A tool-using agent that reads your live Shopify or WooCommerce store, predicts which customers are about to churn, runs missions overnight, drafts lifecycle emails on its own, and lists your catalog on ChatGPT Instant Checkout and Google's agentic-commerce graph.
Bring your own AI keys · Five providers + fallback · No credit card
Built on the infrastructure serious merchants already trust
"Every ecommerce founder I know loses hours every week to manual price tweaks, stale stock alerts, copying spreadsheets into chatbots, and writing the same lifecycle emails from scratch. An autonomous teammate that sees the store, remembers the rules, runs missions overnight, and ships the change on your approval — that's what's missing in 2026. We're onboarding the first ten merchants by hand right now."
— Marko, founder
Why merchants switch in 2026
Manual ecommerce ops
With WooPulse
Five steps
Paste your Shopify or WooCommerce API keys. Products, orders, customers stream in within seconds.
Anthropic, OpenAI, Grok, Gemini, or Ollama Cloud. Two for fallback; council mode synthesises drafts.
Open the dashboard — health score 0–10, three priority actions, what the agent did overnight.
Writes (price, copy, stock, images, bundles) need a one-click confirm before they touch the store.
Toggle missions to run on cron. Findings land in your inbox, summary lands in your email at 08:00.
The full 2026 stack
Built across six shippable phases — every one a single git commit on main. Each feature works on its own, but together they read like one product.
Deterministic 0–10 health score across 5 weighted pillars + top 3 actions. Never an AI hallucination — pure arithmetic.
Per-SKU 30/60/90-day demand projection with reorder-by dates and suggested PO quantities. Otto Group cut stockouts 80 % with the same pattern.
Per-customer 12-month LTV plus 14- and 30-day churn risk, computed locally from each customer's own buying cadence. Six predictive segments replace rule-based RFM.
Mines line-item co-occurrences across pairs & triples, lift-filters the noise, generates bundle names + pitches + margin-aware discounts. One-click publish to your store.
Five-stage editor — Welcome · First Purchase · Browse Abandon · Win-Back · Milestone. Templates personalised per customer by AI; the cron runner sends every 5 minutes.
Six mission presets run on cron — Inventory Health, Slow Movers, Boost Revenue, Sale Prep, SEO Copy, Retention. Findings land in your dedicated agent inbox.
0-100 score per product against the signals ChatGPT, Perplexity, Google's agent graph, and Claude Browse actually retrieve on. Worst-first with autofix hints.
A ~6 kB drop-in <script> that lets shoppers describe what they want in natural language and surfaces matched products with custom AI-written pitches.
Expose your catalog to OpenAI Instant Checkout + Google's Universal Commerce Protocol. Manifest + paginated feed + product detail; checkout via Stripe Connect (Phase 5b).
Pricing, descriptions, images, bundles — four tabs, every change measured automatically. DALL·E 3 imagery, margin-aware repricing, multi-model copy synthesis.
Pending / failed / on-hold orders bucketed by age (Fresh / Warm / Cold) with AI-drafted win-back emails per customer. No external plugins.
Row-level security on every team-scoped table. Agencies and multi-store operators switch teams from the sidebar; data never crosses.
01 · Today, in 60 seconds
A deterministic 0–10 store health score across five weighted pillars, three priority actions with one-click deep-links, plus a recap of what the agent did overnight. Computed in your browser from real data — never an AI hallucination.
Today, in 60 seconds
Refreshed 12 min agoStore health · last 30 days
Solid overall. Biggest lift opportunity is inventory health.
Top 3 actions today
02 · Predictive customers
Per-customer predictions — 12-month LTV, 14-day and 30-day churn risk, next-purchase probability — computed locally from each customer's own buying cadence. Six predictive segments replace the rule-based RFM bucketing that everyone else still ships.
About to churn
€1,840 LTV at riskAnna Bauer
Last 62d ago · LTV €620 · cadence 28d
Marko Petrović
Last 78d ago · LTV €420 · cadence 21d · VIP
Sofia Rossi
Last 41d ago · LTV €310 · cadence 18d
Lukas Weber
Last 35d ago · LTV €280 · cadence 22d
03 · Inventory autopilot
Per-SKU 30/60/90-day demand projection with moving-average + recent-trend correction. Reorder-by dates and suggested PO quantities cover your lead time and safety buffer. Velocity anomalies flag SKUs whose 7-day rate diverges > 50 % from baseline — surges or drops.
Inventory forecast
lead time 14d04 · Sell smarter
Four tabs on one page: margin-aware repricing, multi-model description rewrites, DALL·E 3 image variants, AI-discovered bundle suggestions. Every change records a tracking pixel and the daily optimizer measures the lift automatically. One-click approval gates every write.
Bundle suggestion
+€820 / 30d projectedStoneware Mug + Linen-Bound Diary
Bought together 8× · lift 2.3 · stationery cohort
AI-written selling pitch
"Customers who reach for our hand-thrown stoneware mug also bring home the linen diary — start your day with both. Save 10% when bought together."
The autonomy layer
Klaviyo's Customer Agent runs on cron. Triple Whale's Moby runs on cron. WooPulse does too — six mission presets, lifecycle email sender, daily digest, every one wired to pg_cron on your own Supabase.
Step 1 · Schedule
Six presets out of the box — Inventory Health, Slow Movers, Boost Revenue, Sale Prep, SEO Copy, Win-Back. Toggle which run, when (daily morning, twice daily, weekly Monday, weekday-only, or a custom cron).
Step 2 · Inbox
Every successful mission run drops structured findings into the agent inbox at /inbox — severity-coded, deep-linked, with raw token + duration audit.
Inventory Health · 06:00
3 SKUs need urgent reorder · €1,640 at risk
Win-Back · Tue 07:00
8 customers with €1,840 in 12-mo LTV at risk
Step 3 · Digest
One polished HTML digest summarises mission findings + unread inbox items + key metrics. Local-hour aware, per-category filterable, daily or weekly cadence. Empty days are skipped.
Today's brief
5 missions · 11 findings · 2 high priority
Top: Stoneware Mug stocks out in 4d · reorder 80
Cron pipeline · self-hosted Supabase
*/5 * * * *
agent-mission-runner
runs due missions
0 2 * * *
enqueue-lifecycle-jobs
scans customers → jobs
2-59/5 * * * *
email-sequence-runner
AI → SMTP send
*/15 * * * *
digest-sender
user-local digest
Klaviyo-lite, in your app
A five-stage editor: Welcome · First-Purchase Nurture · Browse Abandon · Win-Back · Milestone. Each template is a subject line + body prompt with {{handlebar}} placeholders the runner substitutes per customer. AI generates the body in your tone; your SMTP sends it; every send is logged with provider, model, and tokens.
Lifecycle pipeline
Active14
In queue
2
Sending
87
Sent · 7d
1
Failed
Welcome
T+1h after registration
First-Purchase Nurture
T+3d after first order
Browse Abandon
Storefront events · 5b
Win-Back
T+45d inactive · ramps to 90d
Milestone
Anniversary of first order
Built for AI shoppers
McKinsey forecasts $900 B–$1 T in US retail revenue from AI-agent-driven shopping by 2030. ChatGPT Instant Checkout is live since Sept 2025. Google's Universal Commerce Protocol shipped Jan 2026 with Shopify, Walmart, Target. Merchants who don't list become invisible.
Per-product 0–100 score across six checks LLMs actually retrieve on: schema.org/Offer completeness, description richness, alt-text quality, FAQ schema, category coverage, SKU/brand identifiers.
A 6 kB drop-in <script> that mounts a Shadow-DOM chat box. Shoppers describe what they want in natural language; the widget keyword-matches with price-ceiling parsing, AI reranks, and writes a one-line pitch per result.
A capability manifest at /.well-known/agentic-commerce.json, paginated normalised catalog feed, single-product endpoint. Supports OpenAI ACP + Google UCP side by side. Dry-run default for the first 14 days.
$1T
McKinsey projected US agentic-commerce revenue by 2030
900M
ChatGPT weekly users with Instant Checkout
+35%
AOV lift from AI-driven product recs (2026 cohort)
80%
Otto Group's stockout reduction with AI forecasting
Honest comparison
Plugins automate rules. Chatbots talk. WooPulse does.
| Generic Woo plugin | ChatGPT & chatbots | Klaviyo / Triple Whale | WooPulse | |
|---|---|---|---|---|
| Reads live store data | ||||
| Updates products on its own | Rule-based | |||
| Predictive LTV + churn | ||||
| Autonomous nightly missions | Partial | |||
| Lifecycle email automation | Plugin | |||
| LLM-SEO audit | ||||
| ACP / Google UCP listing | ||||
| Bring your own AI keys | N/A | |||
| Pricing | $/store + addons | $20+/seat | $45+/mo + usage | Beta · invite-only |
Platform-agnostic by design
The 25-tool agent registry is identical across platforms; only the read / write adapters change. Add a platform without changing the agent.
25 tools fully wired to the WooCommerce REST API. All cron runners + ACP gateway tested end-to-end.
Adapter shipped. Partners onboarding this month shape what lands first.
Third platform reserved for custom spec. Drop-in once endpoints land.
FAQ
Hand-picked merchants only during beta. Bring your own keys. Direct founder onboarding.
We reply within a day — every day, including weekends during launch.