Hay.chat VS PocketPatch

Hay.chat's answer

We're early. We're working with a small group of pilot partners across European e-commerce and SaaS, but we don't publish their names without permission. If you're evaluating Hay, this is the moment to talk to us. You'll get the founders on the call instead of an SDR. And you'll shape what we build next.

Hay.chat's answer

Most AI customer service tools answer questions. They sit on top of your knowledge base, summarise docs, deflect tickets. The customer still waits for a human to do the actual work: process the refund, look up the order, update the account.

Hay does the work. It's built on the Model Context Protocol, an open standard that lets AI tools talk to other systems. So Hay plugs directly into Shopify, WooCommerce, Stripe, Zendesk, HubSpot, and the rest of your stack. When someone asks for a refund, Hay processes it. When they ask where their order is, Hay checks the carrier and replies. No ticket gets queued for a human to close later.

It's also source-available, EU-hosted, and you can read the code. Self-host if you want. Audit how decisions get made. Nothing's hidden.

Hay.chat's answer

A few honest reasons.

Pricing. The industry's converging on roughly $1 per resolution (Intercom Fin, Gorgias, HubSpot Breeze are all in that range), even though the underlying inference costs a few cents. Hay bundles resolutions into flat monthly tiers from €50 to €2,000, with overages between €0.01 and €0.20. Your bill stays predictable as your AI gets better.

EU hosting. We run on infrastructure in the EU. With AI Act enforcement landing in August 2026 and most US-headquartered competitors still routing data through the US, this matters more than it used to.

Transparent code. You can read how Hay makes decisions. You can self-host if you'd rather keep everything in-house. Most alternatives are closed boxes.

Setup. Plain-language playbooks instead of node-based flow builders. You teach Hay the way you'd onboard a new team member.

What we won't claim: that we're cheaper than every competitor, or that nobody else takes actions. Both have changed. But the full combination still hasn't shown up elsewhere: pricing predictability, EU hosting, source-available code, and action-taking AI.

Hay.chat's answer

Mostly European e-commerce companies between 15 and 50 people, doing €2M to €8M in revenue, handling 1,500 to 3,000 support tickets a month. They're usually on Shopify or WooCommerce, often with Zendesk or a shared inbox.

Two or three people on the support team, with a founder or Head of Support making the buying call.

They've usually tried one of two things. A basic chatbot that deflected without resolving, which made customers angrier. Or Intercom or Gorgias at per-resolution pricing, which gave them sticker shock as volume grew.

B2B SaaS teams of similar size also use Hay for billing questions, password resets, and account changes. Our first paying customer is a Norwegian B2B SaaS, so that segment is real even if e-commerce is the bigger market.

Hay.chat's answer

We kept running into the same pattern in customer support. About 80% of every team's tickets are repetitive (where's my order, how do I get a refund, can you reset my password) and 20% are the complex cases that drive loyalty and revenue. The Pareto principle, basically.

The existing tools all attacked the wrong end. They either hid the 80% behind a chatbot that couldn't resolve anything, or they charged enterprise prices that only made sense if you already had enterprise volume. The AI side was mostly black boxes too. Opaque models, mystery pricing, your data going somewhere you couldn't see.

We thought there was a clearer path. Build it on MCP so it could take real actions across the systems support teams already use. Keep the code transparent. Host in the EU. Price it so the bill doesn't punish you for growing.

That's Hay. Still early, still being shaped by the teams using it.

Hay.chat's answer

The stack: Frontend: React with TypeScript, TailwindCSS, Zustand for state. Backend: Node.js with Express, PostgreSQL, Redis for caching and real-time, Bull for job queues. Infrastructure: Docker, Kubernetes, GitHub Actions. EU-hosted on DigitalOcean. Integrations: Built around the Model Context Protocol. 177 actions across nine integrations including Shopify, WooCommerce, Zendesk, Stripe, HubSpot, plus channel connectors for WhatsApp, Instagram, web chat, and email. Models: Hay is model-agnostic. It runs on OpenAI, Anthropic, Mistral, Llama, DeepSeek, and others. No lock-in to one provider. Licence: Source-available under the Hay Community Licence v1.0, structurally similar to Elastic License 2.0 and n8n's Sustainable Use License.