analytics/
Performance reporting + attribution.
autonomous weekly reports · custom dashboards · briefing automation
Brandon Whalen — AI automation + marketing consultant
brandon@frndly:~$ whoami > brandon whalen > ai automation + marketing for small teams > ~$500K/mo in managed ad spend > vibecodes for fun # type 'help' to poke around · or scroll down for the actual work ↓
CAT INVADERS
← → move · space fire · esc quit · r restart hold buttons to move · tap fire to shoot · tap × to quit
Practice areas
brandon@frndly:~$ cat practice_areas/ # using ai workflows as the force multiplier, not the headline
content_ops/
Content production at scale.
research → draft → qa → publish pipelines
paid_media/
Google Ads, Meta, YouTube across a portfolio of clients.
pacing automation · auto-draft ad copy · cross-platform weekly reports
seo/
Technical audits + AI-era content ops.
citation QA · source diversity checks · Ahrefs + GSC pipelines
Selected work
brandon@frndly:~$ ls projects/ # click any tile to read more
the problem
a healthcare client’s paid media reporting ate 4-5 hours every monday. manual data pulls across google ads, meta, and analytics; rewriting the same narrative every week; screenshot-and-paste into a deck.
the approach
pull → analyze → draft → ship. claude agents handle the narrative with a human-in-the-loop review step. n8n orchestrates. output lands in slack + a google doc every monday at 7am.
the outcome
- 4-5 hrs → ~15 min/wk of human review
- report ships before the monday standup instead of during it
- the client loves it
the problem
a legal client needed to ship long-form content at scale without sacrificing source quality or voice.
the approach
multi-stage pipeline: topic intake → research → draft → citation verification → source diversity check → final human pass → publish. each stage is a claude agent with the right context.
the outcome
- 12+ pages shipped per month
- zero citation errors in published work
- voice stays consistent across authors
the problem
managing paid media across multiple entertainment properties meant juggling platforms, budgets, and reporting cycles — each with their own quirks.
the approach
unified pacing + alerting layer that watches spend trajectories, surfaces anomalies, and auto-drafts adjustment suggestions. human approves, pipeline applies.
the outcome
- weekly reviews take minutes instead of hours
- budget flags catch drift before it becomes a problem
- platforms stay in sync without manual cross-referencing
the idea
could an agent actually run a fantasy team — waivers, lineups, trade analysis — and stream its reasoning live as it worked?
the build
fastapi backend with server-sent events. multiple claude agents (scout, analyst, manager) collaborate in real time. i watch the decisions unfold in a dashboard, override when i disagree.
the result
- fully autonomous weekly lineup + waiver decisions
- live streamed reasoning i can interrupt
- competitive against the league
→ https://github.com/brandonrwhalen/fantasy-baseball-copilot
the idea
my mornings were a mess of context-switching between inboxes and tools. i wanted one agent that ingested everything overnight and handed me a prioritized briefing with drafted replies.
the build
parallel sub-agents for each source (email, slack, clickup, calendar). results merge into a momentum-based briefing with micro-steps, time estimates, drafted replies, and scheduling autopilot.
the result
- runs every morning at 7am
- replaces my fragile manual triage
- genuinely changed how i start the day
the idea
rather than bespoke scripts per client, build one orchestration layer that every automation plugs into.
the build
self-hosted n8n on a frndly-digital subdomain, with version-controlled workflow exports. scheduled flows + webhook-triggered flows. MCP servers hang off it for claude agent access.
the result
- 20+ production flows across my portfolio
- one control plane for everything
- new automations deploy in hours, not days
Tools I use
brandon@frndly:~$ ls ~/tools/ # stuff i use daily
- astro/ # what this site is built with
- claude-code/ # primary ai coding agent
- gemma/ # local ai — current gemma release
- ghostty/ # terminal emulator — love it
- hermes/ # open-source ai agent with persistent memory
- mcp/ # open protocol for connecting ai to tools
- n8n/ # workflow automation engine
- oh-my-opencode/ # opencode plugin ecosystem
- ollama/ # local model runner
- opencode/ # ai coding agent for the terminal
- opencode-zen/ # curated model set tuned for coding agents
- warp/ # terminal with ai completions
About
brandon@frndly:~$ cat about.md ---
paid media director / ai automation specialist / multi-client operator.
i run paid media for a portfolio of brands while building the tooling that lets me run paid media for a portfolio of brands. outcome-obsessed, process-light, allergic to unnecessary complexity.
also i make drum samples and mix movie-trailer audio when no one’s looking.
— open to 1-2 new clients / q2 2026
Contact
brandon@frndly:~$ connect → linkedin.com/in/brandonwhalen/