Frequently Asked Questions

Alchevion combines scientific behavioral sciences (Bloom’s Taxonomy, Grounded Theory, Angoff Method) with advanced AI architecture to completely eliminate intuitive biases. Instead of static surveys or template interviews, MentisRA delivers dynamic simulations (Skill Gym) that adapt in real time to candidate responses, backed by transparent behavioral evidence trails—transforming human resource decisions into a verifiable, auditable science.

MentisRA is not merely an automated screening tool; it is a decision-support ecosystem built on Human-in-the-Loop principles and AI Act readiness. Candidates are evaluated free from cognitive biases via our Blind-Screening module; collected data is scored using dual-LLM architecture (Shadow Evaluator), and the final report is audited and validated by expert clinical psychologists to make it fully legally auditable and ethical.

Operating under the guiding motto “Right Candidate, Right Position, Right Future,” Alchevion significantly reduces mis-hire costs and employee turnover. Post-selection, the platform automatically assigns a Gap-Driven Personal Development Plan (PDP) and leverages the R-Memory 2.0 spaced repetition algorithm targeted specifically at identified competency gaps, securing talent transformation and succession planning end-to-end.

The platform goes beyond merely ingesting Job Description (JD) documents. Our organizational psychologists conduct exploratory interviews with leadership and analyze the data via Grounded Theory to construct custom “Culture Cards” and “Red Lines” parameters (JSON). The MentisRA Content Factory utilizes these parameters to generate fully customized crisis and dialogue scenarios, mirroring your exact corporate culture.

In a critical field like human resources, where human lives and careers are at stake, ‘Black Box’ algorithms are strictly unacceptable. MentisRA is designed on a pure Glass Box AI (Explainable AI) architecture. Evaluations are supported by SHAP/LIME analyses, and every score assigned to a candidate by the system’s Dual-LLM controllers is presented with a transparent ‘behavioral evidence trail’—detailing the exact word or behavioral reaction the score relies upon. The entire system is fully open to legal auditing within a strict causal framework.

MentisRA does not operate as an unrestricted, generic natural language processing model. Your company’s custom “Culture Card” and “Red Lines” are embedded directly into the system as “Absolute Constraints” via JSON structures. Our model assumes no external facts proactively; it exclusively labels observed candidate behaviors within a confined scope (Closed-Domain RAG) driven by predefined, strict scientific evaluation rubrics (Grounded Theory). Furthermore, every final calibration undergoes supervision by our expert clinical psychologists, guaranteeing our unwavering Human-in-the-Loop principle.

Absolutely. The Alchevion ecosystem is structured on an ‘API-first’ architecture. With highly secure FastAPI endpoints, scalable Webhook mechanisms, and an advanced RESTful infrastructure, it integrates seamlessly into your company’s existing ATS (Applicant Tracking System) or HRIS platforms. You can trigger evaluation simulations directly from your current ATS interface and configure MentisRA results to flow autonomously into your internal database through our headless architecture support.

Our structural backend is rigorously grounded in Bias-Free Engineering. All protected demographic factors, including the candidate’s age, gender, and ethnicity, are systematically masked from the evaluation model using ‘Blind Scoring’. Additionally, the engine deploys ‘Adversarial Debiasing’ protocols, and its decisions are regularly stressed against the ‘Four-Fifths Rule’. By enforcing a Demographic Parity margin of < 0.05 combined with high scientific internal consistency (Cronbach’s α ≥ 0.80), the system outputs purely objective metrics based solely on capacity and demonstrated competence.

Alchevion’s infrastructure applies PostgreSQL Row Level Security (RLS) architecture. Because data is locked at the database level with a Tenant ID, data leakage or cross-tenant data mixing is mathematically impossible, even if the application layer is bypassed. Full compliance with KVKK, GDPR, and EU AI Act (High-Risk AI Systems) standards is guaranteed. Candidate data is processed anonymously via Blind Screening, and no customer data is ever used to train underlying models without explicit authorization.

All LLM communications run through enterprise APIs protected by strict Zero Data Retention contracts. Data is processed within a Closed-Domain RAG architecture where third-party model providers are technically and legally prohibited from retaining or training on customer data. All data streams are protected end-to-end with TLS 1.3 in transit and AES-256 encryption at rest.

Alchevion enforces a strict Autonomous Separation of Powers between technical optimization and security/data governance. While the CTO team oversees system architecture, API integrations, and model performance, the CISO function independently audits security policies, RLS isolation, and immutable audit trails. In the event of a security anomaly or ethical deviation, the CISO holds autonomous authority to trigger circuit-breaker protocols independently.

During a P1 security incident or data breach, an automated War Room procedure is triggered immediately with direct CISO oversight. Infrastructure is deployed across containerized Docker environments with geo-redundant storage and automated daily backups. Operating under a Zero Data Loss objective, impacted components are isolated immediately, and failover services resume within minutes. Regular third-party penetration testing and SOC2 Type II / ISO 27001 compliance audits ensure continuous resilience.

MentisRA simulations are engineered using scientific evaluation rubrics grounded in clinical psychology and organizational behavior frameworks (Bloom’s Taxonomy, Grounded Theory, Angoff Method). Instead of anxiety-inducing robotic interviews, candidates experience natural, dynamic dialogue scenarios (Skill Gym) mirroring real-world workplace challenges. Evaluation decisions are never left solely to AI; dual-LLM scoring pairs with expert psychologist oversight (Human-in-the-Loop) to deliver objective, empathetic, and psychometrically valid assessments.

The platform dynamically maps employees’ existing skill sets and future potential across a live Skill Tree and Competency Gap Analysis framework. By structuring internal mobility (Talent Mobility Blueprint) and critical leadership succession (Succession Architecture Blueprint) around verifiable behavioral evidence trails rather than intuitive guesswork, MentisRA significantly reduces time-to-fill for key roles and mitigates voluntary turnover.

MentisRA moves beyond diagnostic evaluation by instantly generating Gap-Driven Personal Development Plans (PDP) tailored to each identified competency deficit. Utilizing an embedded R-Memory 2.0 spaced repetition algorithm alongside interactive micro-simulations, the platform empowers employees to continuously practice essential behavioral competencies on the job, maximizing the impact and retention of corporate learning investments.

MentisRA is designed not to replace human HR professionals, but to amplify their strategic capabilities through Workforce Augmentation. By embedding custom organizational values and “Culture Cards” into evaluation models, all hiring and promotion assessments remain deeply aligned with your company’s unique ethos. This synergy between human expertise and AI efficiency eliminates administrative overhead, enabling HR leaders to focus on high-touch strategic talent development and culture building.

In traditional recruitment, the true cost of a mis-hire extends far beyond base salary—reaching up to 12x the annual compensation when factoring in lost productivity, retraining, and replacement costs. MentisRA reduces hiring error rates by 40-80% through its evidence-based architecture. Based on our data, the platform generates a net ROI of approximately 4.5x to 5x when optimally deployed. Mid-market companies typically achieve a payback period of under 3 months, while fully scaled Enterprise clients realize returns even faster.

We utilize a ‘Pay-as-you-grow’ SaaS model designed to eliminate unused license waste. For our advanced “Tier C – Enterprise Package” clients, we provide high-volume Headless API access, dedicated server capacity, and company-specific isolated language models. Rather than relying on traditional static licensing tiers, we structure flexible commercial agreements based on assessment volume (% Volume-Based) and the tangible competency acquisition value delivered to the organization (Value-Based).

Absolutely. MentisRA’s core backend is built end-to-end on a Multi-tenant architecture designed specifically for the enterprise market. A holding company can govern each of its subsidiaries within completely isolated workspaces enforced by Row-Level Security. Additionally, we offer a comprehensive White-label solution for integrated resellers and HR consultancies, enabling them to operate and deliver MentisRA 100% under their own branding and domains.

MentisRA’s native 9-language support operates flawlessly across cross-border operations. For instance, you can run a simulation and interview for a candidate in Berlin in German, manage the operation locally in Turkish, and view the entire group’s behavioral intelligence metrics synthesized into a single common language (e.g., English) on the executive leadership dashboard.