Funnel — All lead types

Live

Full pipeline from raw scraped sites through classification, CRM upload, verification, outreach and signups.

Funnel — Shopify App Signups

Live

Merchants who installed the AmbassadorFlow Shopify app. Each install lands here as a lead (status signed_up → installed → active). Detected via the Shopify Partner API poller.

Funnel — Shopify Brands

Funnel — WooCommerce Brands

Funnel — Agencies

Funnel — Shopify Agencies

Funnel — WooCommerce Agencies

Funnel — Clutch Shopify Agencies

Funnel — Clutch eComm Marketing Agencies

Funnel — SaaS Partners

Funnel — Job Boards

Funnel — Reviews

Funnel — Expos

Competitor Review Leads

loading…

Brands that left negative reviews on competitor Shopify apps, resolved to named decision-makers via MoltSets + website scraping. In-policy leads (≤12 months) get a personalized LinkedIn DM via SendPilot. The Message Variants section below groups leads by complaint scenario so you can review and improve each message over time.

Get Leads

loading…

How much of our Swedish Shopify base the GetLeads database (334M contacts) actually covers, measured shop by shop with free count calls — credits are only ever spent when a record is pulled. The sweep runs on the Mac and mirrors here every 20 seconds; the panel refreshes every 15 while it moves. Every contact bought so far is listed at the bottom.

loading…

Pipeline Flow

loading…

Scrape → classify → verify, end to end. Each stage hands to the next automatically: a page classified a minute ago is in the verifier’s queue a minute later. Auto-refreshes every 15s while open.

Site Traffic

loading…

Every request to ambassadorflow.com, from the Vercel drain. Vercel forgets its own logs after about a day; this keeps 14. Deliveries are batched, so a visit takes up to a minute to appear.

Live Scrape Feed

loading…

Every site streamed live from any running scrape or classification. Raw scrapes keep a rolling window of the most recent ~5,000 sites; classifications persist in full. Auto-refreshes every 8s while open.

MoltSets enrichment — live feed

Decision-maker emails scraped on MoltSets, biggest shops first. Auto-refreshes every 30s.

Leads collected

Yield by band — emails/company; empirical where processed, estimated otherwise

Current run

Latest leads found — ● verified ● catch-all ● unverified ● invalid

Previous runs

Process control

checking…

Start or restart the long-running jobs. Restarting is safe: an interrupted batch writes nothing for the rows it did not finish, so they simply come round again.

Loading…

Script vs model

loading…

What the Python scrapers produced, and what gemma judged on top of it. Every row passes through the scripts; only a subset reaches the model.

Daily sending — emails sent against the cap, and the positive replies they earned

Loading…

Live campaigns — sending right now: today's emails, and the positive replies they earned

Loading…

Results — what the sending has earned

–

Emails per positive reply ↗

–

Positive replies ↗

read them

–

1 · Installed app ↗

–

2 · + Plan activated ↗

–

3 · + Embed live ↗

AmbassadorFlow revenue — what these stores pay us

–

MRR — subscriptions

–

Fee this month

–

Stores referring ↗

Billed by month

Plan mix

Store revenue — what the installed shops are selling, as APP has measured it

–

30-day store revenue ↗

–

Live stores ↗

–

Not live yet ↗

–

Stores measured ↗

–

Last full month

By month

Biggest stores ↗

This system's campaigns — the SE and US runs built and sent here; the 2025 lists are not counted

–

In our campaigns ↗

–

Replied ↗

–

Interested ↗

–

Waiting on you ↗

–

Meetings ↗

–

Signed up ↗

–

Live inboxes ↗

–

Classified by Gemma ↗

Sending & deliverability — volume against capacity, and whether the mail is landing

Loading…

Lead pipeline — scrape → classify → verify: how fast each is running, and when it finishes

Loading…

Leads still queued — how much each live campaign has left, and how long that lasts

Loading…

System health

Live

Scrapers, sending capacity, inbox health, and data freshness across the stack.

Loading…

GTM Agencies — 4 Pipelines + Brands

~75% Complete

Four parallel agency pipelines (Shopify, WooCommerce/Clutch, Clutch Shopify, Clutch eComm Marketing) plus a brand-direct pipeline. Each follows: scrape → enrich → score → campaign → outreach → CRM.

Pipeline 1: Shopify Agencies

Scoring done

Shopify Partner Directory agencies. Source: official partner directory, 2-pass scraper (listing + detail pages). DB: Shopify-Agencies

74%

Scrape 100% Websites 96% Intel 76% Score 97% Campaign stale Outreach 0%

1 Scrape Done 2,416 agencies

Partner directory scrape

100%2,416 agencies

Website URL extraction

100%2,416 URLs

Contact email extraction

97%2,353 emails

2 Enrich Done 2,313 text / 1,823 intel

Website text scraping

96%2,313 / 2,416

AI intel extraction

76%keyword only

3 Score Done 2,351 scored — 2,125 T1

Lead scoring (v2)

97%2,351 scored (2,125 T1 / 217 T2)

4 Campaign Stale 21,177 generated (Mar 13)

Campaign generation

0%stale (Mar 13)

5 Outreach Pending Not started

Email verification

0%MillionVerifier

ReachInbox upload

0%not started

Scoring Dimensions (lead_scorer_v2.py)

Dimension	What it measures	Tier 1 signal
Vertical match	Beauty, supplements, pet, coffee keywords	Beauty/supplements → Tier 1
Service type	Email, retention, CRO vs. dev-only	Email/retention/CRO = high fit
E-commerce focus	DTC/ecommerce signals in website text	DTC focus detected
Agency quality	Partner tier, rating_score, rating_count	Premier + high rating

Tiers: Tier 1 = score ≥30 + vertical keyword • Tier 2 = score 15–29 • AI review = 10–24 (claude -p) • Skip = <10

Stale Campaign Segments (Need Regeneration)

Segment	Leads	Generated	Status
Nutrition (supplements, health)	13,186	Mar 13	Stale
General (remaining Tier 1+2)	5,059	Mar 13	Stale
Premier Plus (Shopify top-tier)	1,117	Mar 13	Stale
Pet & Coffee (subscription)	784	Mar 13	Stale
Beauty (skincare, cosmetics)	685	Mar 13	Stale
Scandinavia (Nordic, any vertical)	352	Mar 13	Stale

Pipeline 2: WooCommerce Agencies

Scored, needs campaigns

WooCommerce agencies from Clutch.co directory (184 pages). This is the largest pipeline by volume. DB: WooCommerce-Agencies

1 Scrape Done 9,195 agencies

Clutch directory scrape

100%9,195 agencies (184 pages)

Website URL extraction

100%9,190 URLs

Contact email finding

38%3,525 / 9,195

2 Enrich Done 7,621 text / 6,658 intel

Website text scraping

78%7,173 / 9,195

AI intel extraction

78%keyword only

Clutch profile enrichment

0%LinkedIn, etc.

3 Score Done 7,173 scored — 3,754 T1 / 1,710 T2

5-dimension scoring

78%7,173 scored (0–120 scale)

4 Campaign Pending Not started

Campaign generation

0%5,464 T1+T2 ready

5 Outreach Pending Not started

Email verification

0%MillionVerifier

ReachInbox upload

0%not started

Scoring Dimensions (score_woocommerce_agencies.py)

Dimension	Max Score	What it measures
Industry match	35	Nutrition, beauty, pet, coffee verticals
Service fit	25	Email, retention, CRO, growth marketing
Platform profile	20	Shopify +10, WooCommerce +5, Wix/Squarespace -5
Agency quality	15	Rating, reviews, Clutch verification
Size fit	5	Team size & min project thresholds

Tiers: Tier 1 = score ≥30 + vertical keyword • Tier 2 = score ≥20 • AI review = 12–25 (claude -p) • Skip = <12

Service Breakdown (Top Detected)

Primary Service	Count	Fit for AmbassadorFlow
Development	4,212	Low (dev-only)
SEO	835	Medium
Paid Ads	707	Medium
Email Marketing	291	High
Social Media	182	Medium
Branding	173	Low

Pipeline 3: Clutch Shopify Agencies

Scraping in progress

Secondary Shopify agency source from Clutch.co (not the official Shopify Partner Directory). Includes non-partner agencies + verified Clutch reviews. DB: Clutch-Shopify-Agencies

1 Scrape Enriching —

Directory listing scrape

Done~10,104 agencies

Profile enrichment

~In progress

Website URL extraction

0%waiting

2 Enrich Pending Waiting for scrape

Website text scraping

0%waiting

AI intel extraction

0%waiting

3 Score Pending Not started

Lead scoring

0%waiting

4 Campaign Pending Not started

Dedup vs Pipeline 1

0%waiting

Campaign generation

0%waiting

5 Outreach Pending Not started

Email verification

0%waiting

ReachInbox upload

0%waiting

Pipeline 4: Clutch eComm Marketing Agencies

Re-scraping directory

Broader e-commerce digital marketing agencies (not platform-specific). These are marketing-first agencies — email, growth, retention, SMS, paid ads — with higher propensity to work with DTC brands. DB: Clutch-Ecomm-Marketing

1 Scrape Enriching —

Directory listing scrape

Done~14,152 agencies

Profile enrichment

~In progress

Website URL extraction

0%waiting

2 Enrich Pending Waiting for scrape

Website text scraping

0%waiting

AI intel extraction

0%waiting

3 Score Pending Not started

Scoring

0%waiting

4 Campaign Pending Not started

Dedup vs Pipelines 1–3

0%waiting

Campaign generation

0%waiting

5 Outreach Pending Not started

Email verification

0%waiting

ReachInbox upload

0%waiting

Pipeline 5: WooCommerce Brands

Classification in progress

Direct outreach to WooCommerce e-commerce brands from Storeleads data (4.6M records). Currently in the classification & rule-building phase — 13 rounds complete, learning which sites are real online stores vs. B2B/service/etc.

1 Data Done 4.6M brands loaded

Storeleads data loaded

100%4,613,637

2 Scrape Borderline text fill running 2.25M have text · filling 34k borderline

Brand text coverage (Storeleads)

49%2.25M / 4.6M have text

Borderline text fill (AI-review)

~34,122 borderline rows (no usable text → blocks Woo classifier)

3 Score Classification 2.3M scored — 882 classified

Rule-based scoring

50%2.3M / 4.6M

E-commerce classification

<1%882 classified (13 rounds)

Exclusion rule learning

80%238 KB rules

AI scoring (claude -p)

0%borderline brands

4 Campaign Pending Not started

Campaign generation

0%not started

Contact email finding

0%not started

Export with geo filter

100%script ready

Brand Scoring Dimensions (score_brands.py)

Dimension	Max Score	What it measures
Category match	40	Beauty, supplements = highest; pet, coffee = secondary
Repeat purchase	25	Evidence of subscriptions, replenishment, loyalty
Revenue / size	25	Estimated annual revenue signals
Opportunity	20	Market size, growth potential
Brand quality	15	Trust signals, reviews, site polish

Tiers: Tier 1 = score ≥50 + beauty/supplements • Tier 2 = score ≥35 • AI review = 15–34 (claude -p) • Hard filters: excludes dropshippers, POD, B2B, restaurants, service businesses

Pipeline 6: Shopify Brands

Gemma classification running

Direct outreach to Shopify e-commerce brands from Storeleads data (2.5M records). Local Gemma 4 (Ollama, free, on the Mac) is classifying the 18,513 target-vertical tier-1/2 brands in hybrid mode.

1 Data Done 2.5M brands loaded

Storeleads data loaded

100%2,491,156

2 Scrape In progress Brand website text scraping

Brand website scraping

~in progress

3 Score Pending Not started (WooCommerce rules transferable)

Rule-based scoring

0%not started

E-commerce classification

0%not started

AI scoring (claude -p)

0%not started

4 Campaign Pending Not started

Campaign generation

0%not started

Contact email finding

0%not started

Shared Infrastructure

Production-ready

Tools and systems shared across all four agency pipelines: enrichment engine, campaign generator, outreach tooling, CRM integration, orchestration.

✓ Built & Working

✓AI Enrichment (extract_agency_intel.py): 3-pass (keyword → Claude Haiku → self-verify)
✓Few-shot learning: intel_examples.json auto-corrects from past mistakes
✓Campaign generator (campaign_generator.py): 6 segments, SpinTax A/B/C variants
✓Email verification (verify_leads.py): MillionVerifier integration
✓ReachInbox uploader: round-robin sender assignment, warmup score ≥80 filter
✓Gmail classifier: interest / objection / unrelated reply detection
✓Landing pages (generate_pages.py): unique slugs per lead, Supabase upsert
✓CRM (6_crm/): Google Sheets setup, lead import, commission calculator
✓Scoring logger: appends run summaries to CRM Scoring tab
✓Orchestration: status.py (live dashboard), run_overnight.sh

☐ Waiting (Cross-Pipeline)

!Post-send automation: auto-reply within 3 min (stub exists, not wired)
!TidyCal demo scheduling: not started
!SPF/DKIM/DMARC: email infrastructure setup
☐3-touch bounced sequence: not started
✓Personalized pages: 14,700+ slugs live on Supabase
☐Cross-pipeline deduplication: agencies appearing in multiple Clutch DBs

Enrichment Cost Model

Component	Model	Cost per 10K
Keyword pass	Regex (free)	$0
AI enrichment	Claude Haiku	$1.50–$7.50
Agency scoring (borderline AI)	Claude Haiku (`claude -p`)	$0 (subscription)
Brand scoring (borderline AI)	Claude Haiku (`claude -p`)	$0 (subscription)
Brand auto-research	Claude Haiku (`claude -p`)	$0 (subscription)

Database Locations

Pipeline	Database Path	Size
Shopify Agencies	`/AmbassadorFlow/work/Shopify-Agencies`	—
WooCommerce Agencies	`/AmbassadorFlow/work/WooCommerce-Agencies`	—
Clutch Shopify	`/AmbassadorFlow/work/Clutch-Shopify-Agencies`	—
Clutch eComm Marketing	`/AmbassadorFlow/work/Clutch-Ecomm-Marketing`	—
Shopify Brands	`/AmbassadorFlow/work/Shopify-Storeleads.duckdb`	—
WooCommerce Brands	`/AmbassadorFlow/work/woocommerce_test_full_text_check.duckdb`	—

GTM Agencies — Total Monthly Cost: $210–$420

Service	What For	Model / Tier	Unit Cost	Est. Volume/mo	Est. $/mo
Claude Code sub	All AI classification + scoring + enrichment	Claude Haiku via `claude -p`	$0 marginal	All volume	$0
ReachInbox	Campaign sending + warmup	Plan-based	~$199–$499/mo flat	5–10K leads	$200–$400
MillionVerifier	Email verification	Bulk API	$0.002/email	5–10K emails	$10–$20
Total GTM Agencies					$210–$420

All AI costs covered by Claude Code subscription (claude -p --model haiku). No per-call API charges. Biggest variable cost is ReachInbox for campaign sending.

Website & Partner Portal

~85% Complete

Next.js 15 marketing site + Supabase-backed partner portal. Full authentication, multi-tier commissions, admin dashboard. Deployed on Vercel.

85%

Marketing site Partner portal Admin dashboard Auth & commissions Stripe payments Theme App Extension

100%

Landing page

100%full feature overview

Competitor comparisons

100%7 pages

Alternative pages

100%dynamic /alternatives/

Personalized landing pages

100%dynamic [slug]

Pricing plans

100%3 tiers

LP visit tracking

100%Sheets + Supabase

98%

Authentication (signup/login)

100%Supabase Auth

Dashboard (KPIs + chart)

100%4 KPIs, 30-day chart

Brands management

100%list, status, revenue

Payouts page

100%history + status

Referral system

100%2-tier, 5% upstream

Revenue breakdown

100%12-month chart

Partner settings

100%name, company, payment

Commission engine

100%20% + 5% referral

Click tracking data

85%collected, no UI

100%

Partner management

100%approve/suspend

Partner detail pages

100%referral chain, stats

Revenue dashboard

100%GMV, fees, top 10

Brand management

100%all brands, status

Payout management

100%create + mark paid

Partner type assignment

100%agency, consultant…

100%

Order webhook

100%attribution + AF fee

Brand status webhook

100%status + touchpoints

Click tracking API

100%IP hash, referral

Platform detection API

100%Shopify/Woo detect

Partner signup API

100%referral code gen

Partner settings API

100%PATCH profile

Industry benchmarks API

100%11 industries

Stripe / payment integration

0%manual payouts only

Email notifications

0%no email library

CSV export

0%not started

Theme App Extension

0%Shopify widget

Hybrid attribution

0%cookie + discount

Webhook HMAC validation

50%Bearer only

Click analytics dashboard

0%data exists, no UI

Two-factor authentication

0%not started

Admin audit logging

0%not started

API documentation

0%not started

Tech Stack

Layer	Technology
Frontend	Next.js 15, React 19, TailwindCSS 4, Radix UI
Auth	Supabase Auth (email/password)
Database	Supabase (PostgreSQL)
Charts	Recharts 3.8
Deployment	Vercel
Types	TypeScript 5

Website — Monthly Cost: $90–$150

Service	What For	Plan	Est. $/mo
Supabase	PostgreSQL database + Auth	Free → Pro ($25 base + row overage)	$60–$100
Vercel	Next.js hosting + serverless functions	Pro ($20/mo + analytics)	$30–$50
Total Website			$90–$150

Note: Supabase currently on free tier (50K rows max). Will need Pro tier ($25/mo + variable) once lead_pages + revenue_events + click_events exceed 50K rows. Vercel Pro includes Web Analytics + Speed Insights.

Competitor Monitor

Live

Weekly source monitor (Hetzner cron, Mondays 09:00) watches each competitor's pricing & feature pages and flags changes to re-verify in the comparison tables. Same alerts go to Telegram.

–

Pages watched

–

Changes logged

–

Unreachable

never

Last check

When	Competitor	Change to re-verify
Loading…

Comparison pages

35 indexed competitor comparison pages are live under ambassadorflow.com/compare (source-verified matrices; the long-tail set is noindex until it earns demand). Google Search Console wiring is pending.

Content Gaps

Live

Google Search Console queries mapped against our published pages. Opportunities are searches where we already rank in striking distance (positions ~4–20) or pull impressions without a dedicated page worth building.

–

Keywords tracked

–

Opportunities

–

Striking distance

–

Monthly impressions

Query	Pos	Impr.	Clicks	Gap
Loading…

Leads

Company	Contact	Email	Type	Status	Campaign	Last Touch

Lead Sources & Provenance

Where these leads came from and how they were filtered, from the raw database down to what we actually emailed.

The funnel

By status

By email type

Store/generic = info@, sales@, hello@ (general inbox). Named = a person's address.

Activity

Campaigns

Conversations & Auto-replies

Live from the ReachInbox unified inbox. Each thread shows the lead's reply and our auto-reply (flagged). Click a conversation to expand the full thread.

Loading conversations…

Sending Inboxes

Inbox	Name	Provider	Status	Health	Daily Limit	Sent Today

Email Sending Capacity

Override /inbox From To

Selected range

This Month

Range Utilization

Daily Sends vs Capacity

Each bar is one day (emails sent vs the /day capacity). Grey bars are weekends, which are excluded from the capacity total.

Funnel

Every domain Gemma has classified, through to a reply. Filters recompute the whole funnel.

Leaks — leads that skipped a step

Lists

Gather companies from All companies, then Prepare. The checks run cheapest first — deduplication, suppression, is-it-a-shop, platform, liveness and language are all free. Verification is the only step that costs money, so it runs last and only ever sees what survived.

Classification

Gemma decides whether a domain is an e-commerce shop and what it sells. Nothing may be emailed before it has an answer here, and it runs locally, so it costs nothing but time.

Email verification

Verification is the bottleneck on every segment — there are always more leads than deliverable addresses. A verdict is bought once and never paid for twice, and a shop whose site is down is skipped before a credit is spent on it.

Which campaign the credits were spent for

Waiting to be verified — live, never contacted

Live — addresses as they come back

live

History — what each day's credits bought

Recent verification runs

Activity

auto-refresh

Every action across the system, newest first — a reply arriving, a suppression, a stage move, a draft written, a job finishing. The Live Feed shows conversations; this shows what happened.

Live Feed

Awaiting our reply

All companies

One row per company, newest action at the top. A company that replied on any channel is marked blocked and every automation skips it — see Do not contact for the full list and for what the block has actually stopped.

Quick segments

🏢 The company ▾

What the business is, independent of us

📍 Where it came from ▾

Which list, scrape or channel put it in front of us

🎯 Fit and score ▾

How good a prospect we think it is

✉️ How to reach them ▾

Whether we hold a usable way in

📤 What we have done ▾

Our side of the conversation

💬 What they did ▾

Their side of it

Awaiting reply

🚦 May we contact them ▾

The suppression gate

🧩 What they run ▾

Detected from their site — 20,065 companies have at least one

Outreach state — across every source. Each company is in exactly one state.

Channel dots: grey = not contacted, blue = contacted, green = replied, hollow = no handle for that channel. Click a row for the full history.

Do not contact

Everyone no automation may contact, on any channel, whatever source they came from. The block is computed from every signal at once — a reply, a hand-written message, a booked meeting, a signup, an unsubscribe, a Telegram pause, a manual stop — so email cannot keep sending to someone who already answered on LinkedIn. An out-of-office auto-reply does not block.

Blocked sends — what the suppression actually prevented

The list

All shops

Awaiting reply

Outreach state — where the shops in this list stand right now. "This list" is our built candidate pool: every domain our scrapers pulled in and Gemma classified, plus leads that entered outreach. Not all of Shopify. Each shop is in exactly one state.

Channel dots: grey = not contacted, blue = contacted, green = replied, hollow = no handle for that channel. Click a row for the full history.

Angles

One angle is one argument for why a visitor should come back, not one ad. Two angles from the same territory are a copy variation rather than a test, so a wave is built by taking the best angle from each territory instead of the loudest five overall. Headlines are capped at 40 characters and descriptions at 30 because the platforms truncate past that; the limit is enforced here, in the API and in the database, so copy that saves is copy that exports.

Experiments

One row per angle × audience. The Meta ad-set id is recorded here when the upload comes back, because Meta reports spend against an ad set and knows nothing about our angles — that mapping has to be written down when the ad set is made, not reconstructed from ad names later. An ad set is not a test until it has run three weeks, and cost per signup is the number to judge it on, never CTR.

Angle	Territory	Audience	Status	$/day	Meta ad set	Started

Build

Pick who the ads are for. Every angle written for that buyer appears below with its creative and the copy that ships with it — the call-out already applied, exactly as it will read in the upload. The words and the picture are on one page because they are one ad.

Creative

The same creative/template.html the PNG renderer uses, rendered here in the browser — so what you see is what uploads. Edit the wording in Angles; this reflects it on save. Download writes a PNG named exactly as the Image File Name column in meta_bulk.csv, so dropping the files into Meta’s bulk import makes the sheet resolve by name.

Results

What each angle cost and what it produced. Spend comes from Meta against an ad set; outcomes come from our own ads_touch rows, keyed on the utm_term the generator stamps with the angle id. Judge on cost per signup, never on CTR — the loudest angle and the one that converts are usually not the same ad. An angle needs three weeks before it means anything, so days is shown next to the money.

Angle	Territory	Spend	Landings	Leads	Signups	Cost/lead	Cost/signup	Days	Verdict

Copy Studio

A segment is a saved filter, so it stays live as brands become sendable. The preview below runs the same resolvers and the same checks the push runs, against real brands from the segment. A lead that cannot fill a slot the body uses is excluded rather than mailed with a gap in it.

Segments

Templates

Signals

Every brand with the signals the copy branches on. Checked means we fetched the site; no programme means we fetched it and found nothing. A brand that is neither is unknown, not clean, and is never counted as having no programme. Detection reads server-rendered HTML and follows one hop, so a programme behind a JavaScript footer reads as none — open a row to see the evidence a call was made on.

Domain	Brand	Category	Country	Signals	Address	Verification	Score
Loading…

Segments

The stage between the pipeline and outreach. Each segment is platform-verified live (we fetch the site, not trust the label), the emails are cleaned for free (syntax, dedupe, live MX, brand-match) before any MillionVerifier credit, and it is split by verified platform. Shopify and WooCommerce become separate campaigns so a shop is never told it is on the wrong platform.

Campaigns

Reply rate is only meaningful for campaigns this system ran. The 2025 campaigns show 100% because only the shops that replied were ever imported, so their denominator is the repliers themselves.

Built here

Results from ReachInbox

Reminders

quiet for at least

Conversations where we answered last and they went quiet. Writing a reminder creates a draft from that conversation — it is never sent until you approve it in Drafts, where it threads back into their own email chain.

Scheduled emails

Live

Every follow-up the sequence scheduler will send, soonest first. Move one, reword one, or cancel it — a pending step is a plan, not a promise. Stop conditions still run at send time: a reply, an install or a manual email stops the whole run regardless of what is listed here.

Loading…

Drafts

Nothing here sends on its own. Approving marks a draft ready; a separate step puts it on the wire, so a mis-click cannot email anyone.

All agencies

Awaiting reply

Outreach state — where the agencies in this list stand right now

Channel dots: grey = not contacted, blue = contacted, green = replied, hollow = no handle for that channel. Click a row for the full history.

Outreach settings

Every rule the automation follows. Changes apply within 30 seconds, no restart.

Waiting for you

What the bot just did

Deploy token

So dashboard deploys stop expiring. Create a token at vercel.com → Account Settings → Tokens with No Expiration, scope it to the ambassadorflows team, and paste it here. Saved to the server env file, never to the code.

…

Merchants — AmbassadorFlow APP

Their data, copied here. Click a row for every field, including the raw response.

📈 Installs per day

📥 Events received from APP

Loading…

Pipeline Board

Everyone we are actually working. Click a card to open it; drag it to another column to move the stage.

Where the conversations are happening

…

Reply Trainer

Draft and approve auto-replies to common prospect questions. Generation runs claude -p on the server. Approve good ones to build the training set.

Loading…

Cold Email Trainer

Pick a lead from the CRM; the server reads their website and drafts a personalized first cold email. Edit and approve.

Loading…

Per-lead Reply Trainer

Pick a lead, paste their reply, and draft a personalized response based on their website. Edit and approve.

Loading…

Objection Handling

Draft and approve the best short answer to each common objection.

Loading…

Cold Email Personalizer

Write or paste a cold email; the trainer suggests the merge variables to personalize it per brand, and rewrites it as a reusable template.

Loading…

Approved Examples

Everything you've approved across the trainers. This is the corpus the replies are tuned from.

Loading…

SaaS Partners

~60% Complete

Scrape Shopify App Store, WordPress.org, and WooCommerce.com for complementary SaaS tools (email, subscriptions, loyalty, reviews, shipping) as partnership leads.

60%

Scrape 6,679 apps Scored 4,698 Tiered top 41 Contacts 20 Email finding 0.5% Outreach 0%

Scrape

3 marketplaces

→

Consolidate

4,618 leads

→

Tier & Curate

41 top-tier

→

Email Finding

0.5% have emails

→

Outreach

Not started

100%

Shopify App Store scraper

100%2,688 apps

WordPress.org scraper

100%3,523 plugins

WooCommerce marketplace

100%468 extensions

Competitor filtering

100%208 excluded

Unified CSV export

100%4,618 deduped

100%

5-point scoring engine

100%4,698 scored

AI pass for uncertain leads

100%score 25–44 band

Competitor penalty scoring

100%auto-exclude

Top-tier ranking

100%41 companies tiered

Tier 1 deep research

100%20 with program details

34%

Contact finding script

100%website + email gen

Contacts found at scale

<1%20 / 4,698

Email verification

0%MillionVerifier

Outreach automation

0%not started

CRM integration

0%not started

Company validation

0%not started

Data by Source

Source	Total	Competitors	Qualified
WordPress.org	3,523	156	3,367
Shopify App Store	2,688	48	2,640
WooCommerce.com	468	4	464

Tier 1 Partnership Targets (Top 20)

Company	Category
Klaviyo	Email & SMS
Omnisend	Email & SMS
Attentive	Email & SMS
Postscript	Email & SMS
Sendlane	Email & SMS
Judge.me	Reviews & UGC
Loox	Reviews & UGC
Okendo	Reviews & UGC
Junip	Reviews & UGC
Recharge	Subscriptions
Skio	Subscriptions
Loop Subscriptions	Subscriptions
Smile.io	Loyalty
LoyaltyLion	Loyalty
Rebuy	Upsell & CRO
Replo	Upsell & CRO
Gorgias	Support
AfterShip	Shipping & Returns
Triple Whale	Analytics
Elevar	Analytics

SaaS Partners — Monthly Cost: $0

All scraping uses Playwright (free, open-source) and the WordPress.org REST API (free, no auth). Data stored in local DuckDB (free). No paid services required for current pipeline.

Future cost: Email finding (if using Hunter.io, Apollo, etc.) would add ~$50–200/mo depending on volume. Outreach automation would use ReachInbox (shared with GTM Agencies).

Job Board Scraper

~90% Complete

Scrapes 14 remote job boards daily for ambassador/referral/influencer marketing roles, filters with 50+ keywords, detects e-commerce companies, enriches with contact info.

90%

14 scrapers 4,823 records Daily schedule E-commerce detection DB persistence pending

95%

API-based scrapers (11)

100%Remotive, RemoteOK…

HTML scrapers (2)

100%BuiltIn, DailyRemote

ATS scrapers (2)

80%hardcoded companies

LinkedIn scraper

70%guest API, 50 max

100%

Keyword filtering (50+)

100%ambassador, referral…

E-commerce detection

100%56 platform signals

Email extraction

100%domain + SMTP

Deduplication

100%per-run dedup

100%

CSV/JSON output

100%4,823 records

Daily scheduling

100%macOS launchd

Rate limiting & error handling

100%graceful retries

Cross-run deduplication

0%not started

Database persistence

0%file-only output

ATS company expansion

0%beyond 30-40 hardcoded

LinkedIn API upgrade

0%guest API limited

14 Active Sources

Source	Type	Status
Remotive	Public API	Active
RemoteOK	Public API	Active
ArbeitNow	Public API	Active
WeWorkRemotely	Public API	Active
Himalayas	Public API	Active
Jobicy	Public API	Active
LinkedIn	Guest API	Limited
WorkingNomads	Public API	Active
NoFluffJobs	Public API	Active
JobSpresso	Public API	Active
DailyRemote	Public API	Active
Built In	HTML scrape	Active
Greenhouse	ATS scrape	Active
Lever	ATS scrape	Active

Job Boards — Monthly Cost: $0

All 14 scrapers use free public APIs or Playwright (free). Data stored locally as CSV/JSON. Scheduled via macOS launchd (free). No paid services.

SMTP email verification feature exists but uses outbound port 25 (free, no API needed).

Competitor Reviews

~75% Complete

Competitive intelligence: scrape negative reviews from 5 platforms, classify pain points with AI (Claude Sonnet), generate personalized outreach, notify via Slack.

75%

Scraped 1,048 Categorized 198 (19%) Outreach 62 (6%) Slack notifications G2/Capterra blocking

92%

Shopify App Store scraper

100%star filtering, pagination

G2 scraper

80%blocking issues

Capterra scraper

80%blocking issues

Trustpilot scraper

100%working

Reddit scraper

100%7 subs, 12 queries

Competitor mapping

100%90+ competitors

13%

Pain point classification

19%198 / 1,048

Outreach message generation

6%62 / 1,048

Uses Claude Sonnet for classification (10 pain point categories) and personalized outreach. ~300–500 tokens input, ~600 output per review.

80%

SQLite deduplication

100%state mgmt

Slack notifications

100%webhook alerts

Continuous monitoring

100%monitor.py

Pricing analysis

100%scrape + report

Reddit pipeline integration

50%standalone only

Trend analysis

0%not started

Automated CI reports

0%not started

Scraper health monitoring

0%not started

Reviews — Monthly Cost: ~$0.04

Service	What For	Model	Volume	Est. $/mo
Claude (Anthropic)	Pain point classification + outreach gen	Sonnet 4.6	~50–100 reviews	~$0.04
Slack	Webhook notifications	Free (incoming webhooks)	~50 notifications	$0
Playwright	Scraping (Shopify, Trustpilot)	Free	—	$0

Near-zero cost. Claude Sonnet is only called for AI-enriched reviews (~300–500 tokens input, ~600 output per review). At 100 reviews/month this is ~80K tokens — under $1.

Trade Show Expos

~50% Complete

Scrape exhibitor lists from beauty, food, pet, and retail trade shows. 35 scrapers targeting 20+ events. Website enrichment as post-processing.

50%

35 scrapers built 10,671 exhibitors 3,219 classified 4 scrapers broken ~5% enriched

100%

Interzoo 2026

100%4,820 exhibitors

SuperZoo 2026

100%821 exhibitors

Pet industry events

100%325 exhibitors

90%

High Point Market

100%1,664 exhibitors

NRF Big Show

100%1,030 (29% enriched)

Shoptalk Spring

100%604 (72% enriched)

Shoptalk Fall

100%135 (89% enriched)

Groceryshop

100%43 (86% enriched)

80%

Beautyworld ME

100%scraper built

NYSCC Suppliers Day

100%scraper built

ABS / Premiere Orlando

100%scraper built

Cosmoprof (intl)

50%partial data

Beauty enrichment

<1%643 total, 0.2%

60%

Expo West

<1%1 / 3,200 (broken)

Sweets & Snacks

0%scraper broken

Summer Fancy Food

<1%12 / 2,400

NRA Show

5%98 / 2,000

Other food/bev events

100%297 (14.5% enriched)

30%

E-commerce classification

100%3,219 classified

URL enrichment script

100%Playwright-based

URLs extracted

~5%most events missing

Export pipeline

100%JSON → CSV

Contact finding

0%not started

Script consolidation

0%36 files, messy

Clean up debug HTML

0%44 files

International events

0%SIAL, Cosmoprof

Events by Category

Event	Category	Exhibitors	Enriched	Status
Interzoo 2026	Pet	4,820	0%	Need enrichment
High Point Market	Retail	1,664	0%	Need enrichment
NRF Big Show	Retail	1,030	29%	Partial
SuperZoo 2026	Pet	821	0%	Need enrichment
Shoptalk Spring	Retail	604	72%	Good
Pet Industry events	Pet	325	0%	Need enrichment
Beauty events (5)	Beauty	643	0.2%	Partial scrapes
Food/Bev events (7)	Food	297	14.5%	Partial scrapes
Shoptalk Fall	Retail	135	89%	Good
Groceryshop	Food	43	86%	Good

Expos — Monthly Cost: $0

All scraping uses Playwright (free) + BeautifulSoup (free). Data stored as local JSON/CSV. No paid APIs.

LinkedIn Monitor

Built, Not Active

Playwright-based LinkedIn post comment scraper. Extracts commenter profile URLs and adds them to a Sendpilot email campaign. Includes watch mode with 5-minute polling.

40%

Scraper built Sendpilot API Watch mode Not configured No production runs

100%

LinkedIn comment scraper

100%Playwright

Watch/polling mode

100%5-min interval

Config & session mgmt

100%Chrome profile

100%error handling

100%

Sendpilot API client

100%add leads, list

Deduplication store

100%JSON-backed

Post URL configuration

0%LINKEDIN_POST_URL

Target post identification

0%high-value posts

First production run

0%never run

LinkedIn — Monthly Cost: $50–$150 (when active)

Service	What For	Est. $/mo
Sendpilot	Email campaign management for LinkedIn leads	$50–$150
Playwright	LinkedIn comment scraping	$0

Currently $0 — not yet activated (LINKEDIN_POST_URL not configured). Sendpilot cost kicks in only when monitoring starts and leads are being added to campaigns.

Other Projects & Resources

Claude Skills Production

34 reusable marketing/growth/ops skill modules with SKILL.md templates

SEO: 5 CRO: 6 Content: 5 Sales: 3 Ads: 3 Other: 12

Podcast Outreach Reference

18 research docs with 100+ channel targets (podcasts, YouTube, newsletters, communities)

Docs: 18 Size: ~500KB

Auto Research Ready

Autonomous ML experiment swarm — GPT training with iterative agent modifications

Framework: PyTorch 2.9 Model: GPT + RoPE

Meta-Dream Daemon Running

Nightly launchd job that syncs project memory across all 9 sub-projects

Last run: Mar 26, 03:05

Empty / Placeholder Directories

Directory	Created	Status
cliffs/	Mar 26	Empty
legal/	Mar 26	Empty
logistic-partners/	Mar 19	Empty
reddit/	Mar 18	Empty
-design/	Mar 18	Empty
-growth/	Mar 18	Empty
-operations/	Mar 18	Empty
-product/	Mar 18	Empty
-sales/	Mar 18	Empty

Everything Waiting To Be Done

Critical / High Priority

!GTM Agencies: Post-send automation — auto-reply within 3 min, demo scheduling (TidyCal)
!GTM Agencies: Email infrastructure — SPF/DKIM/DMARC for sender domains
!Website: Stripe/payment integration for partner payouts
!Website: Email notifications (payouts, approvals, alerts)
!SaaS Partners: Email finding for 4,618 leads (only 0.5% have emails)
!Expos: Fix broken scrapers (Expo West, Sweets & Snacks, Fancy Food, NRA Show)

Medium Priority

☐GTM Agencies: Brand website text scraping (Shopify + WooCommerce stores)
☐GTM Agencies: Full Clutch data integration (500-2,000 more agencies)
✓GTM Agencies: 14,700+ personalized landing pages live
☐GTM Agencies: 3-touch sequence for bounced emails
☐Website: Webhook HMAC signature validation
☐Website: CSV export from portal
☐Website: Theme App Extension (Shopify)
☐Website: Hybrid attribution (cookie + discount code)
☐SaaS Partners: Contact page scraping + outreach automation
☐Expos: Website enrichment on remaining 95% of exhibitors
☐Expos: Consolidate 35 scraper scripts
☐Reviews: Fix G2/Capterra blocking issues
☐Reviews: Integrate Reddit + pricing scrapers into main pipeline
☐LinkedIn: Configure post URL and run first production scrape

Lower Priority

☐Website: Click tracking analytics dashboard
☐Website: Admin audit logging
☐Website: Two-factor authentication
☐Website: API documentation for webhooks
☐Job Boards: Database persistence (currently file-only)
☐Job Boards: Expand Greenhouse/Lever beyond hardcoded companies
☐Reviews: Trend analysis & automated reporting
☐Expos: International events (SIAL, Cosmoprof)
☐Expos: Clean up 44 debug HTML files
☐General: Populate department folders (-design, -growth, etc.)

Cold Email Unit Economics — Inboxkit Phase 1a / 1b

Source doc: gtm-agencies/6_infrastructure/unit_economics.md. Update both when the model changes.

Model inputs (edit to recompute)

Reply rate (%) Positive/replies (%) Download/pos. (%) Activate/dl (%) Rev-gen/active (%) Avg CMR ($/mo) Revenue lift (%) AF take rate (%) Monthly churn (%)

Computed outputs

—

Net close rate

—

Closes/mo (1a, 16,600 sends)

—

Closes/mo (1a+1b, 33,200)

—

AF $/mo per customer

—

LTV per customer

—

Steady-state MRR (1a)

—

Steady-state MRR (1a+1b)

—

Break-even customers/mo

Funnel breakdown

Step	Cautious	Base	Good	Live (your inputs)
Reply rate	1.0%	1.5%	2.0%	—
Positive / replies	25%	30%	40%	—
Downloads app	45%	55%	65%	—
Activates (Shopify OAuth + first program)	40%	45%	55%	—
Revenue-generating	60%	70%	80%	—
Closes/mo (1a)	4.5	13	38	—

Cost structure

Item	Phase 1a	Phase 1a + 1b
Inboxkit subscription	$149/mo (Agency)	$256/mo (Enterprise)
Domain amortization	$18/mo	$36/mo
MillionVerifier + proxies	$10/mo	$15/mo
Total monthly	$180/mo	$310/mo
Annual	$2,160	$3,700

Variable cost per extra email ≈ $0. Automated replies mean fixed-cost infra, no per-send marginal cost.

Infrastructure buildout timeline (Day 0 = domain purchase)

Day range	Milestone	Daily cold volume
0-14	Domain aging — Gmail/M365 penalize <14-day domains	0
15-28	Warmup phase 1 & 2 (5 → 20 warmup/day)	0
29-35	Cold ramp (5 cold/day/inbox)	~255/day
36+	Full Phase 1a volume	765/day

Revenue lags another 6-10 weeks — send → reply → download → activate → first ambassador-driven sale. First meaningful AF revenue: ~week 12-14.

MRR trajectory — Phase 1a only (live)

Month from Day 0	Active customers	MRR	% of ceiling

MRR trajectory — Phase 1a + 1b (1b on at month 4)

Month from Day 0	Active customers	MRR

Three levers to accelerate (in order of impact)

☐Turn on Phase 1b at month ~4. Doubles volume and ceiling. Cuts $80k-MRR crossover from ~month 48 (1a only) to ~month 15-18.
☐Close-rate lift. Each funnel step ~equally leveraged. Halving the largest drop-off lifts ceiling ~20%.
☐Raise avg CMR at activation. $150k → $300k CMR doubles per-customer revenue at same close rate. List quality + premier_plus targeting.

Minimum viable "definitely profitable" thresholds

✓Net close rate ≥ 0.03% (below cautious scenario — comfortable)
✓Avg CMR ≥ $50k/mo ($600k/yr) at activation
✓Churn ≤ 5%/mo (Performance Plan + referral-program retention ~2-3%)

All three inside realistic ranges. Break-even at base CMR ≈ 2-3 customers/mo — ~10× lower than base-case funnel throughput.

Track live once cold sends begin

☐Emails sent, replies received, reply rate
☐Positive replies, download events, activation events, revenue-generating count
☐Funnel rates per step (compare to model)
☐Active customer count, monthly revenue churn %
☐Avg CMR at activation (add to CRM "deal closed" row)
☐MRR & implied ARR; overlay vs model trajectory