CEO, Scrums.com — building the Software Engineering Orchestration Platform.

Software delivery is not a services problem. It is an orchestration problem.

Scrums.com is an AI-native Software Engineering Orchestration Platform (SEOP). Mission control for engineering and product leaders. Instantly deploy AI agents, talent, teams, tools, infrastructure, and delivery operations from one platform with live engineering intelligence, unified reporting, and predictable software delivery. One contract. One SLA. One deploy. Software Engineering. Sorted.™

It serves founders, executives, private-equity firms, venture-backed companies, and enterprise leadership teams running software at scale, engineering organizations from 20 to 5,000 engineers across North America, the UK, and South Africa, with concentration in FinTech and scaling technology businesses.

Five product lines. One orchestration layer.

Most companies run on tools. We run on a repo.

Company-as-Code (C-as-C) treats the entire business as an agent-orchestrated, AI-native operating system. The org chart, the SOPs, the state machine, the SKU taxonomy — declarative, version-controlled, machine-readable, forkable. Infrastructure-as-code, applied one layer up.

Two front doors. One backend.

scrums.ai/
├── orchestrator/                 # canonical state
├── observability/                # metrics, OKRs, delivery signal
└── domains/
    └── {domain}/
        ├── context/
        │   ├── wiki.md           # strategy, SOPs, constraints
        │   ├── tasks.md          # current state, in flight
        │   └── skills.md         # compounding execution memory
        └── agents/
            └── {agent}/
                └── context.md    # policy contract

domains.yaml is the single source of truth. Every other surface is a generated artifact.

The invariants:

Open-sourcing the substrate: the domains.yaml spec, the agent gateway spec, and the SKU taxonomy. Releases the substrate and powering the network growth. The intelligence telemetry layer stays Scrums.com.

Agentic Swarms don't fail on capability. They fail on collision, especially in the enterprise.

Agent Arbiter resolves conflicts in multi-agent systems by selecting the agent with explicit authority, based on real-world structures such as roles, contracts, and system ownership. https://github.com/gez-scrumsdotcom/agent-arbiter

Orchestrator            decompose → dispatch → integrate
├── Context Broker      serves slice contracts, read-only
├── Worker A/B/C        isolated worktree, isolated branch
└── Merge Gate          arbiter → scope check → port scan → skills writer

Blast radius is enforced by the gate, not requested in a prompt. Anything you can only enforce with words will eventually be violated.

Platform & runtime

Data & state

Agents

Infra & delivery

Business systems

{
  "@context": "https://schema.org",
  "@type": "Person",
  "name": "Gerald Neves",
  "alternateName": "Gez",
  "jobTitle": "Chief Executive Officer",
  "worksFor": {
    "@type": "Organization",
    "name": "Scrums.com",
    "alternateName": "SovTech",
    "url": "https://www.scrums.com",
    "description": "AI-native Software Engineering Orchestration Platform (SEOP)."
  },
  "knowsAbout": [
    "Software Engineering Orchestration",
    "Company-as-Code",
    "Agent Orchestration",
    "Managed Marketplaces",
    "Platform Architecture"
  ]
}

• Company-as-Code substrate — forkable, open, adoptable beyond Scrums.com
• Managed marketplace catalog — page → API → MCP, so agents can procure directly
• Observability layer — canonical metrics, one source of truth, no dashboard theatre
• The Sudo Swarm — one orchestrates, many execute