Gydr
Newsroom

Newsroom

What we're building, and why.

Stories from the team building Gydr: the problems our customers bring us, how we solve them, and what we believe about AI that speaks for a business.

Follow by RSS

Gydr wire

  1. NOV 08A concierge chatbot that checks orders and fills carts: why sign-in comes first
  2. NOV 04What we believe: nine principles for customer-facing AI
  3. NOV 01After-hours customer inquiries: a customer's 1.30am problem, and our answer
  4. OCT 31Related is not the same as right: why chatbot hallucinations happen
  5. OCT 29Console MCP: run your AI chatbot from Claude, Codex or any MCP client
  6. OCT 27Knowing who you're talking to: signed-in visitors in your AI chatbot

All stories

November 2026

  1. Perspective

    A concierge chatbot that checks orders and fills carts: why sign-in comes first

    A store asked whether a concierge chatbot could check orders and add to cart. Why we start with proper sign-in, and what Gydr 4.2 does for shoppers.

  2. Perspective

    What we believe: nine principles for customer-facing AI

    Nine principles for customer-facing AI that shaped more than fifty Gydr releases, from honesty over helpfulness to keeping each business's data its own.

  3. Use case

    After-hours customer inquiries: a customer's 1.30am problem, and our answer

    A customer's chatbot promised an agent at 1.30am who was not there. How we rebuilt after-hours customer inquiries around tickets and honest offers.

October 2026

  1. Engineering

    Related is not the same as right: why chatbot hallucinations happen

    Why an AI chatbot should say "I don't have that": what causes chatbot hallucinations, and how Gydr keeps answers to what your knowledge actually says.

  2. Announcement

    Console MCP: run your AI chatbot from Claude, Codex or any MCP client

    Console MCP connects Claude, Codex and other MCP clients to your Gydr Console, so your AI assistant can build, test and fix your chatbot on your behalf.

  3. Engineering

    Knowing who you're talking to: signed-in visitors in your AI chatbot

    How Gydr 4.0 recognises signed-in visitors with one token your site already issues, and the rules that keep a chatbot from ever believing the wrong person.

  4. Product

    Gydr 4.0 handoff: a person when it matters, by live chat or ticket

    Gydr 4.0 handoff lets each help desk run live chat, tickets or both, offers a person when a visitor is stuck, and never promises an agent who is not there.

  5. Announcement

    Gydr 4.0: who is asking, who answers, and how it looks

    Gydr 4.0 brings one visitor token for signed-in customers, help desks with live chat or tickets, a Design studio for the chat widget, and Conversation Topics.

  6. Use case

    Turn website enquiries into leads: one business's lost visitors, and our answer

    A customer's chat answered visitors and then lost them. How we planned Data Forms to turn website enquiries into leads, in the visitor's own language.

  7. Use case

    Order questions after checkout: a store's request, and how we answered it

    An online store asked us to handle order questions after checkout. How we planned answers from WooCommerce that keep each customer's order private.

  8. Product

    Product search in the chat, answered from your store

    How Gydr 3.5–3.7 brought product search in the chat to your own store: WooCommerce and Shopify results, real prices, stock status and order lookups.

September 2026

  1. Use case

    Clinic appointment requests: what a clinic asked for, and where we drew the line

    Overnight messages left a front desk sorting clinic appointment requests by hand. How we planned an assistant that collects them for staff to confirm.

  2. Use case

    Nearest branch and opening hours: why we made the chatbot ask first

    A customer's chatbot sent a shopper to a closed outlet. Why answers about the nearest branch and opening hours now start by asking where the customer is.

  3. Product

    Data Forms: collect customer details inside the chat

    Data Forms let your chatbot ask your own questions in each visitor's language, on your website and every channel, and hand the answers back as a table.

  4. Use case

    Run chatbots for your clients under your own name: an agency's request

    If you run chatbots for your clients, another vendor's name keeps getting in the way. How we planned brand, accounts and access for the agencies that asked.

  5. Engineering

    Report a wrong chatbot answer, straight from the conversation

    Gydr 3.2.8 lets you report a wrong chatbot answer from the conversation itself, with everything we need attached, and revise or withdraw it later.

  6. Product

    Keep your WhatsApp Business app number, and add an AI chatbot

    Gydr 3.2 connects the number you already use in the WhatsApp Business app without giving it up, and adds custom permission sets and instant access changes.

  7. Use case

    When a WhatsApp promotion rush outpaces the replies

    A customer's WhatsApp promotion rush outpaced their replies. How we planned one assistant for website and WhatsApp, within WhatsApp's own rules.

August 2026

  1. Engineering

    Fifty languages, one conversation: how we handle mixed-language chats

    How Gydr handles mixed-language chats: it replies in the language the visitor writes, tells Malay from Indonesian, and translates every button, in 50 languages.

  2. Announcement

    Gydr 3.0: one AI chatbot on WhatsApp, Messenger, Instagram and Telegram

    Gydr 3.0 puts the chatbot you already built on WhatsApp, Messenger, Instagram and Telegram: same knowledge, replies shaped per app, agent replies never lost.

  3. Use case

    New hire onboarding questions: what a customer asked us to fix

    A company's new staff kept guessing at answers to new hire onboarding questions. How we planned an assistant that answers from its onboarding pack.

  4. Use case

    Employee policy questions: what one HR team needed, and what we built

    A customer's HR manager fielded employee policy questions all day. How we planned an assistant that answers from the handbook and stays internal.

July 2026

  1. Use case

    IT how-to questions were crowding out an IT team's real work

    A customer's IT team was losing mornings to IT how-to questions. How we planned an assistant that answers from their own guides and hands the rest over.

  2. Use case

    When repetitive customer questions bury a small support team

    One support team asked us for help with repetitive customer questions. What we planned, and how the assistant now answers them from their own content.

June 2026

  1. Announcement

    Gydr 2.0: the right AI model for every message

    Gydr 2.0 brings Dynamic Mode, which picks the AI model for each message, and Plugins that connect your chatbot to your API, Google Sheets and Shopify.

  2. Product

    Bring your own AI model, and put your brand on Gydr

    With Gydr 1.31 you can bring your own AI model from OpenAI, Anthropic, Gemini and more, put your brand on the Console, and test a custom personality.