The latest release of the AI plugin broke our integration, we have released a fix for this (3.5.0). As the AI plugin is a beta plugin, this can happen in the early stages. We’re keeping a close eye on developments, and will release updates where necessary as soon as possible.
Burst Statistics now ships with an in-dashboard AI chat. Type a question in plain language, and the assistant pulls the answer straight from your Burst data: live visitors, today’s top page, last month’s referrers, sales by country, the status of your tracker. No CSV exports, no dashboard hopping, no learning a query language.
The chat sits behind a small icon in the Burst dashboard. Click it, ask, get an answer. This guide explains exactly what you need to switch it on and what makes the experience different from asking the same question of Google Analytics.
Why a chat that lives next to your data is different
Burst stores your analytics locally, in your own WordPress database. That single design choice makes the AI chat significantly more useful than a chatbot bolted onto a cloud analytics product.
- No sampling. The assistant queries the same tables the Burst dashboard uses. Whether you have 200 sessions a month or two million, the numbers it reports are the real numbers, not an estimate.
- No data leaves WordPress for the query. The AI provider receives your question and the small, structured result of an ability call. It does not need access to your raw event stream. Compare that with Google Analytics, where pulling data through a third-party AI tool typically means routing your analytics through yet another vendor.
- Direct, structured access. Burst exposes a set of read-only abilities through the WordPress Abilities API. The AI uses those as tools, with schemas it can validate against, so it cannot invent metrics or fabricate fields.
- One context, one conversation. You can chain questions: “How many visitors yesterday?” then “Which page did most of them land on?” then “Compare that page’s bounce to last week.” With Google Analytics, each of those is a new report, a new filter, a new export.
This is the kind of workflow local analytics enables and a remote analytics service makes painful. If you have already moved from Google Analytics to Burst, the chat is what the move was for.
What you need before you start
The chat is built on top of WordPress core’s new AI infrastructure, so the requirements are about getting that infrastructure in place.
- WordPress 7.0 or higher. WordPress 7.0 ships with the AI Client and the connectors UI built into core. On 6.9 you can technically run the same stack with an extra Composer-based package, but 7.0 is the supported path.
- The official AI plugin from wordpress.org/plugins/ai. This plugin bundles the AI Client SDK and the Abilities Explorer UI, and acts as the bridge between Burst’s abilities and your chosen AI provider.
- A provider plugin for the AI service you want to use, with an API key configured. WordPress 7.0 supports three official providers at launch: OpenAI (ChatGPT), Anthropic (Claude), and Google (Gemini). Pick whichever you already have an account with.
- The Abilities API setting enabled in Burst. This is the toggle that registers Burst’s analytics abilities so the AI can actually read your data. See the deep dive in Bring AI to your stats: Burst Statistics now speaks the WordPress Abilities API for the full list of what gets exposed.
Step by step
1. Update WordPress to 7.0 or later
From your WordPress dashboard, go to Dashboard > Updates. If 7.0 is offered, install it. Take a backup first if you have not already. On a managed host this is usually a one-click upgrade.
2. Install the AI plugin
Go to Plugins > Add New, search for “AI” by WordPress.org, and install it. Activate it. The plugin adds a Connectors screen under Settings where AI providers register themselves.
3. Install and configure your AI provider
In settings > AI, you will see a link “manage connectors”. If you click it, you can choose your favorite AI connector, and configure the key. Alternatively, you can go to Plugins > Add New, and install the provider plugin for the service you want to use:
- AI Provider for OpenAI if you want to use ChatGPT models.
- AI Provider for Anthropic if you want to use Claude.
- AI Provider for Google if you want to use Gemini.
Activate the one you picked, then go to Settings > Connectors. The provider you just installed shows up with a field for an API key. Paste in the key from your provider’s account dashboard and save. WordPress now has the credentials it needs to talk to that AI service, and every AI-aware plugin on your site can share those credentials, including Burst.
You can install more than one provider. The Burst chat will use whichever one is configured.
4. Enable the Abilities API in Burst
Open Statistics > Settings in WordPress and switch on Abilities API integration. Save the settings. Burst now registers its read-only abilities so the AI can use them as tools.
This toggle is opt-in by design. If you leave it off, nothing is registered and nothing changes. For a complete overview of what each ability does (live visitors, today summary, full data queries, sales, subscriptions, tracking status, license notices, tasks), see the Abilities API article.
5. Open the chat
Go to the Burst dashboard. A small chat icon now appears in the header. Click it to open the modal, type a question, and send. The first response can take a few seconds: the model reads your question, decides which Burst abilities to call, calls them, and writes the answer in plain language.
Past conversations are stored locally in your browser, so you can switch between chats and pick up where you left off. Each conversation is scoped to your user; nothing is shared with other admins.
What to ask
The chat is at its best when you ask the kind of question you would normally answer by clicking through three or four dashboard screens. A few examples to start with:
- “How many people are on the site right now, and what are they reading?”
- “What were my top 10 pages last week by visitors?”
- “Compare pageviews this month against last month.”
- “Which referrers sent the most traffic in the last 30 days?”
- “Is my tracking healthy? When was the last successful test?”
- “For Burst Pro: how much revenue came from the German market last quarter?”
Follow-up questions work too. After the assistant gives you last week’s top pages, you can ask “what was the bounce on the second one?” without restating the context. That continuity is one of the reasons the chat feels different from a static dashboard.
Privacy, costs, and where the data goes
A few notes on what changes when you turn the chat on.
- What the provider sees is your prompt and the small, structured response of any ability the model calls. For example, a top-10 pages query sends ten rows of page URLs and counts to the model so it can phrase the answer.
- API calls to OpenAI, Anthropic, or Google are billed to your account through the API key you configured. Burst itself does not charge anything for the chat.
- Every ability is read-only and gated behind the same capability check as the Burst dashboard. A user who cannot see the dashboard cannot use the chat to read the same data.
- To switch the chat off, toggle Abilities API integration back off in Burst settings. The icon disappears and no abilities are registered on the next request.
Troubleshooting
If the chat icon does not appear, hover over it: the tooltip will tell you what is missing. The common cases:
- The Abilities API setting is off. Enable it in Burst settings.
- The AI plugin is not installed or activated. Install it from wordpress.org/plugins/ai and activate.
- No provider is configured. Install at least one provider plugin (OpenAI, Anthropic, or Google) and add the API key under Settings > Connectors.
- WordPress is on 6.9 without the supporting package. Update to 7.0 for the supported path.
The bigger picture
For years, the trade-off with privacy-friendly analytics was that you owned your data but interacting with it was less polished than the big cloud platforms. The Abilities API plus WordPress 7’s AI infrastructure changes that. Local analytics now has a native interaction layer that the cloud platforms cannot match without sending your data outside their walls.
Set the chat up once, ask a few questions, and you’ll see what we mean. If there’s a question you wish the assistant could answer but it currently cannot, the answer is almost always “add another ability”. Feedback on which ones to prioritise is welcome through the usual support channels.
