ABS Core: Runtime Governance for AI Agents
ABS Core: Runtime Governance for AI Agents
March 2026 — Enterprise Overview
1. Executive Summary
ABS Core is a runtime governance layer for AI agents. We sit between the agent and the systems it wants to access — APIs, databases, files, internal tools — and evaluate every action against defined policies before execution.
In simple terms: ABS Core helps enterprises place AI agents in production without giving up control.
2. What the product is
ABS Core is a self-hosted enforcement layer that:
- Intercepts agent tool calls before execution
- Evaluates against deterministic policies (ALLOW / DENY / ESCALATE)
- Records every decision in a cryptographically signed audit trail
- Escalates sensitive operations to human approvers (M-of-N quorum)
3. Architecture: The Octagonon
ABS Core follows a modular governance architecture with 8 pillars:
- Identity — Sovereign verification of agent identity and model allowlisting
- Policy — Deterministic evaluation via OPA WASM engine
- Execution — Controlled tool call interception
- Audit — Immutable ledger with SHA-256 hash chain
- Approval — Human-in-the-loop M-of-N quorum protocol
- Observability — Operational telemetry and monitoring
- Intelligence — Governance context and anomaly detection
- Forensics — Deep analysis of agent behavior and code
4. Performance
Performance must be discussed accurately:
- WASM engine: ~1.2ms policy evaluation (p50)
- Full governance loop: ~23ms median, ~38ms p95
- LLM call baseline: ~1,200ms (GPT-4)
- ABS overhead: 1-3% of total request time
Enterprise buyers should evaluate performance on real governed workflows, not only isolated benchmarks.
5. Deployment Model
ABS Core is commercialized as enterprise software:
- Deployment: Customer-controlled infrastructure (VPC / self-hosted)
- License: Annual subscription with implementation and support
- Support: SLA contracts for critical operations
This creates a clear separation between recurring software revenue and professional services.
6. Market Traction
- Traction: Active deployments in production (O-Bot and OConnector)
- Sustainability: Consistent revenue with zero churn over 3+ months
- 350,000++ requests/day through ABS Core (O-Bot)
- 350,000+ total governed requests (O-Bot, 12 months)
7. Commercial Use Cases
The strongest enterprise use cases are narrow and high-value:
- Governed internal automations
- Infrastructure or operational write paths
- Sensitive data handling workflows
- Agent systems requiring strong auditability
8. Commercial Engagement Model
The recommended enterprise motion is:
- Technical due diligence
- Narrow pilot on a governed workflow
- Evaluation of deployment topology and integration depth
- Annual licensing or strategic integration
The goal is to prove runtime value in a specific execution path before expanding scope.
9. Investment Opportunity
Strategic Goal: Secure funding for institutional scale and market expansion.
Use of funds:
- Convert existing pilots to annual contracts
- Close new enterprise deals in the financial sector
- Build implementation and SLA support team
Why now: Companies are putting AI agents in production. The question is no longer "can the agent do this?" but "should the agent be allowed to do this, under which policy, and how do we prove it later?"
10. Current State
ABS Core has transitioned from concept to production-validated software:
- Gateway: Policy enforcement with OPA WASM + FSM pipeline
- Audit: SHA-256 hash chain with RSA/Ed25519 signatures
- Identity: Model allowlist blocking unauthorized LLMs
- Quorum: M-of-N approval protocol (1-5 approvers by risk tier)
- Production: 12 months with O-Bot at 12k requests/day
Contact: [email protected] / [email protected]