Ideas for using AI to move HBW outcomes — category visibility, prevention, growth. Not setup polish, actual category leverage.
Every night, automatically check Rx weight loss and ED deal pages to make sure they have the right legal language (FDA-approved vs compounded distinction, FTC subscription disclosure, eligibility callouts). New violations get flagged to Safet automatically.
Rx / GLP-1
She's now built refund-targeting deal content for LHR, LASIK, and cosmetic injections one category at a time. Lock the pattern into a PDS-indexed template library so any rep can pull the right refund-driver-targeting copy skeleton for facials, RF microneedling, body contouring, hair restoration, etc. — without her doing the category research each time.
Cross-HBW (all PDS)
Surgical aesthetics is the only High End L4 that hasn't gotten the weekly enterprise tracker treatment that MH, LightRx, SEV, and Simplicity now have. Pick the top 2-3 surgical groups (plastic surgery practice networks, body contouring chains) and stand up the same SF-state-aware tracker against them.
Surgical aesthetics
Refund-reason work has been per-account (LASIK Specialists, Refresh Vero Beach, Tacoma). Roll it up one level: cluster refund reasons across every merchant within a PDS, so she can see which content-addressable reasons (medical ineligibility, scope confusion, scheduling friction) are structural to the PDS vs. merchant-specific.
Aesthetic injectables + LHR + LASIK
Extend the script that auto-fills pitch decks into a continuous prospecting system. Surface multi-location medspas, dermatology groups, dental DSOs that BD should pursue.
BD / Prospecting
Watch each sub-category within L3 High End. Anything down >20% YoY gets surfaced with a RESTORE recommendation: which deals to restructure, which merchants to renegotiate, which to pause. Right now RESTORE is done category-by-category as you notice problems.
L3 High End
Dental and vision are in her scope but have no recent activity. Run the same TG co-occurrence query she built for Portfolio Adoption (cosmetic injections) on dental and vision — find which adjacent services merchants in those categories tend to also sell, and where the multi-deal portfolio opportunity is.
Dental and vision
The Rx weekly tracker now runs on a Monday cron and exposes refund/CVR/coverage gaps. Run a one-time scan of the GLP-1 and TRT supply pipeline (active deals + pipeline opps + lapsed accounts) against the same playbook to identify which Rx merchants are eligible for the same monitoring treatment.
Medical weight loss (GLP-1) + Men's health (TRT)
Recurring manual work that costs you real hours each year. Each candidate shows what it costs now, what it would cost after, and total annual hours back.
One command creates the empty merchant row in the tracker spreadsheet and matching stub in the JSON file, ready for the pre-screen skill to fill in. Runs before the Rx merchant pre-screen workflow so I stop hand-editing two files to add a merchant. Second phase: auto-fill the 'Approved & Live' program list from Salesforce and dim_deal on each refresh so the tracker never drifts from what's actually live.
When: Weekly · ~50x/year
Today: 5-10 min per new merchant, plus recurring drift-fix time
After: Add a merchant with one command; live-program column stays accurate on its own
Annual: ~15 hours/year plus prevention of tracker-vs-reality mismatches
Recurring fix
Two fixes to the CM Impact Tracker weekly refresh so the hero metric reads correctly for refund-focused changes like SEV, and so deals contaminated by Mother's Day traffic can be excluded on demand. Adds a net financial column (change in M1+VFM plus change in refund dollars) and a pre-period Mother's Day flag with a hero toggle. Lives inside the weekly CM Impact Tracker refresh Katelyn reviews to size her category interventions.
When: Weekly · 50x/year
Today: never done systematically
After: Hero number reflects the full financial impact of a CM change (revenue plus recovered refunds), and Katelyn can hide deals whose pre-period was distorted by the Mother's Day spike with one click.
Annual: Prevention-value: $50K-150K (refund-driven wins like SEV no longer read as losses in the hero, so real interventions stop getting under-credited or reversed)
Drift detection
When something in your chase queue hits N days with no activity from the person you're waiting on, auto-generate a polite nudge draft you can send with one click. You decide whether to send, but the drafting work is done.
When: Daily check · ~10 chase items active
Today: ~30 min/week reviewing chase queue + drafting nudges
After: ~5 min/week reviewing pre-drafted nudges
Annual: ~22 hours/year + faster turnaround on stuck threads
Coordination
After an account health review is published to Groupon IQ, automatically create an Asana task for the assigned AM with the report link and key actions. Lives at the tail of the account-health-report skill, right after GIQ upload. Without it, reviews get published but the AM never sees them in their task queue, so recommendations sit unread.
When: ~30x/year (per account review published)
Today: 5-10 min per review to manually create the Asana task with link + summary
After: Skill creates the Asana task automatically as the final publish step, assigned to the right AM with the GIQ link prefilled.
Annual: ~4 hours/year + prevents reviews from going unread
Distribution
After shipping a fix or update across multiple Enterprise brand trackers, post a single consolidated Google Chat message to Colin, Lindsey, and Dusan with all brand links and headlines. Lives at the end of any Enterprise Tracker release cycle (e.g., when Massage Heights, Hand & Stone, Massage Envy, and European Wax Center are all aligned). Without it, stakeholders get fragmented per-brand pings and lose the cross-brand 'these are now in sync' signal.
When: Monthly · ~12x/year
Today: 15-20 min per occurrence (gathering links, drafting headlines, formatting)
After: One command assembles the consolidated message — all four brand links, current-state headlines, and what changed since last update — ready to paste into Google Chat.
Annual: ~3-4 hours/year
Distribution
Adds a visible 'X uploads retried' counter to the Google Chat message that the HBW Enterprise Tracker posts every Monday after its run. Right now when the tracker quietly retries a failed upload, the recovery is invisible in chat — you only see the final 'success' note. With the counter, you can immediately see whether Monday's run was clean or whether the safety net had to catch something, which is the early signal that the underlying upload path is degrading.
When: Weekly · 50x/year
Today: never done systematically
After: Monday chat post shows retry count alongside success status; degradation visible before it becomes a full failure
Annual: Prevention-value: $500-2000 (one missed Monday tracker run during MBR/5/15 prep)
Drift detection
Adds a small panel to the HBIC Setup tab showing what changed in my Claude setup since yesterday: new rules, new skills, new memory files, and how many scheduled jobs are running. Sits next to the existing setup summary so I see it every time I check the dashboard. Without it, small changes pile up unnoticed until the monthly system audit surfaces a mess.
When: Daily · 365x/year (passive display)
Today: never done systematically
After: Setup drift is visible each morning instead of waiting for the monthly audit to catch it
Annual: ~12 hours/year (shorter monthly audits, earlier catch of bad rules)
Drift detection
Two small fixes on the SEV Laser content pipeline. First, a tighter rule for pulling the location-specific anchor sentence from each SEV landing page so freeway names and full street addresses stop leaking into deal copy. Second, a delete step in the GIQ upload script so renamed reports don't leave stale copies cluttering the catalog. Both live inside the SEV Laser national rollout workflow.
When: Per rollout batch · ~6x/year for SEV, reusable across brands
Today: 10-20 min per batch cleaning anchors and stale reports by hand
After: Anchors come out clean the first time; GIQ stays tidy without manual purges
Annual: ~8 hours/year plus prevention of bad copy shipping to reps
Recurring fix
New skills you could build once and reuse across many projects. Each one compounds — first use justifies it, every reuse after is free.
One skill that scans any deal page for required disclosures: FDA-approved vs compounded language, FTC subscription/renewal, medical eligibility callouts, off-label drug claims. Reusable across Cosmetic Injections, Rx Weight Loss, Hair Restoration, IV Therapy — anywhere regulatory language matters.
First use: Rx weekly tracker (replace nightly scan from automation candidates)
Reusable in: Cosmetic Injections, Rx, Hair Restoration, IV Therapy, Weight Loss, Men's Health
Build effort: ~6-10 hours
Compliance
Takes any category, unit count, and market data → computes the price floor and runs the Botox-style weekly sweep. Same lifecycle-tracking pattern, but configurable per category instead of hardcoded.
First use: Dermal Fillers (extend Botox sweep)
Reusable in: Botox, Dermal Fillers, Lip Filler, IV Therapy, Kybella, anything with unit pricing
Build effort: ~8-12 hours
Pricing + compliance
Currently a POC in Portfolio Pitch Generator. Productionize as a reusable skill that any project can call to: infer services offered, classify merchant type, identify multi-location chains, surface contact info.
First use: BD prospect lists (extend Portfolio Pitch scraper)
Reusable in: Portfolio Pitch, BD prospecting, account research, competitive analysis
Build effort: ~12-15 hours
Merchant intelligence
When a new telehealth merchant needs pre-screening, one command creates the empty tracker row and the research stub — instead of hand-editing the spreadsheet and JSON separately. Long-term add-on: pull 'Approved & Live' merchants from Salesforce automatically so the tracker's live-status column stays synced with what's actually running on Groupon. Without this, every new merchant is a manual copy-paste into two files, and the live column drifts from reality until someone spot-checks it.
When: ~3-4 new merchants/month · 40-50x/year
Today: 10-15 min per new merchant (manual xlsx + JSON entry)
After: Merchant name + URL creates tracker row and research stub instantly; live status auto-refreshes weekly from Salesforce
Annual: ~10-12 hours/year
Recurring fix
The HBW Shopping List Top Growth ranking runs quarterly across ~20 divisions to guide rep prioritization — same underlying pull, new dates, re-rank. Currently rebuilt by hand each cycle. As a byproduct, this can flag deals whose PDS category doesn't match what they actually sell (e.g., a massage deal miscoded as facial) so the category team fixes taxonomy hygiene without running a separate audit. Both live inside the SL cycle Katelyn runs for HBW rep guidance.
When: Quarterly refresh · optional monthly check-in = 4-16x/year
Today: 45-60 min per quarterly refresh (script edits, re-runs, review)
After: One command produces the quarterly ranking; a category-mismatch list drops out as a byproduct for the category team
Annual: ~4-6 hours/year on refresh + prevention value on ~20-40 miscoded deals/year caught earlier
Recurring fix
Two small changes to the weekly CM Impact Tracker dashboard: (1) make the hero number count refund reductions the same as revenue gains, so refund-focused wins like SEV Laser show up correctly instead of looking like flat results; (2) tag deals whose 'before' window overlapped Mother's Day and add a toggle to exclude them, so seasonal spikes don't fake a win. Lives in the weekly Friday tracker refresh she uses to prove category-management impact.
When: Weekly · 50x/year
Today: 30-60 min per week manually caveating results, plus occasional wrong-direction reads that need reversing
After: Hero number tells the truth on refund plays out of the box; contaminated deals are visible and one-click removable.
Annual: Prevention-value: 25-50 hours/year of manual caveating + avoiding wrong impact claims to leadership
Recurring fix
Shows day-over-day changes to Katelyn's setup — new rules, new skills, new memory files, scheduled-job count deltas — directly on the Setup tab of the HBIC dashboard. Currently these small changes only surface during the monthly system audit, so a month of micro-choices accumulate invisibly and have to be untangled all at once. Making drift visible daily catches problems while the context is still fresh.
When: Daily view · scanned ~50x/year
Today: never done systematically (monthly audit catches drift 30 days late)
After: Setup tab surfaces a 'what changed since yesterday / last week' panel so drift is visible in real time
Annual: ~3-5 hours/year on monthly audit untangling + prevention value on rules that would have been caught early
Drift detection
Two small fixes surfaced during the SEV Laser 33-location content rollout. First: pull the distinctive fact sentence from each location's landing page automatically — cleanly, without leaking freeway names or street addresses into the About section. Second: when a GIQ report gets renamed, auto-delete the old version so the folder doesn't accumulate zombie files. Both apply any time a chain merchant needs per-location content built across many sites.
When: 3-5 chain rollouts/year · ~100+ locations touched
Today: 3-5 min per location for the anchor pull + manual GIQ cleanup after each rename
After: Anchor extracted cleanly during scrape; old GIQ reports auto-deleted when renamed
Annual: ~6-10 hours/year
Pattern extension
Improvements to your own automation and dashboards. Surfaced automatically from things your past wins have pointed at.
fix freshness-pill regex so tiles show real refresh dates instead of 'unknown'
surfaced from HBIC 5/15-prep Rx tile stale-data fix + weekly heavy-refresh launchd job · 2026-W29
Scheduled refresh — currently manual, tracked in tasks/followups-open.md as launchd or GIQ automation candidate
surfaced from HBW Northstar Benchmark Tool — canonical taxonomy refactor + Asana description + analyst-ready XLSX · 2026-W29
Wire GCHAT_WEBHOOK into launchd wrapper (~/Library/Scripts/hbw-search-trends-monthly.sh) so Aug 2 refresh actually notifies Katelyn when surfacing files are ready. ~10 min task.
surfaced from HBW Search Trends — taxonomy handoff prep for Aleksandra (Keboola) + Methodology modal + Unclassified-terms surfacing automation · 2026-W29
Wire fetch_sf_state.py into weekly-refresh.sh so Friday launchd fires it before refresh.py; also address the two open handoff items — SEV Protection-vs-Improvement reframe + Mother's Day PRE-window contamination flag
surfaced from CM Impact Tracker v25 — SF Deal_State__c filter to drop paused deals from portfolio aggregates + weekly refresh · 2026-W29
Package the SQL → builder → GIQ upload as a scheduled monthly refresh (launchd or GIQ automation) so the tool auto-refreshes as new deal launches age past the 90-day observation window
surfaced from HBW Northstar Tool v2 — cumulative-cohort refresh + phantom filter + methodology tab + canonical TG sync + GIQ v2 publish · 2026-W29
Round 4: move report-template-eligible rules out of Document construction section into report-template skill knowledge; consolidate Verification & methodology entries into their existing DISCIPLINE bundles.
surfaced from Memory consolidation — 3-round reorg of top-level MEMORY.md into project + skill folders · 2026-W29
Level-2 sentinel enhancement (still pending from Jul 10 entry): include last 20 lines of the flagged log in the gchat alert so root-cause is visible without opening a shell — would collapse diagnosis cycle from minutes-with-terminal to seconds-in-chat
surfaced from launchd EX_CONFIG re-diagnosis + fix — TCC stdio path (not LWCR) · 2026-W29
V2 sentinel — include log tail in alert message (level-2 diagnosis-in-alert) for 10x faster fix time
surfaced from Freshness sentinel + launchd LWCR fix · 2026-W28