• Meet Spacefast: a new home for what you build with AI

    Automattic just launched Spacefast. It puts the apps and websites you build with AI online for you.

    Building with AI is easy now. Putting it on the internet is still hard, because you need a host, a database and a domain. Spacefast sets these up for you. Built-in sign-in is coming soon.

    You connect Spacefast to an AI tool you already use, like Claude, ChatGPT or Cursor. You ask it to publish your project, and you get a link back in the chat within seconds.

    What I like:

    • Each publish gets its own permanent link, so you can go back if something breaks.
    • You choose who sees it: password, invite, link or public.
    • You keep your own AI. Spacefast charges for hosting, not AI use, so you don’t pay twice.
    • The AI has to ask before risky changes, like changing your domain settings.

    I’ve already used it for Drum Lab, a practice page for drum rudiments (basic sticking patterns like the paradiddle: RLRR LRLL). I built it for my son, who is learning the drums, and for me to brush up too. It runs on Spacefast with my own subdomain.

    Spacefast is open to everyone, and there’s a free personal plan.

    Read Automattic’s announcement.

  • My colleague Adrián Gómez wrote a post about Odie, the AI assistant behind WordPress.com and WooCommerce support: Meet Odie.

    Back in August I wrote about how Odie decides when to hand a conversation to a person. Adrián covers a different part: how the support team thinks about Odie, and why putting a bot in front of human support was not a decision to make lightly.

    This line stayed with me:

    A response can be technically correct and still miss the mark if it feels cold or disconnected from the person reading it.

    Adrián says this is why the AI Squad exists. It’s a team of Happiness Engineers (Automattic’s support staff) who review Odie’s conversations. They look for wrong or unexpected answers, and they also check the tone, how clear the answer is, and whether it feels supportive.

    He also shares an example about domains. Happiness Engineers kept seeing the same domain requests, and many of them did not need a person. So after testing and adding guardrails (limits on what the bot is allowed to do), Odie can now handle some of those requests directly.

    What I like is the order. The people doing support noticed the pattern first, and the bot got the new ability second.

    It’s a short read, and a good one if you are curious how our support team works alongside AI.

  • I’m speaking at AI Engineer Sydney

    I’m speaking at AI Engineer Sydney this December. My talk is called “Half our handoffs are the bot’s idea.”

    The talk is about part of my work on the Dotcom AI Squad at WordPress.com and how our AI support assistant decides when to hand a conversation to a person. (A handoff is when the assistant passes the chat to one of our Happiness Engineers, the people who provide support at WordPress.com).

    About one in four conversations with the assistant ends with a person. In about half of those, the handoff is the bot’s idea. I wrote about how this works in When WordPress.com Support Knows It’s Time for a Human, and the talk builds on that post.

    AI Engineer Sydney is part of the world’s leading AI engineering conference series, and features two days of practical, production-focused talks from people building AI into real products.

    This link will give you $200 off a conference or leadership ticket, or $100 off streaming:
    https://register.webdirections.org/ai-engineer-sydney-2026?coupon=adelineaieng26syd

    #aiengineersyd

  • I play the drums, and my son has started learning too. I wanted a quick place we could both go to for rudiments. Rudiments are short practice patterns for your hands, and most drumming is built from them. My son could practise along with it, and I could brush up on paradiddles and the other rudiments I don’t play as often. So I made a small website called Drum Lab.

    The Drum Lab homepage drawn like a notebook page: a handwritten headline, Learn a drum pattern, hear it, then play along, with a faint orange drawing of two drumsticks hitting a snare drum, and Easy read and Dark mode buttons at the top.
    The Drum Lab homepage at drum-lab.adelineyaw.blog (it also has a dark mode)

    I didn’t set up any hosting myself. I used Spacefast.

    What is Spacefast?

    Spacefast is a new product from Automattic. On its homepage it calls itself “the publishing layer for your agents.” You make something with an AI agent, such as Claude, ChatGPT, Cursor or VS Code, and Spacefast puts it online with its own web address.

    Spacefast calls each website a “Space” and here’s what I like:

    • Your Space is private until you choose to share it.
    • Spacefast saves every change as a version, so you can go back to an older one in one step.
    • The free plan covers 20 Spaces, which is plenty for hobby projects.

    How I made it

    The whole process was quick and most of it happened inside Claude Cowork.

    1. Connected Spacefast to Claude. Spacefast has a connector for Claude. Once it’s connected, Claude can publish to Spacefast for you.
    2. Described the site I wanted. I asked for paradiddles, the ways you can change them, and drum fills. A paradiddle is a pattern where your hands go right, left, right, right, then left, right, left, left (written RLRR LRLL). I also asked for simple language so my son could follow it easily.
    3. Published it. Spacefast gave the site its own address straight away: adeline-drum-lab.view.fast.
    4. Moved it to my own domain. I added drum-lab.adelineyaw.blog to the Space in the Spacefast dashboard. It’s a subdomain of adelineyaw.blog, a domain I already owned. Spacefast then showed the DNS records it needed. DNS records are the settings that tell a domain which website to load. My domain is registered with WordPress.com, and Claude added the records there for me.

    One step caught me out. adelineyaw.blog was still attached to another WordPress.com site of mine, and WordPress.com wouldn’t let it point anywhere else until I detached it. Once that was sorted, the records went in, Spacefast verified the domain, and the site loaded at drum-lab.adelineyaw.blog

    Making it easier to read

    My son is dyslexic so I asked Claude to make the site easier for him to read. It followed the British Dyslexia Association’s style guide and added an “Easy read” switch at the top of the page.

    When Easy read is on, the site changes to:

    • A plain, clear font (Lexend) with bigger text and more space between letters, words and lines.
    • A plain cream background instead of lined paper, with no drawings behind the text.
    • Straight, simple boxes instead of tilted ones.
    • Orange for the right hand and blue for the left. The guide suggests avoiding red and green together, so I swapped out the red.

    It works in both light and dark mode, and each device remembers the setting so my son only has to turn it on once. Anyone can use it, so if the handwritten style is hard to read, give it a try.

    The Drum Lab homepage with Easy read switched on: a plain cream background, a clear bold sans-serif font with extra spacing, and simple boxes, with no handwriting fonts or background drawing.
    The same page with Easy read switched on

    What I found

    • It was fast. I went from an idea to a live site on my own domain in one sitting.
    • I mostly stayed in one place. Most of the work happened inside Claude Cowork. I only opened the Spacefast dashboard to add my domain.
    • The only snag wasn’t Spacefast. My domain was still attached to an old site. If you’re using a domain you already own, check it isn’t attached to anything else first.
    • It did the job. My son and I now have one place to practise rudiments together, in a style that works for him, and I’ve had a reason to brush up on my paradiddles.

    If you have a small project like this, try building it with Spacefast and your favorite AI agent.

  • Most companies make people work through layers of help before they reach anyone who can fix the problem, so those with the hardest problems try the hardest. Job Thomas, Automattic’s interim Chief Customer Experience Officer, says AI didn’t create that pattern, it made it easier to see, and that a deflection rate is “a measure of how good we are at not talking to customers.” I run response reviews for the WordPress.com support bot. What I check isn’t the number, it’s whether someone got the right help quickly.

    Read his post →

  • Most people have been stuck with a support bot that won’t let them through. I work on the opposite problem: teaching ours to notice when it should step aside.

    I wrote a post on the WordPress.com blog about how it makes that call. Here’s the short version.

    Before the assistant answers anything, it runs a few checks. Have you already asked for a person? Have you asked this same question before? Have you contacted support about it recently? Is this the kind of topic a bot shouldn’t handle on its own? If the answer to any of those is yes, it starts moving you toward a human instead of trying again.

    It also offers a person without being asked. That happens when a problem needs someone to look into an account properly, when the assistant isn’t confident in its own answer, or when what it already suggested didn’t work.

    And if you ask for a person outright, it doesn’t argue with you. It might ask one or two quick questions in case it can fix the problem on the spot, but the offer to hand over stays on the table. You choose.

    The part I care about most is what happens after that. When it hands you over, the whole conversation goes with you — what you asked, what it tried, what didn’t work. Nobody has to explain their problem twice.

    The number that surprised me: about one in four conversations ends up with a person, and in about half of those, the assistant is the one who suggested it. It offers first more often than people expect.

    That’s the goal, really. Solve what it can. When it can’t, get out of the way quickly.

    Read the full post on the WordPress.com blog →

  • It’s been about nine months since I took on the AI Integration Specialist role, and I wanted to share some of the things I’ve learned along the way. These aren’t polished frameworks, they’re real observations from doing the work every day.


    Accuracy is trust

    When an AI bot gives a wrong answer, it doesn’t just fail to help, it actively erodes trust. Users who receive incorrect information from our Bots are less likely to engage with the bot next time. That’s why we treat accuracy as the single most important metric. Getting it right matters more than responding fast.


    Review everything, assume nothing

    One of the biggest shifts in my thinking has been around evaluation. You can’t improve what you don’t measure, and you can’t measure what you don’t look at. Our regular feedback reviews are the foundation of everything we do. Every conversation the bot has is an opportunity to learn.


    Cross-team alignment is non-negotiable

    AI work touches almost every part of our support workflow. If the AI Squad operated in isolation, we’d quickly drift out of sync with what the business actually needs. Regular check-ins and close collaboration with various stakeholders keep us grounded and focused.


    The human side of AI leadership

    Leading a squad isn’t just about data and strategy. It’s about people. Providing constructive feedback, supporting professional growth, and making sure everyone on the team feels valued and heard, that’s where the real impact happens. A motivated, well-supported team does better work, and that shows up in the bot’s performance.


    Still learning

    Nine months in, I’m more convinced than ever that the intersection of AI and customer experience is where some of the most meaningful work is happening. I don’t have all the answers, but I’m learning something new every day. And that’s exactly where I want to be.

  • After nearly a year of working full-time on AI in customer support at WordPress.com, I’ve been thinking a lot about where this is all heading. Not in a hype-cycle, buzzword kind of way, but based on what I’m actually seeing in our data, our conversations, and our day-to-day work.


    AI is getting better at knowing what it doesn’t know

    One of the most important developments I’ve seen is AI getting better at recognizing its own limitations. Early on, bots would confidently give wrong answers. Now, the best systems know when to say “I’m not sure about this, let me connect you with a human.” That self-awareness is a game-changer for user trust.


    The human-AI handoff is the new frontier

    The moment a conversation moves from bot to human support is critical. Do it too early, and you lose the efficiency benefits of AI. Do it too late, and the user is frustrated. Getting that transition smooth and seamless is one of the biggest areas of focus for teams like ours, and I expect it to be a major differentiator for support organizations in the coming years.


    Quality evaluation needs to evolve

    How we evaluate AI support quality today is still largely manual. Our AI squad reviews conversations, scores them against criteria, and identifies patterns. That’s valuable, but as volume scales, we’ll need smarter ways to evaluate at scale. Think automated quality scoring, anomaly detection, and real-time accuracy monitoring. The teams that figure this out first will have a serious advantage.


    AI doesn’t replace empathy

    No matter how good AI gets, customer support will always need a human heart. AI can handle the “what” and the “how” but the “I understand how frustrating this must be” still needs to come from a real person. The future isn’t AI vs. humans. It’s AI and humans working together, each doing what they do best.


    What I’m excited about

    I’m genuinely optimistic about where we’re heading at WordPress.com. The work we’re doing with our Bots, the investments in AI quality, and the cross-team collaboration across divisions and teams, it all points toward a support experience that’s faster, smarter, and more caring. And I’m proud to be part of building that future.

  • One of the most common questions I get from colleagues is: “What exactly does the AI Squad do?” So I thought I’d write a proper overview of how our team works, what we focus on, and why it matters.


    The mission

    The AI Squad exists to make WordPress.com’s AI support bot as accurate, helpful, and ‘human-feeling’ as possible. Our bot handles thousands of support conversations, and our job is to make sure those conversations are genuinely useful for the people on the other end.


    How we work

    Our process revolves around a few key activities:

    // feedback reviews — Squad members regularly review our bot’s responses to evaluate quality. We look at accuracy, tone, completeness, and whether the bot genuinely resolved the user’s issue. These reviews feed directly into improvements.

    // data analysis — I track bot statistics, accuracy rates, CSAT scores, and escalation patterns. When something trends in the wrong direction, we investigate and act.

    // cross-team collaboration — We don’t work in a silo. We partners closely with teams to ensure our priorities align with the broader division strategy.

    // weekly reporting — Every week we publish detailed updates. This keeps everyone informed and creates a transparent record of our progress.


    Why it matters

    AI in customer support isn’t about replacing human support. It’s about handling routine questions quickly and accurately so that our Happiness Engineers can focus on complex, high-value interactions. When our bot gets it right, users get faster help and our Happiness Engineers can spend their time where it counts most.

    That’s the balance we’re always working towards.

  • In July 2025, I stepped into a new role at Automattic: AI Integration Specialist for WordPress.com. After years as a Happiness Engineer helping WordPress.com users directly, I’m now leading a team focused on making AI-powered customer support smarter and more effective.


    Why the shift

    For years, I watched AI tools evolve from simple chatbots into genuinely useful support assistants. At WordPress.com, we introduced our AI support bot and I became increasingly involved in reviewing its responses, analyzing its accuracy, and figuring out how to make it better.

    When the opportunity came, it felt like a natural next step. I already knew the support landscape inside and out. Now I could focus entirely on the intersection of AI and customer experience.


    What the role involves

    My day-to-day includes collaborating with colleagues, on different teams to keep our AI initiatives aligned with the bigger picture. I coordinate bot response reviews with Squad members, publish weekly updates, and dig into our Bot’s trends and patterns to find opportunities for improvement.

    It’s a mix of strategy, data analysis, coordination, and hands-on evaluation. Every day is different, and I love that.


    What’s ahead

    I’ll be sharing more about this journey and what I’m learning about AI in customer support, the challenges we’re tackling, and the wins along the way.