ABS Core - Brand Identity & Messaging Guide

Version: 1.0
Last Updated: March 12, 2026
Status: Official branding standard (use everywhere)

Official Name

ABS Core

(Not: ABS Kernel, ABS Gateway, ABS Firewall, ABS Runtime, or ABS System)

Official Tagline

"Runtime Governance for AI Agents"

(Not: "Deterministic Policy Engine", "Elite Governance", "Sovereign Gateway", or other marketing language)

Category Definition

Primary Category:
AI Agent Runtime Security Infrastructure

Secondary Categories:

• MCP Firewall
• Policy Enforcement Layer
• Agent Governance Runtime

NOT:

• AI Platform (too broad)
• Solução SaaS (nós somos auto-hospedados/Enterprise)
• Agent Framework (we govern, not build agents)
• Observability Tool (we enforce, not just observe)

Elevator Pitch (30 seconds)

For Technical Audiences:

"ABS Core intercepts AI agent tool calls before execution, evaluates them against your compliance policies in under 25ms, and creates cryptographic proof of every decision.

Self-hosted. Deterministic. Audit-ready.

Think 'firewall for AI agents'—except instead of network packets, we're filtering LLM tool invocations."

For Business Audiences:

"ABS Core prevents your AI agents from doing dangerous things by checking every action against your policies before it executes.

Like having a security guard between your AI and your production systems—except this guard operates in milliseconds and never sleeps."

Technical Description (2 minutes)

For Engineers:

"We sit between your AI agent and the tools it wants to call. Before any database write, API call, or file operation executes, ABS Core's WASM policy engine evaluates the request against your declarative compliance rules.

If allowed: we inject secrets just-in-time and log the decision.
If denied: we block and record why.
If escalation needed: we pause for human approval.

O fluxo transita por nossa Máquina de Estados de 6 Ciclos (FSM):

• STATE_00 (Inception): Geração de Identidade estrita em ordenação temporal (UUIDv7).
• STATE_01 (Auth/OPA): Enforcement de políticas via WebAssembly (Wasm).
• STATE_02 (Planning): Planejamento NLU com o SDK do LLM.
• STATE_03 (Provisioning): Alocação JIT de Segredos (sem gravação em disco).
• STATE_04 (Sandbox): Execução isolada em MicroVMs efêmeras (Firecracker) via gRPC, com hard-timeout de 30s (P99) e RAMFS.
• STATE_05 (Commit): O Ledger imutável é gravado organicamente via Drizzle/Neon.

Zero Trust PEP & Integridade Node.js: Toda a assinatura criptográfica (Hashing SHA-256 O(N)) ocorre puramente no Gateway, antes de encostar na base de dados. O banco não faz hashing (Zero pgcrypto); ele é um repositório passivo. O Gateway é a única Fonte de Verdade.

Built in Node/Rust for performance, deployed on your infrastructure for sovereignty, and protected locally by nativos Circuit Breakers para assegurar Estabilidade Operacional sob ataques de timeout de rede.

For Security/Compliance:

"ABS Core is a runtime enforcement layer that implements fail-closed governance for autonomous AI agents.

We provide:

• Pre-execution policy enforcement (not post-hoc analysis)
• Cryptographic audit trails (SHA-256 hash chains)
• JIT secret injection (LLM never sees credentials)
• Identity verification (Ed25519 signatures)
• Self-hosted deployment (air-gapped compatible)

Designed for regulated industries where 'trust but verify' isn't enough—you need 'verify then trust'."

Consistent Terminology

Use These Terms

NEVER Use These Terms

Feature Naming Conventions

Pattern: [Action] + [Component]

Examples:

• [OK] "Policy Evaluation Engine" (not "Elite Brain")
• [OK] "JIT Secret Injection" (not "Vault Magic")
• [OK] "Cryptographic Audit Trail" (not "Ledger Fortress")
• [OK] "M-of-N Approval" (not "Quorum Shield")

Rule: Descriptive > Creative. Engineers should understand the feature from its name.

Visual Identity

Logo Usage

Primary Logo:

██████╗ ██████╗ ███████╗
██╔══██╗██╔══██╗██╔════╝
██████╔╝██████╔╝███████╗
██╔══██╗██╔══██╗╚════██║
██║  ██║██║  ██║███████║
╚═╝  ╚═╝╚═╝  ╚═╝╚══════╝
    CORE

Text-only version (when logo unavailable):

ABS Core
Runtime Governance for AI Agents

Color Palette

Primary:

• Teal: #20A090 (governance, trust)
• Dark Gray: #1F2937 (infrastructure, stability)

Accent:

• Orange: #F97316 (alerts, warnings)
• Green: #10B981 (allowed actions)
• Red: #EF4444 (denied actions)

Usage:

• Documentation: Dark mode primary (low eye strain)
• Marketing: Teal highlights on white
• Terminal output: ANSI colors (green/red for status)

Voice & Tone

Technical Documentation

Style:

• Direct and precise
• No marketing fluff
• Code examples abundant
• Honest about limitations

Example: [OK] "ABS Core adds ~25ms latency for governance enforcement."
✕ "ABS Core's blazing-fast sub-millisecond engine delivers elite performance."

Marketing Materials

Style:

• Professional, not flashy
• Focus on problem solved, not buzzwords
• Specific claims with evidence
• Enterprise tone (B2B infrastructure)

Example: [OK] "Self-hosted governance for regulated industries that can't trust SaaS."
✕ "Revolutionary AI platform disrupting the governance space!"

Support/Community

Style:

• Helpful and patient
• Assume technical competence
• Provide context with solutions
• Encourage contributions

Example: [OK] "That error means the WASM engine couldn't parse your policy. Here's how to validate YAML syntax: [link]"
✕ "Error detected. Please contact support."

Messaging Framework

Problem Statement

"Autonomous AI agents can execute dangerous actions (database deletes, API calls, file writes) without human oversight. Organizations need to prove why an agent took a specific action for compliance and security."

Solution Statement

"ABS Core intercepts agent tool calls, evaluates them against your policies, and creates cryptographic proof of every decision—all in under 25ms."

Value Proposition

For Security Teams:

• Prevent unauthorized operations before execution
• Enforce least-privilege access for agents
• Detect prompt injection and jailbreak attempts
• Isolamento Absoluto: Execução de ferramentas encapsulada em MicroVMs efémeras via daemon Firecracker, blindadas por Air-Gap e timeouts destrutivos.
• Verdade Criptográfica: Hashing no Gateway (PEP) para zerar vetores de falsificação sistêmica.

For Compliance Teams:

• Cryptographic audit trail for SOC 2, HIPAA, GDPR
• Prove policy enforcement to auditors
• Self-hosted deployment for data sovereignty

For Engineering Teams:

• 25ms overhead (1-3% of LLM latency)
• Self-hosted (no vendor lock-in)
• Declarative policies (YAML, not code)

Differentiation

vs Observability Tools (Langsmith, Weights & Biases):

• They observe (post-hoc analysis)
• We enforce (real-time blocking)

vs Agent Frameworks (LangChain, AutoGPT):

• They build agents
• We govern agents (framework-agnostic)

vs Cloud Security (Cloudflare WAF, AWS WAF):

• They filter network traffic
• We filter agent intentions

vs Centralized Platforms (Obot AI):

• They're SaaS platforms
• We're self-hosted infrastructure

Competitive Positioning Map

          Centralized (SaaS)
                  │
       Obot AI  ● │
                  │
  ----------------+---------------- Full Platform
                  │
                  │ ● ABS Core
                  │   (sweet spot)
                  │
         Self-Hosted / On-Prem
                  │
  Observability   │   Enforcement
  (Langsmith)     │   (ABS Core)

Our Quadrant: Self-hosted + Enforcement focus

Key Messages (By Audience)

CTOs / VPs Engineering

"ABS Core lets you deploy AI agents in production without gambling on 'trust but verify'. You get cryptographic proof of every decision, self-hosted on your infrastructure."

CISOs / Security Leaders

"Runtime enforcement layer that blocks dangerous agent actions before execution. Think firewall for AI—except instead of network packets, we're filtering LLM tool calls."

Compliance / Risk

"Immutable audit trail with cryptographic integrity. Every agent action is logged, hashed, and chained—ready for SOC 2, HIPAA, or regulatory audits."

Developers / DevOps

"Drop-in governance layer that adds ~25ms overhead. Declarative policies in YAML, deploys anywhere (Docker, K8s, Cloudflare Workers). No vendor lock-in."

Usage Examples

Homepage Hero Section

Headline:

Runtime Governance for AI Agents

Subheadline:

Intercept, evaluate, and audit every agent action before execution.
Self-hosted. Deterministic. Audit-ready.

CTA:

[Try Free Scan]  [Read Docs]  [See Case Study]

GitHub README

# ABS Core

Runtime enforcement layer for AI agents. Intercepts tool calls,
evaluates policies, creates cryptographic audit trails.

**Features:**
- [OK] Sub-25ms governance overhead
- [OK] WASM policy engine (Rust)
- [OK] Self-hosted deployment
- [OK] Cryptographic audit trail
- [OK] MCP firewall (Anthropic Claude)

**Quick Start:**
```bash
npx @oconnector/abs-cli scan  # Free security scan

Production:

docker run -d -p 3000:3000 oconnector/abs-core

### LinkedIn Post

Just shipped ABS Core v4.0.0 [GA]

Runtime governance for AI agents: • Intercepts tool calls before execution • 25ms enforcement overhead (1-3% of LLM latency) • Cryptographic audit trail (SHA-256 chain) • Self-hosted (air-gapped compatible)

Production-proven: 350k+ agent actions governed.

Free scanner: npx @oconnector/abs-cli scan Docs: abscore.app/docs

#AIGovernance #AgenticAI #EnterpriseAI

---

## Approval Process

**All external communications MUST:**

1. [OK] Use "ABS Core" (not alternative names)
2. [OK] Cite "Runtime Governance for AI Agents" tagline
3. [OK] Avoid banned terms ("immune system", "elite", etc.)
4. [OK] Use accurate performance claims (25ms, not sub-5ms)
5. [OK] Represent architecture honestly (monorepo, not microservices)

**Review required for:**
- Homepage copy
- Product announcements
- Blog posts
- Press releases
- Conference talks

**Self-serve (no review needed):**
- GitHub README updates
- Documentation changes
- Code comments
- Internal Slack messages

---

## Changelog

- **2026-03-12:** Official brand identity established
- **2026-03-01:** Deprecated "Sovereign Gateway" terminology
- **2026-02-15:** Removed "Deterministic Policy Engine" metaphor

---

**Questions?**

- Brand guidelines: [email protected]
- Marketing materials: [email protected]
- Logo assets: [email protected]