Connectly's AI Recovers Abandoned Carts via WhatsApp Conversations

01 /The Concept

Connectly's AI chatbot Sofia doesn't remind customers to buy – she starts a conversation with them.

The use case is familiar: a shopper browses an online store, adds sunglasses to a cart, and disappears without checking out. Rather than firing a standard "You left something behind" email, Sofia reaches out via WhatsApp a few hours later and asks what the customer was actually looking for – their use case, their price range – then offers two or three specific recommendations. If the customer says the original pick was too expensive, Sofia assesses their purchase history from integrated CRM data and responds in real time: a real discount for a loyal buyer, free shipping for a casual one.

The decision logic is defined by store owners in a visual flow editor – a block diagram specifying what Sofia can offer, to whom, and under what conditions. Sofia improvises within those guardrails during the actual conversation, drawing on data from integrated platforms: Salesforce, HubSpot, Zendesk, Shopify, and others.

Supported channels include WhatsApp, Instagram DMs, Facebook Messenger, and other messaging apps. Shopify merchants can start at $20 per month for a basic configuration; more capable tiers run $80 or $299 per month, with actual billing scaling based on conversation volume. About 200 brands are currently on the platform, including Clarins, Careem, and Toys R Us. Annualized recurring revenue has crossed $3.5M. Connectly has raised $7.85M in its latest round, bringing its total to $17.25M.

02 /Why It Matters

Connectly isn't breaking new technological ground – AI-powered personalization in e-commerce has been standard for years. What the company is addressing is a problem large enough that even an incremental solution generates serious commercial value.

Abandoned carts represent roughly 66% of shopping sessions on average: seven out of ten visitors who spend time selecting products leave without buying. Industry estimates put the annual revenue gap at $18 billion globally for e-commerce merchants – money left on the table from buyers who have already demonstrated clear purchase intent.

Cartloop, [covered back in 2020](/review/jemocii-na-sluzhbe-prodazh), works the same problem via SMS, using human "customer recovery experts" to run the conversations. The model is effective but hard to scale – Cartloop raised $420K total. RetainIQ, [reviewed earlier](/review/denezhnaja-proslojka), automates recovery through multi-step personalized email sequences and raised $2.31M in its first round. Email is a one-way channel, though: the store fires a message and can't respond to what the customer says next.

Connectly claims that two-way WhatsApp conversations convert to purchases 5x more often than single-shot email campaigns. If accurate, the math is straightforward: the same customer base, the same traffic, 30% more revenue – through a better interaction model alone.

03 /Opportunities

Two directions open up from here.

The first is tools specifically targeting cart abandonment recovery. The problem is large enough that any working solution can find customers and attract investment. Connectly's conversational AI approach is timely – the technology for real-time dialogue quality at this level simply wasn't affordable three years ago, and the WhatsApp Business API has made conversational commerce a real channel rather than a novelty.

The second direction is broader: AI chatbots for automating customer interactions across the entire purchase lifecycle, not just cart recovery. Only about 1.6% of customer interactions are currently AI-automated – industry projections put that figure near 10% by 2026. The same conversational depth that enables Connectly could apply to onboarding, support triage, renewal conversations, and reactivation campaigns.

Glystn, [covered in the summer](/review/samyj-bolshoj-rynok-dlja-razgovornogo-marketinga), pushes conversational AI into community management, generating AI-prompted comments and topic seeds to keep discussion channels active. That company raised $4M in its first round while still in closed beta.

The constraint worth taking seriously: conversational AI only generates value if the conversation quality is high enough that users engage naturally with it. The weakest version of this category is a sophisticated keyword-response bot with a chatty wrapper. The strongest version genuinely improves outcomes for both parties – and that's the much harder product to build.