Every morning at 6:00 AM Pacific, before I've poured my first coffee, my AI agent has already reviewed every open pull request across 8 client repositories, summarized overnight Slack conversations by client, checked uptime on every production site, audited 7,247 Contentful entries for stale drafts, and queued a prioritized briefing in my DMs. By the time I sit down, I have the situational awareness that used to require a project manager, a DevOps engineer, and a very caffeinated Slack monitor.

This is the Command Center — the operational dashboard I built with Alpha Agent to run Last Rev's day-to-day operations. It's not a concept deck or a demo. It's the actual system I use every day to manage a digital agency with enterprise clients, a distributed engineering team, and a pipeline of inbound leads — mostly by myself.

If you're a COO, ops leader, or solo founder wondering how to scale without adding headcount, this is the playbook.

The Problem: Death by a Thousand Tabs

Running a services company means living in a dozen tools simultaneously. GitHub for code. Slack for communication. Contentful for CMS health. Google Analytics for traffic. Zoom for meetings. A CRM for leads. Uptime monitoring. Deployment dashboards. Calendar for scheduling. Each tool has its own notification system, its own dashboard, its own mental model.

The result? Context switching becomes your full-time job. You're not doing ops — you're checking things. A study by Qatalog and Cornell found that workers spend 59 minutes per day just searching for information across disparate tools.1 For an ops leader managing 8+ platforms, that number is conservative.

I needed a single pane of glass. Not a BI dashboard that required manual data entry — a living system that pulled data from every source, applied intelligence, and surfaced only what mattered.

What the Command Center Actually Is

The Command Center is a custom web application built with Web Components, backed by Supabase for real-time data, and powered by Alpha Agent's AI agent for automated analysis. It's organized into five operational tabs:

Admin — The Daily Operating Rhythm

The Admin tab is where I start every morning. It contains four modules that give me an instant pulse check:

  • Queue Summary — A real-time view of the trigger queue. Alpha Agent processes tasks every 60 seconds from a Supabase trigger_queue table. If a client sends a Slack message that needs follow-up, if a PR needs review, or if a content audit flags something — it hits the queue. I can see pending, processing, and completed items at a glance.
  • Daily Feed — An AI-generated digest of everything that happened in the last 24 hours: deploys, PR merges, Slack conversations, meeting summaries, and anomalies. Think of it as a personalized newspaper for your company.
  • Slack Summaries — Every morning at 6:30 AM PT, a cron job scans all Slack channels and generates per-client conversation summaries. Instead of scrolling through 200 messages, I read five paragraphs. If Krista mentions that a client wants to reschedule a meeting, or if a developer flags a blocker — it's in the summary, highlighted by priority.
  • Calendar — Today's meetings pulled from Zoom and Google Calendar, with AI-generated prep notes. Before my 10 AM IAS billing review, the system has already pulled the latest client health score, open PRs, and recent Slack context so I walk in informed.

Code — Engineering Visibility Without Standup Meetings

The Code tab gives me full visibility into our engineering output:

  • PR Dashboard — Every open pull request across all Last Rev repositories (last-rev-marketing-site, lively-marketing, wow-marketing, tarealty-marketing, ias-intranet, contentful-app-monorepo, lastrev-libraries, and more). Right now I can see 30 open PRs, including security bumps from Dependabot and a feature PR from our developer CT for "UIE Templates Updates for Improved Usability When Duplicating." The hourly PR review cron (running 7 AM–4 PM PT) flags any PR that's been open more than 48 hours.
  • Ideas Pipeline — An hourly cron generates product and content ideas, scored by a rubric that considers revenue impact, client relevance, and implementation effort. These aren't random brainstorms — they're grounded in actual data from our client conversations, GA4 trends, and competitive research.
  • Lighthouse Audits — Automated performance scores for every client site. When a deploy tanks Core Web Vitals, I know before the client does.
  • DRY Audit — A nightly code quality pass that checks every app in our workspace for shared component usage, duplicate code patterns, and consistency with our design system.

Client Health — The Scorecard That Replaced Weekly Check-ins

This is where the Command Center earns its keep. The Client Health tab synthesizes multiple data sources into a single health score per client:

  • Uptime Monitoring — Real-time status for every production site, pulled from our uptime monitoring app
  • Client Health Scores — Each client gets a composite score based on: commits in the last 7 days, open PRs (and how stale they are), open issues, last deploy date, and Slack message volume. A client with zero commits, 13 stale PRs, and no Slack activity in a week gets a red health score — which is exactly what our IAS account showed recently, triggering a proactive outreach before the client had to ask.
  • Contentful Health — A full audit of CMS content: 7,247 total entries, 5,983 published, 1,048 in draft, 216 changed but unpublished, and 1,255 stale entries. The system flags drafts that haven't been touched in over 1,000 days — like that HSA calculator sitting in Lively's old Contentful space since August 2021.
  • GA4 Alerts — Automated anomaly detection on Google Analytics data. When sessions drop 28% vs. the 7-day average, I get an alert. When a blog post's pageviews spike 316% (like our /blog/nextjs-tips post recently did), I know to amplify it. No more logging into GA4 and squinting at charts.
  • Community Activity — Tracking engagement across our ecosystem

Financial — Pipeline Without a CRM

The Financial tab connects to our lead management system:

  • Lead Pipeline — Every inbound lead is enriched with company data, tech stack analysis, and a "Last Rev Fit" score. When Acme Health (a digital health platform on Contentful + Next.js) came in, the system automatically scored them 9/10 and generated talking points: "Ask about their patient portal redesign timeline" and "Discuss composable architecture benefits for HIPAA compliance." That's intelligence that would take a sales team 30 minutes of research, generated in seconds.
  • Nightly Calendar Scan — Every night at 9 PM PT, a cron job scans tomorrow's calendar for meetings with external contacts, then pre-researches each company and person so I have context before every call.

Personal — The Human Touch

Even the personal tab has a purpose: a curated news feed, weather briefing at 6 AM PT, and a daily dev journal generated at 6 PM PT that captures what was accomplished, what's pending, and what needs attention tomorrow. It's the async standup I give to myself.

The 63 Cron Jobs That Run the Show

The Command Center isn't just a dashboard — it's an operating system. Behind the UI are 63 scheduled jobs that continuously feed it data, run analyses, and take action. Here's how they break down by category:

Category Jobs Examples
Nightly Reviews 35 PR/Review/Merge cycle, DRY audit, UX review, per-app code reviews for 20+ apps
System 5 Trigger queue processor (every 1 min), Command Center refresh (every 30 min), Kanban worker
Monitoring 4 GitHub PR review check (hourly), weekly PR triage & stale nudger, backlog meeting prep
Content Generation 4 Hourly idea generation, skill ideas, weekly web search discovery
Morning Briefings 4 Weather (6 AM), Slack summary by client (6:30 AM), today's meetings (7 AM), daily dev journal (6 PM)
Research 2 Lead research calendar scan, DMC research
Maintenance 1 Weekly memory hygiene — consolidates AI context files
Other 8 Recipe audit & discovery, nightly brainstorm, meme trend refresh, reminders

The most critical job is the Trigger Queue Processor, which runs every 60 seconds. It polls a Supabase table for pending tasks — anything from "summarize this Zoom recording" to "research this lead" to "deploy this app" — processes exactly one per cycle, and delivers results to Slack. It's the nervous system that connects everything.

Real-World Scenario: A Day in the Command Center

Let me walk through an actual morning — today, February 18, 2026:

  1. 6:00 AM — The nightly PR review completed at 6 AM UTC. It scanned every app directory with a git remote, checked for uncommitted changes, created PRs where needed, and ran code review. The DRY audit ran in parallel, checking all apps load shared components correctly.
  2. 6:30 AM PT — The Slack summary cron fires. It pulls conversations from the last 24 hours: Krista mentioned that Derek wants to use tomorrow's time for SEO work. A developer flagged a web component race condition in the blog listing (which Alpha Agent already fixed and pushed).
  3. 7:00 AM PT — Today's meetings cron pulls my calendar. I have a client sync at 10 AM. The system pre-loads the client's health score, recent PRs, and deployment status.
  4. 7:15 AM PT — I open the Command Center. One tab. Five sections. Everything I need. The client health panel shows IAS is red — 13 stale PRs, zero commits this week. I make a note to address this in the sync. The GA4 panel shows a traffic spike on our blog. The lead pipeline shows a new healthcare prospect scored at 9/10.
  5. Total time to full situational awareness: 4 minutes.

The Architecture: Web Components + Supabase + AI

The technical stack is deliberately simple:

  • Frontend — Vanilla Web Components (no React, no build step). Each module (cc-prs, cc-client-health, cc-ga4-alerts, cc-leads, cc-uptime, cc-slack, cc-calendar, etc.) is a self-contained custom element that fetches its own data and renders independently. 27 modules total.
  • Data Layer — Supabase for real-time tables (trigger queue, daily updates, crons) and static JSON files for snapshots (PRs, leads, client health, GA4 alerts). The sql-sync.js library keeps the frontend in sync.
  • AI Layer — Alpha Agent cron jobs write to JSON files and Supabase tables. The dashboard reads them. No complex API layer — just files and a database.
  • Hosting — Static files served at command-center.example.alphaclaw.app. No server. No SSR. Loads in under 1 second.

This architecture means the dashboard is indestructible. If a cron job fails, the dashboard shows stale data with a timestamp — it doesn't crash. If Supabase goes down, the JSON files still load. If the AI agent is offline, the last-known-good data persists.

What This Replaces

Let's be explicit about the headcount math. Before the Command Center, achieving this level of operational awareness would require:

Role What the Command Center Replaces Estimated Annual Cost
Project Manager Client health monitoring, PR triage, blocker escalation, meeting prep $85,000–$110,000
DevOps Engineer (part-time) Uptime monitoring, deployment tracking, Lighthouse audits, dependency updates $60,000–$80,000
Marketing Analyst GA4 monitoring, content audit, traffic anomaly detection, lead enrichment $65,000–$85,000
Sales Ops Coordinator Lead research, CRM maintenance, meeting prep, pipeline reporting $55,000–$70,000
Executive Assistant Calendar management, Slack monitoring, daily briefings, follow-up tracking $50,000–$65,000
Total replaced capacity $315,000–$410,000/yr

I'm not saying fire people. I'm saying that if you're a 10-person agency or a solo founder, you can't afford those five roles. But you still need that operational coverage. The Command Center gives it to you for the cost of a Supabase free tier and an Alpha Agent subscription.

Lessons for Building Your Own

If you want to build something similar, here's what I've learned:

1. Start with the morning briefing

Don't try to build the whole dashboard on day one. Start with one cron job: a morning Slack summary. Once you taste the productivity gain of walking into your day with full context, you'll be motivated to add more modules.

2. Use boring technology

The Command Center uses vanilla Web Components. No framework. No build step. No dependency hell. Each module is a single JavaScript file that fetches JSON and renders HTML. This sounds primitive, but it means any module can be written, tested, and deployed in 15 minutes. The 27 modules in our dashboard are all under 200 lines each.

3. Embrace eventual consistency

Not everything needs to be real-time. Client health scores update every 30 minutes. PR data refreshes hourly. GA4 alerts run daily. The only real-time component is the trigger queue (every 60 seconds). Match your refresh rate to the decision cadence — most operational decisions don't need sub-second data.

4. Make the AI write to files, not APIs

The simplest integration pattern: a cron job runs, the AI analyzes something, and it writes a JSON file. The dashboard reads the file. No API authentication, no webhook configuration, no error handling for failed API calls. Just files. It's old-school and it works beautifully.

5. Color-code ruthlessly

The Command Center uses exactly three alert colors: green (healthy), amber (needs attention), and red (act now). Every module speaks this language. When I glance at the dashboard, I don't read — I see colors. If everything is green, I move on. A red dot anywhere gets 100% of my attention.

The Compounding Effect

Here's what surprised me most: the Command Center gets more valuable over time, not less. Each new cron job adds another data stream. Each new module creates another lens into the business. The nightly brainstorm agent (running at 4 AM PT) generates ideas that reference patterns spotted by the client health module. The lead enrichment system uses talking points informed by our Contentful audit findings. The whole system cross-pollinates.

After three months of operation, we have:

  • 63 automated jobs running on schedule
  • 27 dashboard modules providing operational visibility
  • 7,247 content entries continuously audited
  • 30+ open PRs tracked across 8+ client repositories
  • AI-enriched lead profiles with fit scores and talking points
  • Zero missed client escalations
  • 4-minute morning briefings replacing 45-minute review sessions

Is This the Future of Operations?

I think so. The traditional model — hire specialists, hold meetings, aggregate information manually — doesn't scale for small teams. And most "dashboarding" solutions (Datadog, Grafana, Tableau) require you to know what you're looking for. The Command Center is different because the AI decides what to surface. I didn't write a query for "flag when a client has 13 stale PRs." The client health module noticed the pattern and raised the alert.

This is the shift from passive dashboards (you look, you find) to active dashboards (it looks, it tells you). And it's available today, not in some future product roadmap.

If you're an ops leader drowning in tabs, or a founder who's been putting off hiring a PM because the budget isn't there — let's talk about building your Command Center.


Footnotes

  1. Qatalog & Cornell University, "Workgeist Report 2021: The impact of tool overload on workers," 2021. Found that knowledge workers spend 59 minutes/day searching for information and 36% say finding information across apps is their biggest friction point. PDF