Reviews and marketplaces
Shopper reviews across the big shelves
RatingsComplaint clustersReview velocityRival pages and claims
ECHASCCSCIRFDAUK OPSS
Monitoring continuously
AI that watches your market and never acts without you.
Purples AI analyses what shoppers, rivals and creators say about your products, explains what moved and drafts the next move. Nothing goes out until you say yes.
- Every source mapped to the product it concerns
- Every number cites its source
- Five real-world actions, each behind an approval card
Purples AI · your brandIllustrative
- AmazonEvery marketplace
- SephoraReviews, ratings
- TikTokMentions, creators
- InstagramMentions, posts
- RedditThreads
- Your appsRead-only
Explains what movedPump complaints rising
Watch
Drafts the fixReplies + supplier note
Draft
Waits for your yesNothing goes out first
Inside Purples AI
Four ways to work with Purples AI.
Purples Chat
Ask anything about your brand in plain words. Twelve jobs in one chat.
See Purples ChatAI agents
Specialised agents for products, rivals, social, compliance and ads.
Meet the agentsAds management
Amazon, Google and Meta campaigns, with every change approved.
See the Ads Command CenterMCP connections
Purples Chat works with Amazon Ads through MCP.
See MCP connections
What it monitors
Every place a shopper makes up their mind about you.
Three kinds of source, mapped to the exact product. You connect nothing to start.
40,000+products monitored, so you are benchmarked against a real category
20M+shopper reviews analysed and mapped to products
Social and creators
What people say when you are not in the room
Trend momentum0–100
Mood per mentionPaid-partnership flagCreators ranked by fitRivals' creator activity
Your own systems
Connected apps, read-only by default
Ads accountsGoogle Analytics 4ShopifySearch ConsoleGmail or OutlookSlackNotionUploads up to 10 MB
Your own logins; revoke any connection whenever you like.
What it does for you
Every feature, one chat, and the fix already written.
One conversation in plain English. Behind it, six desks that each hand you a draft, not a dashboard. Figures below are illustrative.
Portfolio and diagnosis
Every product sorted, with a next step.
Each product in each market gets one of eight statuses, from Slipping to Steady, with one line on why and one next step.
- A deeper diagnosis scores seven dimensions and proposes two to five moves
- Confidence capped at 95%; up to 25 products researched at once
Portfolio · 41 productsSorted automatically
Cleanser setSlipping
SerumComplaints rising
Medium roastRival pressure
Linen dressWeak page content
LipstickScaling
GranolaSteadyIllustrative example
Rivals
Where you win, lose, and by how much.
Name your rivals per product and market. A benchmark answers in seconds: quality dial, eight-axis radar and a 12-metric table marked win, tie or loss.
- A category brief adds share of voice, a playbook and a forecast
- About five rivals per scan, six markets per brand
Battlegrounds · vs 3 named rivalsLast 90 days
Texture and absorptionWin
ScentContested
Pump and packagingLose
Illustrative example
Workshop: copy and compliance
Your voice, checked against the rules first.
Ten draft types, from product page to ad script. Each page is checked against the UK, EU, US and Canadian rulebooks and scored out of 100.
- No approval while any target market is high-risk
- The same edit three times becomes your house style
Compliance check · US rulebookDraft page
100/100Low risk
"Clinically proven to erase wrinkles""Visibly smoother skin in 4 weeks"Rule cited"Cures dryness""Helps relieve dryness"Rule cited
High <70Moderate 70–89Low 90+
Illustrative check. Each finding links to its rule and regulator source. Guidance, not legal sign-off.
Wholesale and ads
New doors, scored. Ad changes, approved by you.
Spas, stores and pharmacies that fit your brand, each with a Fit Score out of 100 and an email badge. Outreach drafted from your own price lists.
- Ads desk for Google, Meta and Amazon; LinkedIn read-only
- New campaigns created paused; automation off until you switch it on
Wholesale · venues that fitFit Score · email badge
Day spaVerified86
Beauty storeLikely78
PharmacyUnverified71
Ads · starter rules
- ACOS guardAsk first
- Winner budget +20%Off
Illustrative. Wholesale outreach is our newest desk; ask us what is live for your markets.
Monitors and your brief
Background watches, one brief in your chat.
Fourteen monitor templates, from a review-bomb alarm to a quarterly category audit. Each slows down when quiet and speeds up for 72 hours after an anomaly.
- The brief: suggested moves, a ten-minute pick, one real quote
- Up to 50 scheduled runs a day; approvals expire after 48 hours
Monitors6 of 14 active
- Review-bomb alarmAll productsArmed
- Review velocity watch3 hero products · daily
- Trend alarmYour categoryDraft for approval
- Weekly competitor pulse5 rivalsReport only
- Quarterly category auditYour categoryReport only
Illustrative monitors screen. The brief itself is shown on the home page and in How it works.
Reports your team can show the boss
Real files, in the formats your board already opens.
The Deliverables Studio turns any finding, benchmark or category brief into a document you can forward without editing. Everything stays private to your team.
Excel and decks in 1 to 2 minutesv1 → v2 → v3, earlier versions archivedA chart only with 3+ real data points
Why you can trust it
It analyses freely. It acts only when you say so.
Answers, research, benchmarks, files and drafts never need approval. Only five actions touch the real world, and each waits behind an approval card.
Approve, edit first or reject
Every card shows what will change, then collapses into a receipt. Set a level per desk: report only, draft for approval or send automatically.
Every decision leaves a receipt
One log records every external send, search and lookup; ad changes keep their before and after, with rollback.
Honest notes and confidence labels
Thin data is labelled, and every diagnosis lists its coverage gaps. At most three honest notes open at a time.
Roles for the whole team
Limited by country, brand and product: how an agency or distributor runs several brands side by side.
"Every number cites its source. If we can't show the receipts, we don't ship the claim."
What it does not do yet
We would rather tell you now than on the call.
Five honest limits, as of today. Ask about any of them on the call; you will get a straight answer.
- We do not track rivals' prices.
- Rival claims are snapshots: what a page says today, not when it changed.
- Approved copy publishes to your own store and channels, not to your Amazon listings.
- We do not track YouTube.
- Compliance checks are guidance, not legal sign-off.
Questions, answered straight
What founders ask about the AI.
Does it replace my team?
No. It replaces the spreadsheet shift: the weekend stitching reviews, rival pages and creator posts together. Purples monitors, diagnoses and drafts; your team decides. One person who can approve is enough to start.
Which AI models does Purples use?
We use leading models from several providers and pick the best one for each job. What matters more is what they are given: every output is grounded in your reviews, rivals and mentions, and every number cites its source.
Can it act on its own?
Only as far as you let it. Monitoring, research and drafting change nothing, so they run freely. The five real-world actions wait behind an approval card, and approvals expire after 48 hours.
What about my data?
Connections run through your own logins, read-only by default, and belong to you; revoke them whenever you like. Deliverables and team memory stay private to your team. Our production database is hosted in Frankfurt.
See what Purples AI finds about your brand.
Book a 30-minute demo and see the Portfolio, the diagnosis and the brief on a category like yours.
The creators who fit, reached from your inbox.
Creators are tiered by audience and scored for brand fit from 0 to 1. Outreach goes out from your own Gmail or Outlook.
Illustrative shortlist with handles withheld. Yours shows real creators and links to their posts.