| Agency Task | What it involves | Frequency | Current time | Command Center | Agent |
|---|---|---|---|---|---|
| Media Planning | |||||
| Brief intake & structuring | Gather client objectives, KPIs, audience, channel preferences into structured doc | Per campaign | 2–4 hrs | AUTOMATE Orchestrator parses brief, validates required fields, routes to Phase 1 | Captain's Table |
| Audience research & segmentation | Analyse historical data, identify segments, model reach/frequency | Per campaign | 4–8 hrs | ASSIST Produces ICP validation, segment maps, opportunity sizing. Human validates. | Lookout Spyglass |
| Budget allocation modelling | Distribute budget across channels using CPM assumptions and historical benchmarks | Per campaign | 4–8 hrs | ASSIST Generates allocation model with rationale. Planner adjusts at PAUSE checkpoint. | Cartographer |
| Competitive analysis | Track competitor spend, messaging, creative, positioning gaps | Monthly | 6–12 hrs | AUTOMATE Continuous competitor monitoring, positioning gap analysis, white space report | Spyglass |
| Content calendar creation | Build 4-week publishing schedule with platform-specific timing | Monthly | 3–6 hrs | AUTOMATE Generates full calendar with TOF/MOF/BOF distribution, platform cadence | Cartographer |
| Media Buying & Optimisation | |||||
| Campaign pacing review | Check spend rate vs. plan; adjust daily budget caps | Daily | 30–60 min/day | HUMAN Requires platform access. System can set pacing rules and flag deviations. | — |
| Bid management | Monitor win rates, adjust bids per audience/placement | Daily | 1–2 hrs/day | HUMAN In-platform execution. System writes optimisation rules (hard pause/scale thresholds). | Quartermaster (rules only) |
| Optimisation rules & thresholds | Define when to pause, scale, or test based on CPA/ROAS targets | Per campaign | 2–4 hrs | AUTOMATE Writes hard rules: "if CPA > €X for 3 days, pause placement." Human approves at PAUSE 2. | Quartermaster |
| Channel budget reallocation | Shift budget between social, display, video based on performance | Weekly | 1–2 hrs | ASSIST Recommends reallocation with evidence. Buyer executes in platform. | Quartermaster Chronometer |
| Campaign Management & QA | |||||
| Creative spec gathering | Collect ad formats, sizes, file types per placement | Per campaign | 2–3 hrs | AUTOMATE Generates spec sheet from campaign plan + platform requirements | Cartographer |
| QA checklist execution | Verify naming, budgets, dates, targeting, frequency caps, tracking pixels | Per launch | 4–8 hrs | ASSIST Generates pre-flight checklist from plan; flags mismatches. Human signs off. | Captain's Table |
| Trafficking & tag management | Upload creatives to ad server, configure line items, test tags | Per launch | 4–8 hrs | HUMAN Requires ad server access. System produces the trafficking sheet; ops team executes. | — |
| Daily campaign health check | Monitor impressions, CTR, frequency; flag anomalies | Daily | 1–2 hrs/day | ASSIST Writes trend narratives, anomaly flags, hypothesis log. Human decides action. | Chronometer |
| Creative & Ad Copy | |||||
| Ad copy writing | Headlines, body copy, CTAs per platform and audience segment | Per campaign | 6–12 hrs | AUTOMATE Generates 15–25 variants per brief, on-brand (style memory from Navigator). Human curates. | Scribe |
| Email sequence writing | Multi-email drip: subject lines, body, CTAs, timing logic | Per campaign | 8–16 hrs | AUTOMATE Full sequence with branching logic, timing, exit conditions. Human reviews. | Scribe |
| Landing page copy | Hero, value props, social proof, CTA — per campaign | Per campaign | 4–8 hrs | AUTOMATE Produces full page copy with conversion structure. Human + designer finish. | Scribe |
| Visual creative production | Design banners, video, rich media assets | Per campaign | 10–40 hrs | HUMAN Creative production stays with designers. System provides briefs + Canva handoff. | Scribe (brief to Canva) |
| A/B test design | Define hypothesis, variants, success criteria, duration | Ongoing | 2–4 hrs/test | AUTOMATE ICE-scored test backlog, hypothesis definition, stat sig criteria. Human approves. | Helmsman |
| Analytics & Reporting | |||||
| Weekly performance report | Pull metrics across platforms, summarise trends, flag anomalies | Weekly | 4–8 hrs | AUTOMATE Trend narrative, anomaly log, hypothesis register. Analyst reviews, doesn't build. | Chronometer |
| Monthly performance deck | 5–10 slides with channel breakdown, insights, recommendations | Monthly | 8–12 hrs | AUTOMATE Full narrative report with business impact framing. Human edits for client tone. | Ship's Log |
| QBR preparation | 25–30 slide deck: performance vs. objectives, learnings, recommendations | Quarterly | 20–30 hrs | ASSIST Generates narrative structure, data synthesis, recommendations. Human adds context + presents. | Ship's Log Chronometer |
| Attribution & data reconciliation | Cross-check platform data (GA, Meta, ad server); flag discrepancies | Monthly | 2–5 hrs | ASSIST Flags mismatches between sources, labels as Hypothesis or Confirmed. Analyst investigates. | Chronometer |
| Experiment results & scaling decisions | Evaluate A/B test results; recommend scale/kill/hold | Ongoing | 2–3 hrs/week | AUTOMATE Statistical evaluation, scaling recommendation, learning logged to Data Vault | Helmsman |
| Strategy & Account Management | |||||
| Strategy synthesis | Merge internal data + market intel into positioning and messaging framework | Per campaign | 8–16 hrs | AUTOMATE Falsifiable positioning, contra-strategy, messaging hierarchy. Strategist reviews at PAUSE 1. | Navigator |
| Client status updates | Weekly call prep: pacing, optimisation actions, next steps | Weekly | 2–3 hrs | AUTOMATE Generates status summary from latest agent outputs. AM personalises and sends. | Ship's Log |
| Client relationship & stakeholder mgmt | Calls, negotiations, expectation management, scope discussions | Ongoing | 10–20 hrs/week | HUMAN Relationships don't automate. System frees AM time to focus here instead of reports. | — |
| Brief writing for client approval | Turn strategy into structured brief document for sign-off | Per campaign | 3–5 hrs | AUTOMATE Generates client-facing brief from Phase 1 strategy output. AM reviews and presents. | Navigator Scribe |
These require platform access, client trust, or creative judgment that the system shouldn't own.
| Task | Why it stays human |
|---|---|
| Ad platform execution | Bid changes, budget adjustments, campaign pauses happen inside DSPs/ad managers. The system writes the rules and recommendations; the buyer executes. No API access to client ad accounts needed during POC. |
| Creative design & production | Visual assets (banners, video, rich media) need designers. The system produces the briefs, copy, and spec sheets that feed the design process — but Photoshop/Premiere stays with humans. |
| Trafficking & tag management | Uploading tags to ad servers, configuring line items, and QA-testing live tags requires ad server access and ops expertise. System generates the trafficking sheet; ops team clicks the buttons. |
| Client relationships | Calls, negotiations, trust-building, scope discussions. The system frees account managers from report-building so they can spend more time here — which is what retains clients. |
| Final approval at checkpoints | Every PAUSE gate requires a human decision: approve, revise, or reject. The system never ships anything without sign-off. That's the design. |
| Phase | Current labour | With Command Center | Saved |
|---|---|---|---|
| Planning & Strategy | 20–40 hrs | 4–8 hrs (review + approval) | ~75% reduction |
| Creative & Copy | 20–40 hrs | 6–12 hrs (curation + design) | ~65% reduction |
| Reporting & Analysis | 15–25 hrs/month | 3–5 hrs/month (review only) | ~80% reduction |
| QBR Preparation | 20–30 hrs | 4–6 hrs (editing + presenting) | ~80% reduction |
| Account Status Updates | 8–12 hrs/month | 2–3 hrs/month (personalise + send) | ~75% reduction |
| Total per client/month | 80–140 hrs | 20–35 hrs | ~75% reduction |
Translation for a 10-client agency: Instead of 6–8 full-time people on production, reporting, and strategy drafting — you need 2–3 people doing review, client relationships, and platform execution. The rest is system output.
Pick one existing client where the agency is already running campaigns. Run the Command Center in parallel — same brief, same timeline. Compare output quality, speed, and variant volume at the end.
K&B does: Configure global_brand.json with client voice, ICP, and banned vocabulary. Wire connectors (Notion, Slack). Run The Lookout (internal data analysis) + The Spyglass (competitor/market audit) in parallel. Feed both into The Navigator for strategy synthesis.
Agency provides: Client brief, brand guidelines, 6 months of campaign performance data, audience segment definitions.
Output: Falsifiable positioning statement, competitor audit, audience opportunity map, messaging hierarchy.
PAUSE 1 checkpoint: Agency strategist + account manager review Phase 1 output. Approve, revise, or reject. Nothing proceeds without sign-off.
Then: The Cartographer produces funnel map, content calendar, KPI targets. Feeds into The Scribe (ad copy, email sequences, landing page copy) + The Quartermaster (budget allocation, optimisation rules) in parallel.
Output: Campaign plan, 15–25 ad copy variants, email sequence, media plan with hard pause/scale rules.
PAUSE 2 checkpoint: Creative director + media buyer review all Phase 2 output. Select best variants. Approve media plan. Agency team executes in-platform (trafficking, launching).
Then: The Chronometer reads early performance data (if campaign is live) or analyses historical data to produce trend narratives. The Helmsman designs experiment backlog.
Output: Trend report, anomaly log, ICE-scored experiment backlog, scaling recommendations.
The Ship's Log produces the executive summary — business impact narrative, not just metrics.
Comparison report: Side-by-side of what the agency team produced vs. what the Command Center produced for the same brief. Cover: time taken, number of variants, depth of analysis, quality of strategic rationale.
Decision meeting: Present results to agency leadership. If the delta is meaningful → discuss engagement model (Managed, Embedded, or White-Label).
What the agency risks: €2,500 and one brief. What they see: A full campaign cycle — strategy through performance — produced in 4 weeks with dramatically less labour. The output speaks for itself.
| Metric | What "good" looks like |
|---|---|
| Time to strategy | Phase 1 output in <48 hours (vs. 1–2 weeks agency-side) |
| Copy variant volume | 15–25 on-brand variants per brief (vs. 3–5 manually) |
| Strategy quality | Agency strategist rates positioning as "equal or better" vs. what they'd produce |
| Report depth | Hypothesis register + anomaly log included (vs. standard metrics-only report) |
| Labour hours | Agency team spends <15 hours total on review across 4 weeks (vs. 60–100 to produce equivalent) |
| Experiment backlog | ICE-scored backlog of 8–12 testable hypotheses delivered (vs. typically 0–2 ad hoc tests) |
| Brand consistency | Zero banned-word violations. Tone matches brand guidelines across all outputs. |
Don't lead with agents, AI, or Claude. Lead with their margin problem.
Their reality: Clients want more variants, faster turnaround, and deeper analysis — for the same budget. Team is stretched. Reporting eats 30% of analyst time. Strategy gets squeezed into half-day workshops instead of proper research cycles. Junior staff write copy that needs 3 rounds of revision.
Your opening line: "What if your team could focus on client relationships and creative judgment — and everything else was drafted before they sit down?"
Then show them this document. Walk through the mapping table. Let them point at the tasks that hurt most. That's your POC scope.