Release Risk Control System · Founder-led beta

Release risk control
for QA and engineering teams

Dulvarn analyzes pull requests, CI signals, changed areas and test impact to help your team decide what to test — and whether a release is GO, CONDITIONAL GO or NO-GO.

Built for teams that want fewer production surprises, clearer release decisions and stronger QA visibility.

Request beta reviewView demo workflow
Human-controlled decisions·GitHub-first workflow·Audit trail
northwind/checkout-service·#1284
live analysis
68Risk score
CONDITIONAL GO
ScopePartial NRT
Release decision · RD-2148
Refactor coupon stacking + idempotent payment retries
@m.kovac·12m ago·CI passing
Test integrity74/100
Coverage on changed areas
Code risk68/100
Payment retry hotspot
Changed areas
checkoutcartauthpayment cfg
Blocking signals
  • Missing impacted tests on payment retry path
  • Medium code risk in coupon stacking module
  • No coverage on partial-refund regression scenario
Human decisionPending review
Open decision

Live decision example from a checkout-service PR · representative data

Decision-oriented
GO · CONDITIONAL · NO-GO
GitHub-first
Status checks on every PR
Test scope clear
Smoke · Partial · Full NRT
Human-controlled
AI explains. Humans decide.
Auditable
Every override is logged
The problem

Releases break when teams lack visibility.

Most release incidents are not surprise bugs — they are known unknowns. Risky changes, weak tests, ambiguous CI signals. The work to evaluate them is real, but scattered.

PRs change risky areas without clear test impact

Regression scope is guessed manually, repo by repo

CI passes — but product risk remains unclear

QA signs off without enough context on what changed

Release decisions are scattered across Slack, Jira, GitHub

Teams don't know if Smoke, Partial NRT or Full NRT is enough

Workflow

From pull request to release decision

One linear flow. Every signal is captured. Every decision has a reason. Humans stay in control of what ships.

  1. Step 01

    Connect repository

    Install the Dulvarn GitHub App on the repos you ship from.

  2. Step 02

    Analyze pull request

    Dulvarn parses the diff, changed areas, and PR metadata.

  3. Step 03

    Evaluate signals

    Code risk, CI signals, test impact and historical hotspots.

  4. Step 04

    Recommend test scope

    Smoke, Partial NRT or Full NRT — with the reasoning.

  5. Step 05

    Produce release decision

    GO, CONDITIONAL GO or NO-GO posted as a status check.

  6. Step 06

    Human reviews and decides

    QA or release manager confirms, overrides or escalates.

  7. Step 07

    Store in audit trail

    Decision, signals and override reasons stay traceable.

Modules

Six modules. One release decision.

Each module captures a real workflow problem. Together they turn scattered PR, CI and QA signals into one auditable release decision.

Module 01

PR Risk Analysis

Understand what changed and where risk is concentrated — file, area, hotspot.

checkoutcartpayment cfgauthtax
Module 02

Test Scope Recommendation

Decide whether Smoke, Partial NRT or Full NRT is needed for this change.

Smokeavailable
Partial NRTRecommended
Full NRTnot needed
Module 03

Release Guard

Convert risk signals into GO, CONDITIONAL GO or NO-GO — posted as a status check.

Status checkCONDITIONAL GO
dulvarn/release-guard · checks.run #4827
Module 04

Test Integrity

Detect weak coverage, missing impacted tests and risky blind spots.

78
86
64
92
74
Module 05

Audit Trail

Every decision, override and signal is traceable.

  • decision_generated
  • human_override_added
  • status_check_posted
Module 06

Human Decision Layer

AI explains risk. Humans make the final release call. Always.

QA Lead approves
j.harlan · with override note
Built around real workflows

Built around real QA and release workflows

Dulvarn is built around practical release, regression and CI/CD problems. It helps QA and engineering teams turn scattered signals into clear, human-reviewed release decisions.

  • Human-controlled release decisions
    Override required, with a reason — every time.
  • Transparent risk reasoning
    Every score links back to a specific signal.
  • GitHub-first workflow
    Status checks, PR comments, branch policies.
  • CI/CD friendly
    Reads your existing pipelines — no rewrite.
  • Practical QA workflows
    Smoke / Partial / Full NRT, not vague AI advice.
  • No fake AI autopilot
    Dulvarn does not release. Your team does.
Founder-led beta

Join the founder-led beta

Dulvarn is currently being tested with a small number of QA and engineering teams that want better release risk visibility before production. Limited slots, hands-on onboarding.

Request beta accessBook release risk audit
What we test
Release decisions on real PRs
What you get
Early access · workflow review
What we ask
Honest feedback · 1 review call
Pricing
Early-access advantage
Every override is logged · audit trail enabled · no fake autopilot.