AI Transparency & Security
How we use AI, what we protect, and the commitments we make to every user. No hype, no hidden practices.
SSL/TLS Encrypted
All data in transit
Cloudflare Protected
DDoS & bot mitigation
Stripe PCI-DSS
Payment security
No Data Sales
Your data stays yours
What AI Is Used For
SkillFocusLab uses OpenAI's language models (GPT-4o-mini) to provide personalized coaching and analysis. AI is used in the following areas:
Skill Analysis
After you complete a typing test, grammar check, or other tool, AI analyzes your results and provides specific, actionable feedback to improve your performance.
Resume Feedback
AI reviews your resume text to identify formatting improvements, missing keywords, ATS compatibility issues, and content gaps. It does not store your resume after analysis.
Grammar Explanations
When you submit text for grammar checking, AI identifies errors and explains the underlying grammar rules so you understand the corrections.
Customer Service
Our AI support assistant answers questions about your account, billing, features, and usage. It operates within strict policy guidelines and will recommend human escalation for complex issues.
Cancellation Assistance
If you choose to cancel, AI helps explain your options (pause, downgrade, or cancel) without pressuring you. It respects your decision and facilitates the process efficiently.
Career Tools
Promotion readiness assessment, career skill-gap analysis, writing tone evaluation, and reading comprehension coaching all use AI to generate personalized recommendations.
What AI Is NOT Used For
We believe in clear boundaries. AI on this platform is limited to the use cases above. It does not:
Replace human career advisors or recruiters
AI provides data-driven suggestions. For career decisions involving significant life changes, we recommend consulting qualified professionals.
Store or process sensitive financial data
All payment processing is handled by Stripe (PCI-DSS Level 1 certified). We never see, store, or transmit your card numbers.
Access external private accounts or platforms
AI does not connect to LinkedIn, email providers, social media, or any external accounts. It only processes data you voluntarily submit within our tools.
Make automated decisions that affect employment
No employer, recruiter, or third party receives data from your account. Your scores, analyses, and progress are visible only to you.
Data Handling
| Data Type | Stored | Shared | Purpose |
|---|---|---|---|
| Email & Name | Yes | Never | Account authentication |
| Password | Hashed + Salted | Never | Cannot be reversed or read |
| Typing / Grammar Results | Yes | Never | Progress tracking & Skill Score |
| Resume Text | Temporary | Never | Processed for analysis only, not permanently stored |
| Payment Info | Stripe Only | Never | Handled entirely by Stripe (PCI-DSS Level 1) |
| AI Conversation Logs | Session Only | Never | Not retained after your browser session ends |
Data commitment: Your personal data is never sold, rented, or shared with advertisers. We do not build advertising profiles. Free-tier ads are served contextually by Google AdSense and are not targeted based on your SkillFocusLab activity.
AI Cost & Usage Logic
We believe in full transparency about how AI costs are managed. Here is exactly what model we use, what it costs, and how we control spending.
Model Configuration
| Parameter | Value | Rationale |
|---|---|---|
| Model | gpt-4o-mini | Best cost/quality ratio for coaching tasks |
| Max Tokens | 500-800 | Varies by endpoint; capped to control cost |
| Temperature | 0.3-0.7 | Lower for factual analysis, higher for creative feedback |
| Input cost | $0.00015/1K tokens | OpenAI published rate (as of Feb 2026) |
| Output cost | $0.0006/1K tokens | OpenAI published rate (as of Feb 2026) |
| Avg cost/call | ~$0.001-$0.003 | Typical coaching response costs less than 1 cent |
Usage Limits by Tier
Cost Control Mechanisms
Identical requests return cached responses (5 min - 2 hr TTL), avoiding duplicate API calls and reducing cost by ~30%.
Before any AI call, we strip emails, phone numbers, SSNs, card numbers, and physical addresses from input text.
Every AI call is logged (user_id, endpoint, tokens, cost, latency, model, cached flag). Logs retained 90 days, then purged.
Per-minute (5), per-session (10/15min), and daily caps enforced server-side. Cannot be bypassed from the client.
Why we disclose this: AI costs are a real operational expense. By being transparent about model choice, pricing, and usage limits, we help users understand the value exchange and trust that their subscription fees are managed responsibly. See our methodology for how AI scoring inputs are handled.
Security Measures
Cloudflare Protection
- • DDoS mitigation on all endpoints
- • Web Application Firewall (WAF)
- • Automated bot detection and blocking
- • Global CDN with edge caching
Encryption
- • TLS 1.3 encryption on all connections
- • HSTS headers with preload directive
- • Passwords hashed with cryptographic salt
- • API secrets stored as encrypted environment variables
Rate Limiting
- • 60 API requests per minute per IP
- • 5 AI calls per minute per user
- • 10 support messages per minute
- • Automatic cooldown with retry-after headers
Code Security
- • GitHub CodeQL static analysis
- • OWASP ZAP vulnerability scanning
- • Snyk dependency vulnerability checks
- • Security headers (X-Frame, XSS, HSTS, CSP)
AI Limitations & Disclaimers
We use AI responsibly, but AI systems have inherent limitations. Users should be aware of the following:
AI may generate imperfect suggestions
Language models can occasionally produce inaccurate, incomplete, or contextually inappropriate responses. Our system prompt engineering minimizes this, but it cannot be eliminated entirely.
Users should verify important decisions independently
Career advice, resume optimizations, and skill assessments are informational tools. For significant career decisions, consult with qualified professionals.
Resume suggestions are advisory, not guaranteed
AI-generated resume improvements may increase ATS compatibility but cannot guarantee interview callbacks or job offers. Results depend on many external factors.
AI does not learn from your data
We use OpenAI's API with data processing agreements in place. Your inputs are not used to train or improve OpenAI's models.
Contact & Escalation
Human Review
Required for: account deletion, data export, refund disputes, security incidents.
Response within 48 hours
When does human review apply? Our AI support system will automatically recommend human escalation when it detects: billing disputes over $50, account security concerns, data deletion requests (GDPR/CCPA), technical issues it cannot resolve, or any situation where the user explicitly requests human assistance.
Last updated: February 2026. This page is reviewed and updated quarterly.