The Mentis Platform
A unified cognitive architecture for governed enterprise intelligence
All Bespoke Mentis systems are built on a single governed platform - designed for enterprise deployment, regulatory traceability, and operational fit across complex industries. Not a product. The architecture everything else is built on.
Platform Principles
Four architectural commitments that define every system built on the Mentis platform.
Governance-First behavior
Governance is not applied after a system is built - it is the substrate every system is built on. Policy constraints, constitutional rules, and audit requirements are embedded at the architecture layer, not the application layer.
Traceability & Audit Posture
Every decision, recommendation, and data access leaves a cryptographically verifiable record. When regulators, customers, or internal audit teams ask "why did the system do that?" - you have evidence, not assertions.
Modular Industry Adaptation
The platform is not one system. It is a governed architecture that underlies multiple industry-specific systems. Each system inherits the same governance foundation while adapting to the workflows, data, and compliance requirements of its environment.
Secure Data Integration
Enterprise systems operate on sensitive, licensed, and proprietary data. The platform is designed for customer-controlled data integration: no training on your data, no cross-tenant exposure, strict access boundaries at every layer.
The Architecture Is Not New.
The Brain Designed It. We Engineered It.
Mentis Console and Foresight are built on a modular, multi-core architecture that mirrors the structure of the mammalian neocortex - specialist domains, each expert in its function, coordinated under a constitutional governance layer. This is not metaphor. MIT, Berkeley, EPFL, Numenta, and a 2026 arXiv review independently published the research that validates it.
MIT + Nature Communications, 2025
Modular Agentic Planner (MAP)
The prefrontal cortex plans through 5 specialist subsystems - each with a defined function, coordinated by a meta-orchestrator. Merging these into a single monolithic model produces measurably worse performance. This is the governing layer of the Mentis Platform.
View sourcearXiv 2026 - Salatiello · ICLR
"Modularity is the Bedrock of Natural and Artificial Intelligence"
The No Free Lunch Theorem proves monolithic AI is provably suboptimal for multi-domain tasks. Specialist modules with domain-specific inductive biases are mathematically necessary. This is the architectural foundation of Mentis Console and Foresight.
View sourceBerkeley - Alpha Framework, Lawrence Berkeley National Lab, 2025
Governed Modular AI in Safety-Critical Environments
Berkeley deployed a governed multi-agent system at a particle accelerator - the highest-stakes physical infrastructure. Plan-first orchestration, explicit dependency graphs, and human approval gates. Preparation time reduced by two orders of magnitude. Safety maintained.
View sourceNumenta - Thousand Brains Project, 2024
Cortical Column Architecture
The mammalian neocortex operates through ~150,000 semi-independent cortical columns communicating via a Cortical Messaging Protocol (CMP). Abstract knowledge emerges from integration across specialist modules. The Mentis Platform IS this architecture, instantiated in governed AI software.
View sourceAhead of the Curve
Bespoke Mentis was building on these architectural principles before these papers were published. Mentis Console - the platform everything is built on - is the first governed, brain-inspired, multi-core AI engineering system in production software. Foresight is the first deployment of this architecture in regulated pharmaceutical intelligence.
Platform Capabilities
Structured ingestion of internal documents, licensed datasets, and operational data. Full source attribution, versioning, and update tracking - so the system always knows what it knows and where it came from.
Constitutional constraints applied at the inference layer. Policy-aware outputs, role-based access controls on what the system can surface, and behavioural rules that cannot be overridden at runtime.
Multi-step processes with structured handoffs, human-in-the-loop approval gates, and durable execution. Systems that operate inside real workflows, not in isolation from them.
Continuous quality checks, drift detection, and operational audit logs. Systems do not degrade silently - their behavior is observable, measurable, and improvable.
RBAC, encryption at rest and in transit, tenant isolation at the database layer, and key management. Built to map into SOC 2, ISO 27001, and HIPAA-aligned control frameworks from day one.
Cloud, dedicated VPC, or hybrid deployment depending on data residency requirements, security posture, and operational constraints. Each deployment mode maintains the same governance guarantees.
Platform Architecture
A layered architecture designed so governance is foundational - not an add-on - and so every output system inherits the same guarantees.
Engagement Model
Understanding your environment, constraints, data landscape, and governance requirements.
Architecture blueprint tailored to your workflows, compliance requirements, and operational fit.
Controlled build and deployment in a bounded environment before full rollout.
Expansion across use cases, teams, and systems as the deployment matures.
Ready to Build on a Governed Foundation?
Tell us about your environment. We'll assess fit, outline a system design, and walk you through what a governed deployment looks like in practice.
Common Questions