A practical path from a self-declared record to profile-specific assessment, ongoing assurance and custom institutional programmes.
These are scoped offers, not a universal safety ranking. Every SOV result is bound to a defined resource, version, evidence set, authority envelope and operating context.
Sovrentic is building an inspectable assurance scheme for agentic resources. SOV is Sovereign Operational Verification.
01 / Comparison
Compare the SOV offer
SOV-Agent, SOV-Tool and SOV-Skill are resource profiles, not successive rungs of a safety ladder. No prices are published while the offers remain in development.
Comparison of the SOV-1 Declaration, SOV-Agent, SOV-Tool, SOV-Skill and SOV-C Custom Assurance offers across commercial model, scope, evidence, testing, lifecycle, status and boundaries
Feature
SOV-1 DeclarationSelf-declared · not assessed · proposed
Assessing an AI agent or multi-agent configuration
Resource scope
Agents and multi-agent configurations
Structural record
Included
Identity and version binding
Assessed for the named configuration
Authority and permission envelope
Assessed within scope
Evidence references
Evidence package and review
Behavioural or scenario testing
Agent scenarios, autonomy, delegation and abuse boundaries
Human assessment
Planned assessor review where required
Change triggers and reassessment
Planned profile rules
Ongoing assurance
Optional recurring service
Status and revocation
Planned workflow
GRC and compliance mapping
Mapped evidence for organisational decisions
Public record or ledger anchoring
Optional
What it does not establish
No universal safety or legal-compliance conclusion
SOV-Tool
Proposed
Commercial model
Paid assessment, planned
Best for
Assessing an agent-invoked tool, API or connector
Resource scope
Tools, APIs and connectors
Structural record
Included
Identity and version binding
Assessed for the named tool version and contract
Authority and permission envelope
Assessed for access and side effects
Evidence references
Evidence package and review
Behavioural or scenario testing
Misuse, boundary and side-effect testing
Human assessment
Planned assessor review where required
Change triggers and reassessment
Planned profile rules
Ongoing assurance
Optional recurring service
Status and revocation
Planned workflow
GRC and compliance mapping
Mapped evidence for organisational decisions
Public record or ledger anchoring
Optional
What it does not establish
Does not certify every agent using the tool or provide a universal security guarantee
SOV-Skill
Proposed
Commercial model
Paid assessment, planned
Best for
Assessing a reusable skill or instruction package
Resource scope
Skills, instructions and bounded capabilities
Structural record
Included
Identity and version binding
Assessed for the named skill version and dependencies
Authority and permission envelope
Assessed for invocation boundaries
Evidence references
Evidence package and review
Behavioural or scenario testing
Composition, prompt-injection and abuse testing
Human assessment
Planned assessor review where required
Change triggers and reassessment
Planned profile rules
Ongoing assurance
Optional recurring service
Status and revocation
Planned workflow
GRC and compliance mapping
Mapped evidence for organisational decisions
Public record or ledger anchoring
Optional
What it does not establish
Does not establish safe behaviour in every model or context
SOV-C Custom Assurance
Bespoke engagement · Planned
Commercial model
Bespoke consulting engagement, planned
Best for
Designing a tailored organisational or institutional programme
Resource scope
Custom portfolio, system or issuer workflow
Structural record
Tailored
Identity and version binding
Defined for the engagement
Authority and permission envelope
Tailored control model
Evidence references
Bespoke evidence plan
Behavioural or scenario testing
Custom testing programme
Human assessment
Expert assessment team
Change triggers and reassessment
Contract-defined
Ongoing assurance
Tailored
Status and revocation
Contract-defined
GRC and compliance mapping
Tailored mapping
Public record or ledger anchoring
Configurable
What it does not establish
Not a numbered level; comparability requires explicit mapping
02 / The Sovrentic offer
A clear path to assurance
01
Declaration
Start with a structured, self-declared record.
Self-declared · not assessed
02
Assessment
Evaluate a defined agent, tool or skill against a published scope.
Paid assessment
03
Ongoing assurance
Revisit evidence and status when the resource changes.
Recurring service
04
Custom programme
Design a scoped programme for complex portfolios and institutional needs.
Bespoke engagement
Choose the level of assurance that matches the resource, the decision and the operating context.
Optional public anchoring may support integrity, provenance and transparent record history. It does not by itself prove behavioural truth, runtime identity or current non-revocation. Evidence remains off-chain and status must be checked against an authoritative source.
Shape the first profiles
Start with a scoped record. Decide what assurance you actually need.
The SOV offers are in development. No assurance record is issued today, and no accreditation, regulatory endorsement or payment flow is implied.