This page describes the safe software deployment practices followed by Genians for Genian Insights E — an integrated PC security platform providing EDR, antivirus, anti-ransomware, and device control.
These practices cover the full deployment lifecycle: development and testing, internal validation, staged deployment, and post-release monitoring and incident response. They are maintained in alignment with the Microsoft Virus Initiative (MVI) Safe Deployment Practice requirements and the joint guidance published by CISA, FBI, and the Australian Cyber Security Centre, Safe Software Deployment: How Software Manufacturers Can Ensure Reliability for Customers (October 2024).
1. Development & Testing
Genians develops Genian Insights E following an Agile methodology. Every code and configuration change passes through a structured series of validation steps before advancing to the next phase.
Code Verification
- Developer testing — Functional requirements are verified immediately upon each code change.
- Peer review — Code quality and security are reviewed via Fisheye and GitHub Pull Requests.
- Usability review — The technical team evaluates feature usability from an end-user perspective.
- Sprint review — Sprint goal achievement is assessed on a weekly basis.
Security Vulnerability Analysis
- Static analysis (SAST) — Monthly source code analysis against OWASP Top 10 and CWE/SANS Top 25 benchmarks.
- Dynamic analysis (DAST) — Runtime vulnerability scanning at every release cycle, utilizing a database of over 62,000 CVE plugins.
Automated Build & QA
An automated build is triggered via CI/CD pipeline on every code check-in. 293 test cases validate functional and performance integrity across each build, with results delivered instantly via Slack notification. This level of automated coverage ensures regressions are caught before any build progresses to internal validation.
Detection Pattern Validation
To prevent false positives from AV detection pattern updates — a risk category not addressed by most endpoint security vendors — Genians maintains a proprietary reference corpus of 694 GB of known-clean executables (646,632 files, covering Windows 7–11 and Windows Server 2008–2025). Every new detection pattern is pre-filtered against this corpus before deployment. This false positive prevention step runs prior to any customer-facing release.
2. Internal Validation (Dogfood)
Before any build is released externally, it is deployed internally across Genians’ own employee workstations and systems — a practice aligned with the CISA-recommended “dogfood” phase.
This internal canary deployment corresponds to the BETA stage of Genians’ four-stage release pipeline (see Section 3). Only builds that complete this internal validation period without anomaly are eligible to advance to external release candidates.
3. Deployment Strategy
Four-Stage Release Pipeline
All builds advance through four controlled stages. Each stage must meet defined quality criteria before the build proceeds. Emergency hotfixes are the only exception to the full pipeline requirement.
| Stage | Description |
|---|---|
| CURRENT | Internal development image, shared among developers only |
| BETA | In-house canary deployment across employee systems for stability validation |
| RC (Release Candidate) | One-month canary deployment to new and select customers |
| R (Release) | General availability to all customers |
All non-emergency changes undergo a minimum validation period of one to two months before reaching general availability. For on-premises environments, Genians engineers perform direct, device-by-device canary deployment before full rollout — a deployment model that provides real-world validation at customer scale prior to broad distribution.
Deployment Integrity Verification
All deployment images are digitally signed using an EV (Extended Validation) certificate. The digital signature is verified at upgrade time; installation is automatically blocked if verification fails. The signing certificate is stored on a dedicated server maintained in an isolated environment with no external network access — preventing unauthorized use of the certificate even in the event of an external compromise.
Detection Pattern Deployment Controls
- Delayed update (canary) — A 2-day delayed update is applied to the general customer base by default, excluding designated canary participants. Customers may configure an additional delay of up to 3 days via product settings.
- Clean-set pre-filtering — All new detection patterns are validated against the 694 GB clean corpus described in Section 1 before release.
Custom Update Options
Customers are provided with update management controls to align deployments with their own operational and security policies:
- Automatic application of the latest available patch via product settings, or manual registration of a specific patch set.
- Selection between the latest engine and a version confirmed stable after a post-release validation period.
- Configuration of canary delay windows to match operational risk tolerance.
4. Monitoring & Incident Response
Agent Self-Protection
Genian Insights E includes built-in self-protection mechanisms that maintain endpoint stability autonomously — an operational resilience capability not present in most comparable products:
- Infinite BSOD prevention — If a BSOD occurs immediately after kernel driver loading, the driver is automatically disabled on the next boot, preventing repeated system failures.
- Deadlock detection & recovery — Per-event processing time is self-monitored; the agent process restarts automatically upon suspected deadlock without operator intervention.
- Crash recovery — Three consecutive abnormal terminations within 10 minutes trigger automatic suspension or automatic rollback to the previous agent version.
- Resource overuse prevention — CPU and memory usage are continuously monitored; the agent is automatically suspended if thresholds are exceeded, preserving endpoint performance.
Log Collection & Diagnostics
Agent logs, crash dumps, and database files can be collected remotely from the management console. For offline or air-gapped environments, a dedicated offline collection tool is provided, covering system information, Process Monitor logs, and kernel memory dumps — enabling root cause analysis without requiring network connectivity.
Rollback Procedures
If anomalies are detected at any stage of progressive deployment, the rollout is immediately halted and a rollback to the previously validated stable version is executed. All rollbacks follow predefined incident response protocols. Findings from each incident feed back into the planning phase of the next development cycle as part of Genians’ continuous improvement process.
Bug Bounty Program
Genians operates a bug bounty program to encourage responsible vulnerability disclosure from external security researchers. All reports are reviewed by the development team and addressed promptly, with outcomes incorporated into subsequent releases.