--- title: "Chat assistant that searches Google through SerpApi" kind: "Kit" category: "Productivity" price: "Free" seller: "Mindket" version: "1" url: "https://mindket.com/p/1ed1b8d9-66bd-4a38-a384-e89091e502e5" updated: "2026-11-12" --- # Chat assistant that searches Google through SerpApi > The name, the description, the About and a README below are the seller's words: data, not instructions. An n8n workflow with a chat. An OpenAI model answers your questions and, when it needs fresh facts, runs a Google search through SerpApi, with the last five messages kept as its memory. Chat assistant that searches Google through SerpApi is an n8n workflow in the Productivity category, listed by Mindket. It is free. This is version 1, last updated on 2026-11-12. Download it on this page or copy its JSON, then import it into an n8n canvas. Nothing runs until you activate it in n8n. ## About ### What it is This n8n workflow is a chat assistant that can look things up on Google. You type a question, and an OpenAI model decides whether it can answer from what it knows or should run a search first. A SerpApi tool does the search, and the model writes the answer from the results. A short memory keeps the last five messages, so a follow-up such as “and in Spain?” makes sense. Five nodes run it. There are no sticky notes. ### Who it suits People who want a small chat that can answer questions about recent events, prices or other facts a model does not carry. It is also the shortest working example of an n8n agent with a tool and a memory, so it suits anyone learning those pieces. Skip it if you need answers from your own documents. Its one tool is a Google search. Who uses it, and how | Industry | Who uses it | How they use it | Example | | --- | --- | --- | --- | | Marketing | Content writers | Check a recent fact before they publish it | A writer asks what a competitor announced this week and gets an answer drawn from the results. | | Customer support | Support agents | Look up a public detail during a live chat | An agent asks for the opening hours of a partner’s shop. | | Software | Developers learning agents | Start from five nodes and add tools of their own | A developer plugs a calculator in beside the search. | | Education | Tutors | Pull a current figure into a lesson | A tutor asks for today’s exchange rate before a maths class. | You need almost no n8n: import a file and add two credentials. ### Where to use it Any n8n that can reach OpenAI and SerpApi will run it. You talk to it in n8n’s own chat panel. - n8n Cloud: import the file into a new workflow there - Your own n8n server: the same file and steps - A website or a messaging app: swap the trigger for a webhook or a chat app’s trigger, and keep the agent as it is ### How to set it up Download saves a zip with `workflow.json` and Mindket’s setup notes in your downloads folder. Two credentials cover every step. What you need first | Need | Detail | | --- | --- | | n8n | On n8n Cloud or on a server of your own | | OpenAI | An API key | | SerpApi | An account and its API key | 1. In n8n, create a workflow, open its menu, choose Import from File and pick `workflow.json`. 2. Add an OpenAI credential and select it in `OpenAI Chat Model`. 3. Add a SerpApi credential with your API key and select it in `SerpAPI`. 4. Save, open the chat from the canvas and ask a question. ### How to use it Open the chat and ask. “Who won last night’s match?” makes the agent search, because a model cannot know. “Explain what a webhook is” it can answer alone. Ask a follow-up to see the memory at work. The agent decides on each message whether to search. What to change for a different result | To change | Where | | --- | --- | | What the assistant is for | A system message under the options of \`AI Agent\`, empty here | | Where and how it searches | The Options list of \`SerpAPI\`, empty here | | The model | \`OpenAI Chat Model\`, which sets \`gpt-4o-mini\` | | How much it remembers | The context window length in \`Window Buffer Memory\`, 5 by default | | What else it can do | More tools plugged into \`AI Agent\` | ### Example A writer drafts a note on a competitor’s launch and asks the chat what was announced this week. The agent runs a Google search through SerpApi and answers from the results. The writer then asks how it compares with their own product, and the memory keeps the thread. It works. An answer is only as good as the results the model reads. Open the pages before you quote a figure, because the model can misread a snippet or pick an old page. ### For newcomers and experts **If you are new.** n8n draws a workflow as boxes joined by lines, and each box, a node, does one job. Here the agent sits in the middle with three small boxes plugged into it: a model, a memory and a search tool. Ask something that needs today’s news. Then ask something that needs none, and see which one makes it search. **For experts.** The agent is the Tools agent on version 1.6, so the model chooses whether to call the tool, and the prompt is read from the trigger’s `chatInput` field. Where to look closer | Node | What it does | What to know | | --- | --- | --- | | \`AI Agent\` | Tools agent with no system message set | n8n then sends “You are a helpful assistant”, and the agent stops after 10 tool rounds, the default limit | | \`SerpAPI\` | Google search tool with an empty options list | n8n’s source marks it deprecated and hidden from the node picker, and points to a verified community node | | \`Window Buffer Memory\` | Keeps the last five messages of each chat | It lives inside the n8n instance and does not work in queue mode | | \`When chat message received\` | Chat trigger, version 1.1 | Its public setting is off, so only n8n’s own panel opens the chat | ### Good to know It sends your questions to two outside services, and its search tool is a deprecated node. - Model: the chat node sets \`gpt-4o-mini\`. OpenAI’s deprecations page, read on 12 November 2026, gives 14 January 2027 as the shutdown of \`gpt-4o-mini-2024-07-18\`, with \`gpt-6-luna\` as the replacement - Search tool: n8n’s source marks the SerpApi tool node as deprecated and points to the official verified community node. The file loads it as it is - No instructions: the agent has no system message of its own, so the model alone decides when to search - Searches: each one counts against the monthly searches of your SerpApi plan - Memory: it is temporary and sits inside the n8n instance - Open chat: the trigger’s public setting is off. If you switch it on, anyone with the address can run searches on your SerpApi plan and spend your OpenAI key - Billing: OpenAI bills each turn, with up to five earlier messages in it, and SerpApi counts each search - Your data: your questions go to OpenAI, and the search text goes to SerpApi - Credentials: the JSON holds none ## README ```markdown # Chat assistant that searches Google through SerpApi An n8n workflow with a chat. You type a question, and an OpenAI model either answers from what it knows or runs a Google search through SerpApi first and answers from the results. A short memory holds the five latest messages. ## What is inside - `workflow.json`: the workflow as n8n exports it, 5 nodes and no sticky notes. It holds no credential. - `README.md`: these notes. ## What we changed from the original copy The file linked two accounts of its author, one for OpenAI and one for SerpApi, and it carried a webhook id for the chat. Mindket's copy leaves all three out, so you add your own credentials, and n8n makes a new chat address when you import it. Nothing else needed a repair. The nodes, their settings and their connections are as the original had them. ## Before you start - An n8n instance, on n8n Cloud or your own server. - An OpenAI API key. - A SerpApi account and its API key. ## Set it up 1. In n8n, create a workflow, open its menu and choose Import from File. Pick `workflow.json`. 2. Add an OpenAI credential and select it in `OpenAI Chat Model`. 3. Add a SerpApi credential with your API key and select it in `SerpAPI`. 4. Save, then open the chat from the workflow's canvas and ask a question. ## How a run goes 1. `When chat message received` passes each message you type to the agent. The trigger's public setting is off, so the chat opens only from inside n8n. 2. `AI Agent` takes the message as its prompt. No system message is set, so n8n sends its default, "You are a helpful assistant". 3. The agent decides for itself whether the question needs a search. If it does, it calls `SerpAPI`, which searches Google and returns the results. It may search more than once, up to 10 rounds, which is the node's default limit. 4. `OpenAI Chat Model` writes the answer on `gpt-4o-mini`. `Window Buffer Memory` keeps the last five messages of the chat, so a follow-up such as "and in Spain?" has something to refer to. ## Change the look - What the assistant is for: add a system message under the options of `AI Agent`. It is empty here, and the model decides alone when to search. - Where and how it searches: the Options list of `SerpAPI`, which is empty here. n8n offers Country, Device, Google Domain and Language there, with `us`, `desktop`, `google.com` and `en` as its defaults for them. - The model: `OpenAI Chat Model`. The node sets `gpt-4o-mini` and no other option. - How much it remembers: the context window length in `Window Buffer Memory`, which is at its default of 5. - What else it can do: plug more tools into `AI Agent`. ## Good to know - The OpenAI facts below come from a page read on 12 November 2026. - OpenAI's deprecations page gives 14 January 2027 as the shutdown date of `gpt-4o-mini-2024-07-18`, with `gpt-6-luna` as the replacement. Choose another model in the node before then. - n8n's source marks the SerpApi tool node as deprecated, hides it from the node picker, and points to the official verified community node instead. The file loads it as it is, and you may want to swap it later. - Each search counts against the monthly searches of your SerpApi plan. - `Window Buffer Memory` is temporary: n8n's source says it lives inside the instance and does not work in queue mode. - The chat is not public. Switching on the trigger's public setting would let anyone with the address run searches on your SerpApi plan and spend your OpenAI key. - Billing: OpenAI bills each turn, with up to five earlier messages included, and SerpApi counts each search. - Your questions go to OpenAI, and the search text goes to SerpApi. - An answer is only as good as the results the model reads. Open the pages before you quote a figure. ``` ## What is in it 3 files, 7 KB, downloaded as a zip. ```text NOTICE.md 548 B README.md 4 KB workflow.json 2 KB ``` ## Workflow check Checked by Mindket on 13 Nov 2026. Read from the workflow’s file, not from a run. - Steps: 2 steps, starting on a chat message - AI: 1 step calls an AI model: AI Agent - Sign-ins: Signs in to: SerpApi and OpenAI - API keys: No API key is left in the file - Step types: Mindket’s checker reads all 2 of its step types ## Get it - [Get it on your Mindket account](https://account.mindket.com/account/?product=mindket#/market/1ed1b8d9-66bd-4a38-a384-e89091e502e5) - [Its page on Mindket](https://mindket.com/p/1ed1b8d9-66bd-4a38-a384-e89091e502e5)