CookUnity
Growth Co-Pilot40 chef brands · Snowflake × Braze × Meta/Google · live creative + segment + LP tools
Growth Tools
3 liveProduction-style tools wired to the AI co-pilot. Each one makes a real Claude call and returns structured output.
Channel-aware copy generation across Meta, Google, YouTube, email, and SMS — with per-variant angle + rationale.
Plain-English audience → generated SQL against the warehouse, Braze payload, audience size estimate, send-time recommendation, and a draft message.
Generate 3 hero variants (emotion-led / outcome-led / social-proof-led) per campaign with rendered previews and a test recommendation.
Chef Brand Engagement — 30 days
Live dataDaily engagement share across 40 chef brands. Higher = larger slice of network volume on that growth touchpoint.
Chef Brand Leaderboard
Engagement share across all 4 growth touchpoints · last 24h
- 1Rick BaylessYou8.0%0.0
- 2Esther Choi4.5%0.8
- 3Ludo Lefebvre3.7%0.0
- 4Marcus Samuelsson3.6%0.3
- 5Jose Garces3.5%0.1
- 6Michael Solomonov3.1%0.2
- 7Michael Schwartz3.1%0.2
- 8Cat Cora3.1%0.1
- 9Rocco DiSpirito2.9%0.1
- 10Fabio Viviani2.9%0.2
- 11Pat LaFrieda2.8%0.2
- 12Ahmet Kiranbay2.6%0.5
- 13Sergio Tominaga2.6%0.1
- 14Cindy Hutson2.5%0.1
- 15Hong Thaimee2.4%0.2
- 16Hemant Bhagwani2.4%0.2
- 17Pierre Thiam2.3%0.1
- 18Akhtar Nawab2.2%0.1
- 19Harold Dieterle2.2%0.1
- 20Lorena Garcia2.2%0.1
- 21BCakeNY by Miriam Milord2.1%0.1
- 22Kevin Meehan2.0%0.2
- 23Pino Luongo2.0%0.5
- 24Tal Ronnen2.0%0.1
- 25Reset with Angelia Cole2.0%0.1
- 26Brandon Kida2.0%0.0
- 27Bubb's BBQ by Dustin Taylor2.0%0.0
- 28Mike Ding2.0%0.2
- 29Michelle Bernstein1.9%0.2
- 30Gusto Grill by Chef Maribel1.9%0.4
- 31John DeLucie1.9%0.2
- 32Be Well by Chase Evans1.9%0.0
- 33Shirley Chung1.8%0.5
- 34Felipe Donnelly1.8%0.0
- 35Antony Nassif1.8%0.1
- 36Trevor Lui1.7%0.0
- 37Patrick Kriss1.7%0.0
- 38Sophie's Cuban1.7%0.2
- 39Salt of the Earth Bakery1.7%0.3
- 40Einat Admony1.7%0.1
Active Alerts
Auto-detected from Meta/Google ad insights, Braze events, Snowflake rollups, and LP test monitors
Marcus Samuelsson Meta CTR fatigue — 7 days running
Paid AcquisitionMarcus Samuelsson's Meta CTR declined from 1.42% to 0.98% over the last 7 days while CPM held flat. Pattern matches creative fatigue (no offer change, no audience change, just a stale set of 3 hero creatives). Recommend pausing the bottom 2 ad sets and shipping 3 fresh variants before EOW. Studio team has the brief in queue.
Pino Luongo Welcome-flow activation dropped 8pts
CRM LifecyclePino Luongo's Welcome flow activation fell from 41% to 33% over 10 days. Looking at the Braze logs, the most recent segment refresh changed the cuisine-preference field shape; the Welcome flow's first-touch SMS is now skipping ~25% of new subscribers because the segment filter no longer matches. Roll back the segment definition or update the flow to the new schema.
Lorena Garcia LP variant test stalled — call it
Conversion (LP)Lorena Garcia's LP A/B/C test on the Father's Day gift-card hero hit day 14 with no significant winner (p = 0.41). Sample size is sufficient. Recommendation: stop the test, ship the control, and queue a fresh round of variants with a stronger differentiator on the hero copy.
Esther Choi Paid surge — but the win is bought
Paid AcquisitionEsther Choi's Meta acquisition is up 40% week-over-week after the May 18 creative refresh — but Meta CPM rose 18% on the same window. Net CAC is up 11%. The volume is real and the campaign deserves the budget, but the unit-economics gap is widening. Recommend a budget cap by Friday and a deeper-funnel creative angle by next week.
3 chef brands have CRM Welcome flows pending refresh
Cat Cora, Michael Solomonov, and Shirley Chung Welcome flows are running on prompt versions that pre-date the May 1 voice update. Activation rates are holding but tone has drifted off-brand on a sample audit. Queued for the Creative Studio + CRM Lifecycle team to refresh by EOW.
Ahmet Kiranbay — first-cohort volatility expected
Ahmet Kiranbay is the newest chef brand on the platform (38 days live) and is showing day-over-day swings of plus or minus 18 percent across all touchpoints. Pattern matches every new chef launch in the first 90 days. Suppressing further volatility alerts on this brand until day 90. Cohort-1 LTV is tracking on the optimistic end of the model.
Recent Growth Inquiries
Briefs + questions from the Growth team · chef brand routing highlighted · 40 inquiries tracked total
CRM Lifecycle agent built the segment. Generated Snowflake SQL against the subscriber + order tables, returned the Braze segment payload + estimated audience size (3,847 users), and drafted a 3-touch reactivation sequence with send-time optimization per recipient.
CRM Lifecycle agent built the segment. Generated Snowflake SQL against the subscriber + order tables, returned the Braze segment payload + estimated audience size (3,847 users), and drafted a 3-touch reactivation sequence with send-time optimization per recipient.
Routed to the Paid Acquisition agent for Hemant Bhagwani. Pulled the last 14 days of Meta campaign-level data, segmented by ad set, and surfaced a CPL anomaly tied to a creative refresh on May 22. Recommendation: rebalance budget to the top 2 ad sets and refresh the bottom 2 by Monday.
Routed to the Paid Acquisition agent for Lorena Garcia. Pulled the last 14 days of Meta campaign-level data, segmented by ad set, and surfaced a CPL anomaly tied to a creative refresh on May 22. Recommendation: rebalance budget to the top 2 ad sets and refresh the bottom 2 by Monday.
All Chef Brands
40 chefs · engagement across 4 growth touchpoints · 30-day trends
Inquiry Templates
10 canonical growth-team requests · sampled across 4 touchpoints · 40 captured agent replies
What I'd Want to Dig Into in 30 / 60 / 90
7 pillars · 70 starting hypotheses · anchored against the public JD. Click any pillar to expand.
📊Performance Marketing AI Suite
Phase 1Days 1-45Bid strategy analysis, budget pacing, creative performance scoring, and anomaly detection across SEM, Meta, Affiliate, and YouTube — running as agentic workflows that surface decisions instead of dashboards that require interpretation.
Performance Marketing AI Suite
Phase 1Days 1-45Bid strategy analysis, budget pacing, creative performance scoring, and anomaly detection across SEM, Meta, Affiliate, and YouTube — running as agentic workflows that surface decisions instead of dashboards that require interpretation.
Build AI-powered tools to automate bid strategy analysis, budget pacing alerts, creative performance scoring, and anomaly detection across paid channels (SEM, Meta, Affiliate, YouTube, etc.)
- 1Audit current paid-channel reporting and identify the 5 decisions Growth makes most often (rebalance, refresh, pause, scale, kill).
- 2Anomaly detection v1: CPL drift, CPM spike, conversion-rate drop, day-pacing skew — paged to Slack with the suggested action attached.
- 3Bid strategy agent: pulls last 14 days per ad set, flags exhaustion, recommends budget rebalance with rationale.
- 4Creative performance scoring: per-creative CTR + conversion + day-decay, ranked weekly into 'scale / hold / refresh / kill' buckets.
- 5Channel-level cost-of-incrementality checks (not just last-click) tied to the LTV model.
- 6Auto-paused experiments: if CAC exceeds the LTV threshold for the chef brand by >25% for 5 days, pause the ad set automatically with audit log.
- 7Per-chef-brand budget guardrails (max daily, max CAC, target ROAS) encoded as config and enforced by the agent.
- 8Daily executive digest: 3 paragraphs to the Head of Growth on what moved + what's queued. Auto-drafted nightly, reviewed at 7am.
- 9Pre-launch checklist agent: before any new campaign goes live, validates UTMs, conversion event setup, audience overlap, and budget pacing.
- 10Quarterly perf-marketing tooling retro: which agents the team relies on, which got retired, which need a v2.
🎨Creative Studio AI Loop
Phase 1Days 1-60Ad copy and image variation generated at scale, brief automation that closes the loop between analytics and the studio team, and creative output volume that grows without proportional headcount.
Creative Studio AI Loop
Phase 1Days 1-60Ad copy and image variation generated at scale, brief automation that closes the loop between analytics and the studio team, and creative output volume that grows without proportional headcount.
Develop AI workflows for ad copy generation, creative brief automation, image asset variation at scale, and creative performance feedback loops between analytics and the studio team.
- 1Ship generate_ad_variants in production (it's already in this demo) and wire it to the studio team's brief intake form.
- 2Channel-specific format rules encoded as constraints (Meta short form, Google headline+desc limits, email subject/preview, SMS 140 chars).
- 3Brand-voice library per chef brand: persona, tone, vocabulary, prohibitions — sourced from chef interviews + winning creative analysis.
- 4Brief automation: structured intake form -> generated variants -> studio review queue -> winning variant attribution.
- 5Image variation pipeline: Gemini multimodal + style-transfer for chef-led photography variants at scale (food angle, lighting, plating).
- 6Creative-perf feedback loop: every shipped creative tagged with brief metadata; weekly digest shows which angles + voice patterns won.
- 7Banned-phrase + brand-safety linter: catches off-brand or restricted language pre-ship.
- 8Localization expansion: same creative engine generates SP / FR / PT variants for international cohorts with cultural-fit guardrails.
- 9Studio queue dashboard: what's drafted, what's in review, what's shipped, win-rate per brief author.
- 10Quarterly creative-loop retro: did volume grow without head-count cost? Did win-rate hold? What's the next angle to test?
🎯CRM Lifecycle Agentic Workflows
Phase 1Days 1-60Plain-English audience descriptions flow into Snowflake SQL, Braze segments, send-time-optimized messages, and triggered campaigns — minimal manual intervention. The JD's headline workflow, productionized.
CRM Lifecycle Agentic Workflows
Phase 1Days 1-60Plain-English audience descriptions flow into Snowflake SQL, Braze segments, send-time-optimized messages, and triggered campaigns — minimal manual intervention. The JD's headline workflow, productionized.
Implement AI-driven personalization and send-time optimization across email, SMS, push, and in-app channels; build agentic workflows that pull Snowflake data, generate segment logic, and trigger Braze campaigns with minimal manual intervention.
- 1Ship nl_to_braze_segment in production (it's already in this demo) and wire it to the lifecycle team's segment intake form.
- 2Schema-aware SQL generator: the agent knows the Snowflake schema and validates joins, filters, and aggregations against the catalog.
- 3Audience-size estimator that runs the generated SQL in a cheap warehouse + reports estimated reach before commit.
- 4Send-time optimization: per-recipient model trained on historical Braze open/click data; agent suggests send window with confidence interval.
- 5Multi-channel orchestration: a single brief can spawn email + SMS + push + in-app variants with channel-appropriate copy.
- 6Welcome flow refresh: rebuild the activation sequence per cuisine preference, with eval gate on activation lift.
- 7Winback automation: subscribers who skip 3+ weeks auto-enter a winback flow with chef-brand-aware offers, capped per-recipient frequency.
- 8Lifecycle eval set: golden test segments + expected reach + expected send-time, run nightly to catch model + schema drift.
- 9Compliance + frequency caps: hard limits encoded so the agent can't accidentally over-send to a user across multiple campaigns.
- 10Quarterly lifecycle retro: campaign volume, activation rate, churn delta, complaint rate — all tied back to the agent-generated workflows.
🚀Landing Pages & Conversion Engine
Phase 2Days 45-150LP variants generated per campaign, headline testing automated, and personalization logic driven by audience segment or traffic source — no more manual LP build for every promo cycle.
Landing Pages & Conversion Engine
Phase 2Days 45-150LP variants generated per campaign, headline testing automated, and personalization logic driven by audience segment or traffic source — no more manual LP build for every promo cycle.
Automate LP variant generation, headline testing ideation, and personalization logic by audience segment or traffic source.
- 1Ship generate_lp_variants in production (it's already in this demo) and wire it to the LP build pipeline.
- 2Multivariate test harness: deterministic bucketing + sticky 90-day cookie + server-side variant attribution + UTM + click-ID capture.
- 3Per-traffic-source LP swap: Meta audience sees a different hero than Google search audience sees, driven by the agent's audience-fit classifier.
- 4Headline + hero + CTA library tagged by win-rate, retired automatically when significance drops below threshold.
- 5LP brief automation: campaign + audience + offer + brand -> 3 variants in the queue within 5 minutes.
- 6Personalization tokens: chef name, cuisine, offer dynamically inserted per-user with safety guardrails (no PII in hero copy).
- 7LP performance digest: weekly per-variant lift report tied to downstream trial signups + paid activation.
- 8Auto-stop logic: tests with no significance after 14 days + sufficient sample size get auto-stopped with a recommendation.
- 9Per-segment lift attribution: the same variant lifts 8% on Meta lapsed and tanks on Google search — surface and act on it.
- 10Quarterly LP-engine retro: which patterns won across audiences, which got commoditized, what to ship next.
🔧Growth Ops Internal Tooling
Phase 2Days 60-180Internal AI tools — dashboards, Slack bots, report generators — that compress repetitive analytical work and accelerate decision-making across every Growth subteam.
Growth Ops Internal Tooling
Phase 2Days 60-180Internal AI tools — dashboards, Slack bots, report generators — that compress repetitive analytical work and accelerate decision-making across every Growth subteam.
Build internal AI tools — dashboards, Slack bots, report generators — that reduce repetitive analytical work and accelerate decision-making across the team.
- 1Daily Growth Slack digest: auto-drafted summary of yesterday's anomalies + wins + queued decisions, delivered at 7am.
- 2Self-serve report generator: anyone on the Growth team types 'what's the activation rate for Plant-Forward Cohort 2 vs Cohort 1' and gets a report.
- 3Per-subteam dashboards: Performance / Creative / Lifecycle / LP / Brand each get their own quality-of-life surface.
- 4Slack bot for any tool in this app — the same agent + tool set exposed as /growth in Slack so the team can ask without context-switching.
- 5Recurring report automation: weekly chef-brand performance, monthly cohort LTV, quarterly executive review — all auto-drafted.
- 6Trigger library: any Growth event (campaign ship, segment update, LP variant launch) can fire a Slack/email/dashboard hook.
- 7Internal AI assistant for the Growth team — answers 'how do I X', 'show me the docs for Y', 'who's the owner of Z' — grounded in the team's wiki + chat history.
- 8Per-tool usage analytics: which tools are loved, which are ignored. Retire the ignored ones, double down on the loved ones.
- 9Cost-tracking dashboard for every AI call: spend per tool per subteam, budget alerts before headline-grabby overruns.
- 10Quarterly Growth Ops retro: time-saved per week per role, decisions accelerated, what to build next.
🧠Growth Product — AI Features in the Product
Phase 3Days 150-300AI-powered features that accelerate funnel conversion and retention: recommendations, dynamic content, smart nudges. The work moves from marketing surfaces to the product itself.
Growth Product — AI Features in the Product
Phase 3Days 150-300AI-powered features that accelerate funnel conversion and retention: recommendations, dynamic content, smart nudges. The work moves from marketing surfaces to the product itself.
Partner with Product and Engineering to identify where AI-powered features (recommendations, dynamic content, smart nudges) can accelerate funnel conversion and retention.
- 1Recommendation engine v1: chef brand recommendations based on past orders + similar-subscriber behavior + freshness boost for new brands.
- 2Dynamic site content: hero + featured chefs + dish carousel personalized per signed-in user.
- 3Smart nudges: cart abandonment, plan-upgrade prompts, chef-discovery nudges, all driven by predicted-LTV + propensity.
- 4Per-user reorder predictor: which dish a subscriber is likely to repeat next, surfaced in-app.
- 5Cuisine exploration agent: in-app chat that asks 1-2 questions and steers a new subscriber to their best-fit chef brand.
- 6Pause / cancel intercept: when a user opens the pause flow, an agent surfaces the chef brand they'd miss most + a single-week skip option.
- 7Personalized weekly menu reveal: per-subscriber subject line + body + featured chef tuned by past taste + lifecycle stage.
- 8Eval harness for in-product AI: every recommendation logged with predicted vs observed engagement; weekly model-drift alerts.
- 9A/B test framework that respects subscription-cohort effects (a new feature might lift activation but tank 90-day retention).
- 10Quarterly Growth Product retro: which AI features moved the needle on funnel conversion + retention, which got rolled back.
🧱Tooling, Standards & Team Enablement
Phase 3Days 90-365Repeatable skills, plugins for popular LLM clients, internal training, and a Growth team that uses AI as a multiplier — not as a side project. The JD asks for AI-native team behavior; this is how to get there.
Tooling, Standards & Team Enablement
Phase 3Days 90-365Repeatable skills, plugins for popular LLM clients, internal training, and a Growth team that uses AI as a multiplier — not as a side project. The JD asks for AI-native team behavior; this is how to get there.
Build repeatable skills or plugins for popular LLM clients to empower your team. Educate and mentor team members to become more AI-Native.
- 1Reusable tool registry: every tool in the chat is also a Claude/ChatGPT plugin and an n8n/Make module the team can pipe into.
- 2Internal AI office hours weekly: drop-in for any Growth team member with a 'can AI help with X' question.
- 3Per-role training: Performance + Creative + Lifecycle + LP + Brand each get a 1-hour intro + a per-role tool playbook.
- 4Internal documentation site: every tool, every workflow, every prompt pattern, indexed and searchable.
- 5Skill-sharing pipeline: any team member can submit a prompt or workflow; reviewed weekly and added to the library if it generalizes.
- 6Standards: prompt versioning, eval coverage requirements, observability metadata expected on every tool call.
- 7Cost-of-AI literacy: every team member sees their own tool-call spend; quarterly review tied to ROI.
- 8Vendor + model selection framework: when to use Claude, ChatGPT, open-source, fine-tune, or RAG — encoded as a decision tree.
- 9Annual AI-readiness review: where the team was vs where it is, what's next, what to deprecate.
- 10External hiring impact: candidates seeing this stack should want to join. The tooling itself becomes a recruiting signal.