Every business unit at Greater Goods gets its own team of AI agents that work alongside humans — handling operations, building institutional knowledge, and coordinating across departments. Praesidium2 is the platform that makes this possible: the command center that gives every team AI superpowers.
The Problem
Teams Are Stretched Thin
Every department juggles too many operational tasks with too few people. High-value work competes with routine reporting, status updates, and coordination overhead.
Knowledge Walks Out the Door
When someone leaves, their expertise goes with them. Processes, vendor relationships, institutional context — all lost. New hires start from scratch every time.
Coordination Is Expensive
Cross-department collaboration requires endless meetings, Slack threads, and email chains. Getting alignment on a simple status update can take days.
Manual Work Consumes Hours
Report generation, data entry, status tracking, PPC bid management — repetitive tasks that are necessary but take time away from strategic thinking.
The Solution
AI Teams for Every Business Unit
Each department gets its own team of AI agents — purpose-built for that unit's work. Marketing agents assist with campaign optimization. Product agents help track supply chain status. Tech agents support standup preparation and code drafting.
You Lead, Agents Support
Agents take care of routine operational tasks — generating reports, posting status updates, pulling data, and updating Jira tickets — so your team has more time for decisions, strategy, and creative problem-solving.
Agents Learn Over Time
Every agent builds memory and skills as it works. Lessons learned, preferences, procedures — all preserved. Institutional knowledge is preserved and shared — teams build on what came before, and onboarding new members is faster.
Custom Tools, Built Faster V1.5
Need a tool that pulls HubSpot metrics? Describe it to a forge agent. It writes the code, tests it, and publishes it — freeing up engineering to focus on the complex, high-value work only they can do.
How It Works
Three layers working together — your team talks to agents, agents talk to your systems.
BROWSER DASHBOARD
Projects, Chat, Standups, Approvals, Reports
AGENT TEAMS (Anthropic SDK + LangGraph)
34 agents across 5 teams — memory, skills, tools, coordination
DATA & INTEGRATIONS
V1: Slack, Jira, GitHub — V1.1: +Confluence, Zoho, Google — 45+ planned
Business Units
Tech Team
James Maes, CTO
Engineering, product development, DevOps, and architecture
Build Team: James Maes (CTO) + 2 engineers + unlimited AI agents • see roster below
8 LOTR AgentsMarketing
Bryan Chupp, CMO
Amazon PPC, content creation, social media, and analytics
6 WWE AgentsCorp Claw
Chris Chupp, CEO
Executive coordination, finance, HR, and cross-department operations
6 Star Wars AgentsProduct Dev
Matt Carpenter, VP
Manufacturing, supply chain, vendor management, and quality
7 Avengers AgentsMe Health
Bryan Potter, VP
RPM health business, patient operations, and compliance
7 Grey's AgentsEngineering Leadership Team
The Praesidium2 build team is James Maes (CTO) + 2 engineers + unlimited Claude Code agents. The full Engineering Leadership Team shown here provides oversight, review, and domain expertise.
Key Numbers
Business Infrastructure
HITL Approval Workflow
Human-in-the-loop orchestration via LangGraph checkpointing with tamper-proof approval hashing. By default, every Jira ticket, Slack message, and report generated by an agent requires explicit human approval before reaching production systems — but approval thresholds are operator-tunable. As confidence in agent behavior grows, operators can selectively relax or remove approval gates on a per-integration basis, up to full autonomous control.
Success Metrics & KPIs
Target: measurable improvements in team productivity, execution velocity, and cross-team alignment across all 5 business teams. Tracked via time-saved metrics, task throughput, agent utilization reporting, and qualitative feedback from team leads.
Human Feedback Loop
Thumbs up/down feedback on every agent output. Feedback is stored, analyzed, and used to improve agent behavior over time. Teams see quality trend dashboards.
Phased Rollout
Engineering deploys first (Phases 1-6), Marketing joins at Phase 7, remaining 3 teams added one per phase after stabilization. Me Health deferred until HIPAA controls are validated.
Rollback & Recovery
Undo any agent action, emergency kill switch for runaway agents, and GitOps-based service rollback. Every action is logged for full audit trail and reversal capability.
Datadog Observability
Distributed tracing across all 5 services, custom agentic metrics (task latency, LLM token spend, approval throughput), and HIPAA-compliant log scanning. Full-stack visibility from browser click to agent action.
Business-Value Kill Criteria
30-day and 60-day gates measure real business impact. If agents aren't saving time or delivering value, deployment pauses for root-cause analysis before proceeding.
V1 Scope & Roadmap
V1 Launch
3 integrations (Slack, Jira, GitHub), Agent Hydration Protocol, HITL orchestration, blind secret pipeline, Semantic Search API, Datadog observability, chat, dashboards, and the full qqq data layer. Engineering team live.
V1.1 (Post-Launch)
3 additional integrations: Confluence, Zoho, and Google Workspace. All 5 business teams onboarded and operational.
V1.5 (Post-Launch)
Tool Forge (agent-built custom tools), Memory/RAG (long-term knowledge), and Meeting Orchestration (automated standups and facilitation).
Timeline
KPI Baseline
KPI baseline capture, stakeholder sign-off, environment provisioning
Core Platform
qqq data layer (92 tables), auth, gateway, Tool Runner, CI/CD pipeline, dashboard UI
Agent Runtime
Agent Hydration Protocol, Anthropic SDK + LangGraph (scoped) orchestration, HITL approval workflow, blind secret pipeline
Integrations + Search
Slack, Jira, GitHub integrations, Semantic Search API (pgvector), human feedback loop
Testing + Engineering Deploy
End-to-end testing, security hardening, Datadog observability, AWS deploy, Engineering team goes live
Stabilization + Marketing
Production stabilization, metrics validation, Marketing team onboarded as second business unit
Expand + Enhance
Remaining 3 teams added progressively. V1.1 (3 more integrations) post-launch. V1.5 (Tool Forge, Memory/RAG) post-launch
7-phase build plan: Pre-Build + 6 build phases + 1 stabilization phase. 3-person build team (CTO + 2 engineers) augmented by unlimited Claude Code agents. Existing AWS EKS + Faber CI/CD. Re-sequenced build: Tool Runner (Hands) before Agent Workers (Brain).
What Changes for Leaders
"Your daily standup runs automatically — you get a summary in Slack."
Agents facilitate meetings, capture action items, and post summaries to the channels you already use. You start your day with context instead of a calendar block.
"Your agents handle Jira and reports — you review, approve, and give feedback."
Agents create tickets, generate reports, and track status. Every output goes through human approval during rollout, and your thumbs up/down feedback trains agents to get better over time.
"Need a custom tool? Describe it and an agent builds it for you."
The Tool Forge (V1.5) lets you work one-on-one with an AI to build exactly the tool you need. Simple tools ship immediately; complex ones still get the engineering rigor they deserve. Everyone moves faster.