Architecture & Governance
In Enterprise Local Node mode, raw customer data is processed inside the customer-controlled environment. AI workloads can be configured to use local providers such as LouhiGPT Local or vLLM Local. No raw data is transmitted to external systems unless the customer explicitly enables a hosted provider.
Raw strategy documents stay local
Raw signals stay local
Vector indexes stay tenant-scoped
Audit logs remain under customer control
External sharing is disabled by default
Every model run is recorded
Every human validation is auditable
Every action card status change is auditable
Every report export is traceable
Provider status is visible through System Health
No employee scoring
No individual emotional profiling
No autonomous HR decisions
Human validation required for high-impact findings
Risk analysis targets programs, portfolios and execution structures — not individuals
TALKTO+ can analyze aggregated project climate signals from bounded collaboration contexts. It does not create individual emotional profiles, employee scores or HR decision outputs.
TALKTO+ is designed around privacy-by-design principles. Customer data processing, retention, access control and deployment boundaries are configured for each enterprise environment. Each deployment is the customer's own data processing context — no assumption of shared infrastructure applies.
Note: TALKTO+ does not claim GDPR certification. Each customer deployment requires its own legal and data protection review.
TALKTO+ is designed to support AI Act readiness through human oversight, auditability, role-based access, clear data boundaries and transparent system health. Workplace deployment requires customer-specific legal review, worker information processes and governance alignment.
Note: TALKTO+ does not claim full AI Act compliance. Regulatory readiness depends on each customer's specific deployment context and jurisdiction.
Controls
Security Controls
RBAC and tenant isolation
HMAC, timestamp and nonce for webhooks
Sanitized logs
Secrets management
Provider health checks
Zero-egress misconfiguration detection
model_runs audit log
Report metadata tracing
Internal swarm disabled by default
Architecture Pack
Request Security Architecture Pack
Detailed documentation on deployment boundaries, audit logs, data flows and AI provider configuration.
Contact Us →