Skip to main content

Feature Roadmap

Mandaitor is evolving rapidly to provide the most comprehensive trust infrastructure for autonomous AI agents. While our core delegation, verification, and audit capabilities are already available, we are continuously expanding the platform to address emerging market needs, regulatory requirements, and complex multi-agent workflows.

This roadmap outlines the features we are currently developing or planning for future horizons.

Recently Completed GA Hardening​

The enterprise-beta hardening track has closed the core Horizon 0 slices required for pilot readiness: fine-grained RBAC adoption for high-risk tenant actions, verifiable evidence-pack export manifests, tenant audit/SIEM export, and Trust Operations visibility for export integrity. Wave 4 adds the pull-request CI gate for the golden evidence-pack verifier and publishes evidence_pack_exports in public trust data so operators can review manifest digest, schema version, verifier status, event count, and the latest EVIDENCE_PACK_EXPORTED audit event from the dashboard.

Wave 5 completes the Horizon 0 GA-readiness closeout. Retention defaults, evidence-pack verification posture, tenant audit/SIEM export boundaries, customer-managed KMS-at-rest scope, data-residency attestation status, and GA SLA posture are now represented in a machine-readable GA readiness policy and projected through generated trust data. The remaining boundary is product launch governance: beta accounts continue to operate on best-effort support, while eligible GA plans will receive the published SLA and formal GA support terms.

Horizon 2 (Q3-Q4 2026)​

Multi-Agent Delegation Chains​

As AI agents increasingly collaborate, they will need the ability to sub-delegate authority to other agents. This feature will enable a chain-of-trust model where an agent can grant a subset of its own mandate to a downstream agent. The system will ensure that delegated authority strictly narrows and never exceeds the original constraints.

Real-Time Constraint Enforcement Engine​

We are upgrading our verification model from periodic point-in-time checks to a continuous, real-time streaming engine. This will actively monitor agent actions as they occur, allowing Mandaitor to instantly revoke mandates or trigger alerts the moment behavioral constraints are violated, moving the platform from periodic to continuous governance.

Compliance Report Generator​

To reduce the compliance burden on our customers, we are building automated report generation tools. These tools will synthesize data from mandate audit trails and format it specifically for regulatory frameworks such as the EU AI Act (Article 14), PSD3 SCA, and HIPAA, complete with human-readable summaries and integrated evidence packs.

Horizon 3 (Q1-Q2 2027)​

Federated Trust Network​

Agentic commerce requires agents from different organizations to interact securely. We are developing a federated trust network that will allow cross-organization mandate verification without relying on a centralized authority. Organizations will be able to establish bilateral or multilateral trust relationships, enabling agents to present verifiable mandates across boundaries.

Behavioral Fingerprinting & Anomaly Detection​

To combat sophisticated attack vectors like prompt injection or account compromise, we are introducing ML-based behavioral profiling. This system will detect when an agent's action patterns deviate significantly from its historical behavior, even if the individual actions technically remain within the mandate's defined scope.

Trust Scoring & Contract Enforcement​

We are expanding our semantic analysis capabilities to provide dynamic trust scores for AI agents based on their compliance history and violation patterns. Furthermore, we are exploring active runtime contract enforcement, which will not only detect but automatically block, modify, or escalate agent actions that violate behavioral contracts.

Auto Recovery​

In the event of detected drift or mandate violation, Mandaitor will introduce automated correction mechanisms. This feature aims to automatically roll back states, modify problematic actions, or redirect agent workflows to ensure compliance without requiring manual human intervention.

FIDO Alliance Standards Contribution​

Mandaitor is actively working to formalize our Verifiable User Instructions (VUI) and Proof-of-Mandate (PoM) specifications. We plan to contribute these as reference implementations to the FIDO Alliance working groups, helping to shape the industry standards for agent authentication and trusted delegation for commerce.

Note: This roadmap is subject to change based on market feedback, regulatory shifts, and technological advancements. For the current status of implemented features, please refer to our Beta Limitations page.