Experimentation intelligence, inside your AI assistant

See what the market tested. And what it kept.

ABSee gives your AI assistant 7,500+ real experiments from 650+ brands — the page before, the version tested, and whether it was shipped or rolled back. So every hypothesis comes with evidence behind it.

Get a demo Connect your AI
Works withClaudeChatGPTGeminiconnects in seconds

7,500+ experiments from 650+ brands, including

Mobile size-selection drop-off
+ABSeeOpus 5↑
Try another question

Research finds the problem. ABSee brings the evidence for the fix.

How it fits your process
Your clickstream · or your own research

Connect your analytics and ask where it leaks. Or bring the research you've already done.

DesktopMobile
Product pageSize pickedAdd to bagCheckoutPurchase
WhereSize selection, mobile45% of mobile sessions exit here vs 27% on desktop
WhoNew visitors · items over £80Footwear, mostly
Written by you
Fit uncertainty is making mobile shoppers leave at size selection.
Fit uncertainty?Size availability?Size chart hard to use?

ABSee doesn't replace your research or your judgement. Your team knows your customers — the hypothesis stays yours.

ABSee

Real experiments on sizing from other brands. The closest four, and what happened to each.

UX principleReduce uncertainty at the moment of choice. The changes that stuck made the size step itself clearer; adding links around it did not.
You decide

Rank by how big the leak is and how strong the evidence is. Evidence for and against each idea, side by side — before you spend a sprint on it.

#Your hypothesisLeakEvidenceEffort
1Surprise delivery cost causes cart exits22% · allLow
2Fit uncertainty drives mobile size exits45% · mobileLow
3Too many colourways overwhelm the choice9% · mobileMed

Funnel and hypotheses are illustrative. Screenshots are real experiments from ABSee's library.

Set up in a minute

Copy. Paste. Start asking.

No dashboard to learn, no integration project. ABSee plugs into Claude, ChatGPT or Gemini — whichever your team already uses.

01
Copy your ABSee linkOne line. Nothing to install.
02
Paste it into your AI assistantIt asks you to sign in to ABSee.
03
Start asking"What's been tested against fit uncertainty?"

            
    
          
    
          
    ✓
    ABSee is connectedYour AI assistant can now search experiments, outcomes and patterns.
    
        
    
        
    Placeholder address — the live connection isn't built yet.
    
      
    
    
    
  
    

    




    
    
    
      Real pages, real outcomes
      
    
        
    Not case studies. What brands actually ran.
    
        
    Every experiment is captured from the live site — the page shoppers saw, the version being tested, and whether the brand shipped it or rolled it back.
    
      
    
      
    
    
    

    



    

    
  
    
    Ask ABSee
    
    Ask the questions you already ask. Get answers backed by real experiments.
    
    
    
      
    
      
    
    
    
  
    

    

    


    
    
    
      The difference
      
    Generic AI gives you best practice. ABSee adds the evidence.
    
      
    "Should we add a free-shipping progress bar to the cart?"
    
      
    
        
    
          
    Without ABSeeAI on its ownGeneric best practice
    
          Free-shipping progress bars are a proven best practice. They encourage shoppers to add more to reach the threshold, which can increase average order value.
          No outcomes · No examples · Could apply to any site
        
    
        
    
          
    With ABSeeYour AI + Real experiments, real outcomes
    
          
    It depends where it goes. R.M.Williams kept a free-shipping tracker on the full cart page. Aéropostale rolled back the same idea in the mini-cart.
    
          
    
        
    
      
    
    
    

    



    
  
    
    
    
      Coming next · Simulation
      
    Run your idea past a million shoppers. Before you build it.
    
      
    
        
    
          1,000,000
          
    simulated shoppers per run — built from your real customers and calibrated on real experiment outcomes.
    
          
    Not AI role-play. ABSee's own shopper model, trained on how real visitors behave.
    
          
    Likely behaviour differencesRisksConfidence range
    
        
    
        
    
          
          
    A · Control
    
          
    B · Variant
    
        
    
      
    
      
    It won't pick your winner — it shows which ideas have earned a real test. Concept.
    
    
    
  
    

    



    
  
    
    Your data, your edge
    
    Market evidence built in. Connect your own data to find the leaks too.
    
    
    
      
      
    
        
    
          Your clickstream · optional
          
    GA4BigQueryContentsquareAmplitude
    
          Finds where sessions drop, on which device, for which shoppers.
        
    
        
    
          Your past test results · optional
          
    WingifyKameleoonOptimizelyDYDynamic Yield
    
          Answers "Have we tested this before?" Private to your team.
        
    
      
    
      
    
        
        
    
          
          Market evidence
          7,500+ real experiments from 650+ brands — before & after screenshots, outcomes and the UX thinking behind them.
          Always on
        
    
      
    
      
    
        Your AI assistant
        Claude, ChatGPT or Gemini
        
      
          
    • Where does our funnel leak?
      • 
          
    • What have other brands tried against it?
      • 
          
    • Have we tested this before?
      • 
        
    
      
    
    
    
    
    Your data stays private to your team — it sharpens your answers and is never shared or added to the market evidence.
    
  
    

    



    

    
  
    
    Plans
    
    From what others tested to evidence for your hypothesis.
    
    
    
      
    
        Information
        
    Library
    
        
    What have others tested?
    
        
    • Search real experiments
    • Before & after screenshots
    • Brands and categories
    • Patterns across the market
    
        Get a demo
      
    
      
    
        Decisions
        
    Intelligence
    
        
    What's the evidence for our hypothesis?
    
        
    • Everything in Library
    • Outcomes: shipped or rolled back
    • Evidence for and against
    • UX principles behind each pattern
    • Your clickstream and past tests
    • Competitor tracking
    • Prioritisation
    
        Get a demo
      
    
      
    
        Confidence · coming later
        
    Simulation
    
        
    Has this idea earned a real test?
    
        
    • Control vs variant, before you build
    • Likely behaviour differences
    • Risks and confidence range
    
      
    
    
    
  
    

    

    


    
  
  
    
    
    Ask ABSee
    
    
    Evidence behind every hypothesis. Bring the leak you've found — leave with the evidence for how to fix it.
    
    
    
      Get a demo
      Connect your AI