allowlists
  • Header, query, and path transforms
  • Concurrency and timeout controls
≠
DIFF

Semantic comparison

Separate meaningful behavior changes from expected dynamic values and identifiers.

  • Status, value, and type comparison
  • Schema and nullability changes
  • Configurable ignore rules
⌁
PERFORMANCE

Latency regression detection

Surface candidate endpoints that move beyond the performance thresholds you define.

  • Per-request latency comparison
  • Endpoint P50, P95, and P99
  • Configurable release policies
◎
INTELLIGENCE

API behavior inventory

See what clients use, which schemas are active, and where replay coverage is incomplete.

  • First and last observed times
  • Error rate and schema history
  • Replay coverage by endpoint
04 / USE CASES

Built for consequential changes.

Use production-derived evidence wherever backend behavior can drift from what customers depend on.

SHIP WITH EVIDENCE

Validate candidates against the behavior customers use today.

Replay representative production requests after deploying to staging. Group regressions before promoting the release.

  • Catch status, schema, and value regressions
  • Estimate affected production traffic
  • Build repeatable golden collections
EXAMPLE RELEASE GATE
0Critical regressions
≤20%P95 latency increase
100%Golden suite complete
MODERNIZE SAFELY

Prove that a replacement API preserves real client behavior.

Test framework rewrites, database migrations, and service extractions against real request shapes.

  • Compare implementations side by side
  • Find active legacy endpoints
  • Detect type and nullability drift
MIGRATION SIGNALS
183Active endpoints
12Client versions
94%Behavior covered
CONTROL ROLLOUT RISK

Compare candidate behavior before increasing exposure.

Hold canary promotion when behavior or latency moves outside your release policy.

  • Reuse evidence at each rollout stage
  • Track regressions by release
  • Cancel unsafe sessions immediately
CANARY POLICY
5%Initial exposure
15mObservation window
PASSPromotion decision
LEARN BEFORE IMPACT

Turn past edge cases into permanent coverage.

Save sanitized requests that exposed previous bugs, then replay them before every relevant release.

  • Preserve failure-inducing scenarios
  • Prevent recurrence without rebuilding payloads
  • Connect findings to real behavior
REGRESSION LIBRARY
24Saved collections
2,842Golden requests
0Repeated incidents
05 / SIGNAL, NOT NOISE
GET /customers/:id
PRODUCTION · 200
{
  "id": 4123,
  "status": "active",
  "balance": 25000
}
CANDIDATE · 200
{
  "id": 4123,
  "status": "active",
  "balance": null
}
HIGH
Schema regression detected$.balance changed from integer → null

Compare behavior,
not bytes.

ReplayGuard understands JSON types, schemas, arrays, ignored fields, status codes, and latency. Dynamic values stay quiet. Breaking changes stand out.

  • ✓ Type-aware JSON and schema diffing
  • ✓ Configurable dynamic-field ignores
  • ✓ Performance regression thresholds
  • ✓ Grouped findings by affected endpoint
06 / BUILT TO FAIL SAFE

Production data deserves
serious boundaries.

Sensitive payloads stay in your environment. Capture failures never take down your API. Uncertain replay destinations are blocked.

Read the security brief ↗
01Redact before storageHeaders and JSON paths are sanitized at the capture boundary.
02Allowlisted replay onlyHost, CIDR, method, path, redirect, and DNS checks default to deny.
03Your account. Your data.PostgreSQL, S3, KMS, and secrets remain in your AWS account.
07 / GOVERN WITH CONFIDENCE

Control who can see,
change, and replay.

ReplayGuard gives platform and security teams clear ownership without slowing down the engineers validating releases.

IDENTITY

Use your identity provider

Authenticate through customer-managed OIDC and map existing groups to product roles.

  • Five purpose-built access roles
  • No ReplayGuard password database
  • Server-side permission enforcement
ACCOUNTABILITY

Keep an audit trail

Understand who accessed traffic, changed policy, or initiated a replay.

  • Append-only administrative history
  • Authorization failure records
  • Replay, export, and policy events
OPERATIONS

Set explicit limits

Bound traffic retention and replay behavior at the project and environment level.

  • Retention and body-size policies
  • Concurrency, rate, and timeout controls
  • Emergency cancellation for active runs
08 / DEPLOY YOUR WAY

Inside your AWS boundary.

Run ReplayGuard in the infrastructure your team already trusts. Start with ECS/Fargate or EKS and keep payloads under your own encryption and retention policies.

EKSECSFargateRDSS3KMS
YOUR AWS ACCOUNT
Capture
fail open
→
Sanitize
at boundary
→
Store
your KMS key
Candidate API
←
Safe replay
←
Traffic set
09 / FITS YOUR STACK

AWS-native by design.

Use services your platform and security teams already operate. Production payloads remain in your account.

COMPUTEEKSECSFargateALB
DATA & SECURITYRDSS3KMSSecrets Manager
IDENTITYOIDCOktaEntra IDGoogle Workspace
OBSERVABILITYOpenTelemetryCloudWatchPrometheusGrafana
10 / SIMPLE START

Start with the traffic that matters.

Early-access plans are designed around captured and replayed traffic—not seats. Final Marketplace pricing will be published before general availability.

DEVELOPER

Explore safely

For individual engineers validating a service or proof of concept.

  • 1 project
  • Manual capture and replay
  • 7-day retention
  • Community support
Join early access
MOST RELEVANT
TEAM

Build release confidence

For engineering teams bringing production-derived tests into delivery workflows.

  • Multiple projects and services
  • Golden traffic collections
  • 30-day retention
  • Priority onboarding
Talk to Toln
ENTERPRISE

Control the boundary

For regulated and high-scale organizations with strict networking and support needs.

  • Customer-hosted data plane
  • Custom retention and volume
  • Enterprise identity and audit
  • Private Marketplace offer
Discuss requirements

No credit card required for early-access conversations. Infrastructure costs remain in your AWS account.

11 / QUESTIONS

Before you bring
production traffic.

Clear answers for platform, security, and engineering teams evaluating ReplayGuard.

Does production data leave our AWS account?+

The customer-hosted model keeps capture, payload storage, PostgreSQL, replay workers, and encryption keys in your account. Toln does not need raw production payloads to operate the product.

Can ReplayGuard affect production availability?+

The capture path is designed to fail open. If recording, storage, or ReplayGuard itself is unavailable, application traffic continues to the upstream service.

How do you prevent dangerous replays?+

Replay defaults to safe methods and validates schemes, hosts, CIDRs, resolved addresses, redirects, methods, and paths. Mutating requests require explicit policy and confirmation.

What happens to authorization headers and PII?+

Known credential headers are masked by default. Customer-defined JSON paths are redacted at the capture boundary before payloads enter persistent systems.

Does this replace unit and integration tests?+

No. ReplayGuard complements deterministic tests with scenarios derived from actual client behavior, unusual payloads, legacy integrations, and real traffic distributions.

Which APIs are supported first?+

The initial release focuses on HTTP REST APIs with JSON bodies. GraphQL, gRPC, OpenTelemetry ingestion, and session-aware replay are planned extensions.

T/
BUILT BY TOLN LLC

Tools for teams who carry the pager.

Toln builds infrastructure software around a simple idea: engineering teams should have better evidence before making consequential changes.

PRIVATE EARLY ACCESS

Your next release should
come with evidence.

Bring your APIs. We’ll help you turn real traffic into a safer release process.