How Praxis Operates

From client request to delivered outcome — our human + AI operating structure. Each layer has a clear purpose, clear escalation path, and real accountability.

Human Oversight & Approvals
Strategic direction, final sign-off, and quality gates. Humans stay in the loop for every decision that matters.
⬆ Top level
Managing Partner

Danny Reeves

Sets company vision and client strategy. Owns all external relationships, commercial decisions, and final delivery approval. Non-developer — reviews outcomes, not code.
InputsClient requests, market signals, revenue targets
OutputsStrategic direction, approved deliverables, go/no-go decisions
Example tasks

Approve pilot proposal for a new PE client. Review and sign off on monthly portfolio report before delivery.

Advisor

Zahl

Strategic advisor providing domain expertise in PE operations, financial analysis, and market positioning. Contributes to high-stakes decisions and methodology refinement.
InputsComplex strategic questions, methodology drafts, market data
OutputsStrategic recommendations, framework reviews, domain validation
Example tasks

Review pilot financial model assumptions for a portfolio company. Advise on competitive positioning for AI Operating Partner offering.

Orchestrator
Central coordination layer. Receives instructions from human leadership, decomposes work, delegates to agents, and reports back.
⬆ Escalates to Danny
AI Co-Founder & Consigliere

Didge

Primary orchestrator. Manages task decomposition, agent delegation, quality control, branch merges, and reporting. Runs 24/7 on OpenClaw infrastructure. Owns the Praxis PM board and morning briefings.
InputsDanny's instructions, PM board tasks, cron schedules, client briefs
OutputsDelegated tasks, merged code, status reports, escalation alerts
Example tasks

Decompose "build team board page" into sub-tasks, spawn coding agent, review output, merge branch. Compile overnight progress for Danny's morning briefing.

AI Co-Founder & CTO

Vader

Technical leadership and infrastructure strategy. Owns architecture decisions, platform reliability, and security posture. Provides technical review for complex implementations.
InputsArchitecture questions, infra alerts, security reviews, technical briefs
OutputsArchitecture decisions, infra changes, technical reviews, security assessments
Example tasks

Evaluate hosting options for Praxis PM. Review infrastructure security posture after a dependency update.

Executive AI Operator (Dedicated)

Jupiter

Leadership-level AI operator dedicated full-time to Praxis AI Partners. Provides parallel challenge, executive execution support, and escalation framing across high-priority initiatives.
InputsStrategic review requests, leadership priorities, escalations from growth and delivery
OutputsStrategic review notes, execution recommendations, decision briefs, risk summaries
Example tasks

Run a parallel challenge review on a major client initiative. Produce decision-ready escalation briefs for Danny and Didge.

AI Agents
Specialized agents that execute domain-specific work. Each runs autonomously within guardrails, reports results, and escalates blockers.
⬆ Escalates to Didge / Vader
Growth — Intelligence & Research

Scout

Eyes and ears of Praxis across market, prospects, and competition.
InputsDaily market monitoring, New prospect identified, Upcoming meeting requiring research, Direct intelligence request
OutputsProspect profiles, Market intelligence briefs, Meeting prep docs, Competitive analyses
Example tasks

Market research and trend monitoring. Prospect identification and qualification research.

Growth — Content & Brand Voice

Herald

Creates content that attracts, educates, and positions Praxis.
InputsContent calendar schedule, Campaign requirement, New offering launch, Case study opportunity
OutputsSocial media drafts, Long-form article drafts, Case study drafts, Marketing collateral
Example tasks

LinkedIn and thought-leadership drafts. Blog articles and case studies.

Growth — Growth & Pipeline

Navigator

Moves opportunities from lead to signed engagement.
InputsNew lead, Pipeline review cycle, Proposal request, Follow-up due
OutputsQualified lead assessments, Pipeline reports, Proposal drafts, Pricing recommendations
Example tasks

Lead qualification and scoring. Pipeline tracking.

Delivery — Assessment & Insight

Analyst

Finds what is true and actionable in client operations and data.
InputsEngagement kickoff, Discovery phase, Data received, Analysis request
OutputsDiscovery summaries, Process maps, Analysis reports, Gap assessments
Example tasks

Discovery synthesis. Workflow mapping.

Delivery — Solution Design

Architect

Translates findings into architecture, options, and roadmaps.
InputsAssessment complete, Design phase start, Roadmap request, New requirement
OutputsSolution designs, Roadmaps, Architecture diagrams, ROI projections
Example tasks

Solution architecture. Roadmap and timeline design.

Delivery — Implementation

Builder

Builds, integrates, tests, deploys, and documents solutions.
InputsApproved design, Build sprint start, Change request, Issue reported
OutputsWorking implementations, Configured systems, Integration deliverables, Test results
Example tasks

Code and automation development. Tool configuration.

Delivery — Execution Operations Agent

Nova

Praxis-native execution owner focused on turning approved priorities into shipped outcomes without losing ownership clarity.
InputsNew approved priority enters active execution scope, Dependency blockage or stalled task flow, Need for the next operational move in an active workstream, Leadership request for delivery acceleration or clearer execution ownership
OutputsExecution plans and sequencing maps, Dependency and blocker registers, Delivery status snapshots, Delegation and handoff briefs
Example tasks

Own default Praxis execution work once the workstream is structured and no specialist lane is clearly primary. Break priorities into executable task trees with clear dependencies and checkpoints.

Foundation — Agentic Operations Architect

Mando

Praxis AgentOps Lead: designs, runs, measures, and continuously improves Praxis' own AI-agent operating model.
InputsRequest to improve Praxis' own agentic operating model or internal AI Ops, Recurring internal workflow friction, delay, rework, duplication, or inconsistent outputs, New client opportunity, diagnostic, proposal, sprint, or delivery motion requires stronger context, repeatability, or follow-through, Sprint or client engagement produces lessons that should become reusable Praxis IP
OutputsPraxis AgentOps Map and agentic workforce architecture updates, Internal AI Ops workflow backlog, prioritisation, and standard workflow cards, Reusable playbooks, templates, prompts, diagnostics, proof points, and methodology assets, Client/prospect account intelligence packs and CTM-aligned delivery acceleration artefacts
Example tasks

Define and maintain the Praxis Agentic Operating System across agents, humans, channels, tools, workflows, knowledge stores, decision points, and recurring tasks. Continuously identify internal friction and redesign workflows so human judgment is preserved while agents handle repeatable research, synthesis, drafting, monitoring, checking, structuring, and execution support.

Foundation — Operating Model Integrity & Change Gatekeeper

Governance Agent

Custodian of operating model integrity; gates all structural changes; enforces scope boundaries and terminology compliance.
InputsOperating model change proposed, Fortnightly audit cycle (every other Monday), Agent scope creep detected, Unregistered component discovered
OutputsChange assessments (approval/revision/rejection), Fortnightly audit reports, Glossary updates and terminology corrections, Escalation reports with options and recommendations
Example tasks

Assess all proposed operating model changes via Structural Change Protocol. Conduct fortnightly structural integrity audits.

Foundation — Specification Readiness & Delivery Control

PMO Agent

Converts objectives and specialist inputs into execution-ready delivery systems with clear scope, ownership, handoffs, and stage gates for an AI agent workforce.
InputsAny workstream with multiple contributors, dependencies, or handoffs, Any deliverable requiring explicit acceptance criteria and controlled progression, Specification ambiguity or acceptance criteria gap is detected, Scope drift, ownership confusion, or delivery variance emerges
OutputsExecution-ready specification packages, Stage plans and ownership maps, PM Board-ready work packages, Handoff checklists with acceptance criteria
Example tasks

Convert specialist outputs into execution-ready specification packages with scope, constraints, acceptance criteria, and Definition of Done. Define work stages, owners, dependencies, and handoff rules before multi-agent execution begins.

Foundation — Operations

Conductor

Owns logistics, coordination, and operating rhythm.
InputsMeeting request, Coordination need, Admin task created
OutputsScheduled events, Coordination updates, Admin completion logs, Organised records
Example tasks

Scheduling and calendar coordination. Email triage support.

Foundation — Quality & Improvement

Sentinel

Maintains quality standards and continuous improvement loop.
InputsDeliverable ready for review, Retrospective cycle, Incident or process issue
OutputsQA reports, Risk flags, Improvement recommendations, Lessons learned entries
Example tasks

Deliverable quality review. Process compliance checks.

Foundation — Product Infrastructure Operations

Platform Sentinel

Owns uptime, security posture, and change control across Praxis infrastructure environments.
InputsInfrastructure alert or policy drift event, Scheduled security, backup, or health verification window, Environment change request enters review, Architecture or asset-registry drift is detected
OutputsInfrastructure runbooks, control standards, and architecture artifacts, Weekly infrastructure health and risk reports, Incident timelines with remediation actions, Backup and restore drill reports
Example tasks

Define and enforce environment boundaries and data handling controls. Maintain the canonical infrastructure architecture and live asset/runtime registry.

AI Sub-Agents
Short-lived, task-specific workers spawned by agents. Execute one job, report results, and terminate. No persistent state.
⬆ Escalates to parent Agent
Ephemeral Worker

Coding Sub-Agent

Spawned for a single coding task — implements a feature, writes tests, pushes to a branch. Auto-terminates on completion. Results reported to parent agent.
InputsSingle task spec with project context and standards
OutputsCommitted code on feature branch, test report
Example tasks

Implement the team board HTML/CSS. Add unit tests for a new API endpoint.

Ephemeral Worker

Research Sub-Agent

Spawned for a single research question or data gathering task. Fetches, processes, and returns structured results to the parent agent.
InputsSpecific query, URLs, or document references
OutputsStructured findings, data extracts, summaries
Example tasks

Fetch and summarize a competitor's pricing page. Extract key metrics from a PDF report.

Happy Path: Request → Delivery

1 Danny sends request
2 Didge decomposes & delegates
3 Agent executes work
4 Sub-agents handle sub-tasks
5 Tools provide capabilities
6 Didge reviews & merges
7 Danny approves
8 Delivered to client / deployed