BriefingClaw

Multi-Agent Architecture

Six specialized AI agents orchestrate executive briefing preparation — frontier models for research, local models for private data.

Frontier Model (Claude / GPT / Gemini)
Local Model (Granite 8B via Podman AI Lab)
Trigger
💬
Briefing Request
"Prepare a briefing package for [executive], [title] at [company]. SAB status, sponsor, and context."
8 demo scenarios available: Sarah Chen · David Park · Rachel Morrison · Pepper Minton · Ziggy Stardust-Chen · Luna Wavelength · Max Bandwidth · Sage Cloudberry
Phase 1 — Decomposition
☁️ Frontier
🎯
Oprah-tor
Orchestrator
Decomposes the request into parallel tasks. Dispatches to specialist agents. Validates results and synthesizes the final briefing package. "You get a task! YOU get a task!"
OpenClaw Gateway · Podman Container
Phase 2 — Intelligence Gathering (parallel)
☁️ Frontier
🔍
Sherlock Ohms
Executive Research
Deep person-level intelligence. Career arc, recent news, conference appearances, communication style, technology priorities.
ZeroClaw CLI · Native Binary · 3.4 MB
☁️ Frontier
🏢
Bloom-borg
Account Intelligence
Company-level analysis. Revenue, tech stack, competitive dynamics, job posting signals, strategic priorities.
ZeroClaw CLI · Native Binary · 3.4 MB
💻 Local
📋
Déjà View
SAB Historian
Cross-program memory. SAB themes, past briefings, engagement stage, executive sponsor history, open follow-ups.
OpenClaw Skill · Podman Container
Phase 3 — Assembly & Protocol (parallel)
💻 Local
📄
Draft Punk
Briefing Architect
Synthesizes all intelligence into polished deliverables: executive dossier, backgrounder, sponsor talking points, recommended agenda.
OpenClaw Skill · Podman Container
💻 Local
⭐
Alfred Bitworth
VVIP Protocol
Checks VVIP tier, generates protocol checklist, drafts sponsor alert email, flags scheduling conflicts, logs the engagement.
OpenClaw Skill · Podman Container
💻 Local
🤝
Sponsor Coach
Readiness Assessment
Lightweight micro-agent. Checks whether the assigned sponsor is prepared (recent engagement, topic familiarity, expertise match). Returns GREEN/AMBER/RED plus one action.
OpenClaw Skill · Podman Container
Phase 4 — Success Prediction
💻 Local
🎲
The Oddsfather
Briefing Success Predictor
Takes the complete intelligence bundle and scores it across 5 dimensions: relationship foundation, pipeline stakes, risk signals, briefing readiness, outcome alignment. Outputs success probability plus the three actions that move the odds. "I'm gonna make him a prediction he can't refuse."
OpenClaw Skill · Podman Container
Output — Complete Briefing Package
📦 Deliverables (assembled in ~90 seconds)
Executive Dossier
Briefing Backgrounder
Sponsor Talking Points
Recommended Agenda
Conversation Starters
VVIP Protocol Checklist
Sponsor Alert Email
Engagement Log Entry
🖥️ Infrastructure Stack
Runtime Podman Desktop + Podman AI Lab
Model IBM Granite 8B (local, air-gapped)
Model Claude / GPT / Gemini (cloud API)
Agents OpenClaw (container) + ZeroClaw (native)
Scale Red Hat OpenShift AI (enterprise)
🔐 Model Routing Philosophy
Cloud Research agents — public web data, complex reasoning
Local Data agents — private SAB/CRM/VVIP data stays on device
Power where you need it. Privacy where it matters.
Built with open source · Runs on your laptop · Red Hat CTO Office · GACEP Spring 2026