Gemini + Gmail: AI research reports emailed ready to share
You open a “quick research” task and suddenly you have 20 tabs, half-finished notes, and that sinking feeling you’ll have to explain your sources later. Then someone asks, “Can you send me a clean brief by EOD?” and you’re back to copy-paste.
This Gemini Gmail reports setup hits marketing managers and analysts first, honestly. But content leads doing competitive scans feel it too. The outcome is simple: one prompt becomes a structured Japanese HTML report, emailed and ready to forward.
You’ll see what the workflow produces, why it’s reliable, and how you can tweak it without turning this into a “mini engineering project.”
How This Automation Works
See how this solves the problem:
n8n Workflow Template: Gemini + Gmail: AI research reports emailed ready to share
flowchart LR
subgraph sg0["When clicking ‘Execute workflow’ Flow"]
direction LR
n0@{ icon: "mdi:play-circle", form: "rounded", label: "When clicking ‘Execute workf..", pos: "b", h: 48 }
n1@{ icon: "mdi:robot", form: "rounded", label: "Query Generator", pos: "b", h: 48 }
n2@{ icon: "mdi:brain", form: "rounded", label: "Google Gemini Chat Model", pos: "b", h: 48 }
n3@{ icon: "mdi:robot", form: "rounded", label: "Structured Output Parser", pos: "b", h: 48 }
n4@{ icon: "mdi:brain", form: "rounded", label: "Google Gemini Chat Model1", pos: "b", h: 48 }
n5@{ icon: "mdi:cog", form: "rounded", label: "Tavily_Search_Tool", pos: "b", h: 48 }
n6@{ icon: "mdi:brain", form: "rounded", label: "Google Gemini Chat Model2", pos: "b", h: 48 }
n7@{ icon: "mdi:message-outline", form: "rounded", label: "Send a message", pos: "b", h: 48 }
n8@{ icon: "mdi:robot", form: "rounded", label: "Research Agent", pos: "b", h: 48 }
n9@{ icon: "mdi:robot", form: "rounded", label: "Report Agent", pos: "b", h: 48 }
n10@{ icon: "mdi:swap-vertical", form: "rounded", label: "query", pos: "b", h: 48 }
n10 --> n1
n9 --> n7
n8 --> n9
n1 --> n8
n5 -.-> n8
n2 -.-> n1
n3 -.-> n1
n4 -.-> n8
n6 -.-> n9
n0 --> n10
end
%% Styling
classDef trigger fill:#e8f5e9,stroke:#388e3c,stroke-width:2px
classDef ai fill:#e3f2fd,stroke:#1976d2,stroke-width:2px
classDef aiModel fill:#e8eaf6,stroke:#3f51b5,stroke-width:2px
classDef decision fill:#fff8e1,stroke:#f9a825,stroke-width:2px
classDef database fill:#fce4ec,stroke:#c2185b,stroke-width:2px
classDef api fill:#fff3e0,stroke:#e65100,stroke-width:2px
classDef code fill:#f3e5f5,stroke:#7b1fa2,stroke-width:2px
classDef disabled stroke-dasharray: 5 5,opacity: 0.5
class n0 trigger
class n1,n3,n8,n9 ai
class n2,n4,n6 aiModel
The Challenge: Turning messy research into a shareable report
Manual research isn’t just “time-consuming.” It’s chaotic. You search, skim, save links, forget which claim came from where, then rewrite everything into something your team will actually read. If you work in Japanese but your sources are mixed (Japanese and English), it gets worse because you’re translating, standardizing terminology, and trying to keep the nuance intact. The end result is often late, inconsistent, and hard to reuse for the next project.
None of these alone is the problem. Together, they are.
- You lose about 1–2 hours per report just opening, scanning, and re-scanning sources.
- People can’t trust the brief because citations are scattered across chat messages, docs, and browser bookmarks.
- Two team members researching the same topic will produce totally different structures, which makes comparison work painful.
- “Send it by email” becomes a separate task, so reports often live in drafts until someone remembers.
The Fix: One prompt → multi-angle research → Japanese HTML brief → Gmail delivery
This workflow turns a single research request into a ready-to-share email report. You start by entering a topic (the default example is “n8nとdifyの違い”), and Gemini generates several optimized search queries so the research isn’t one-dimensional. Tavily then runs those searches with deeper parameters and returns summarized findings plus relevant sources. From there, Gemini and the agent steps consolidate what came back, remove duplicates, and assemble a structured HTML report in Japanese that reads like something a human analyst would send. Finally, Gmail sends the HTML report to your chosen recipient(s), so distribution happens automatically instead of “when you get a minute.”
The workflow begins with prompt setup, then shifts into query generation and research orchestration using Gemini plus Tavily. After synthesis and formatting, the final output is delivered as an email you can forward to a client or drop into an internal thread without reworking the content.
What Changes: Before vs. After
| What This Eliminates | Impact You’ll See |
|---|---|
|
|
Real-World Impact
Say you create two research briefs a week for competitors and tooling comparisons. Manually, you might spend about 20 minutes per search angle, and you’ll usually need at least 5 angles, plus another hour to write it up and format it, so roughly 3 hours per brief. With this workflow, entering the topic takes 2 minutes, then you wait for the AI research and synthesis to finish (often under 20 minutes), and you spend about 10 minutes reviewing before forwarding the email. That’s about 2 hours back per brief, without cutting corners.
Requirements
- n8n instance (try n8n Cloud free)
- Self-hosting option if you prefer (Hostinger works well)
- Google Gemini API for query + report generation
- Tavily API to run deep web research
- Gmail account to send the final HTML email
Skill level: Intermediate. You’ll connect accounts, add API keys, and edit a few prompts safely.
Need help implementing this? Talk to an automation expert (free 15-minute consultation).
The Workflow Flow
Prompt is provided. You run the workflow manually and set the research topic in the “Input Prompt Setup” step (this can be swapped later for a form or a sheet trigger).
Gemini expands the topic into better searches. The “Search Query Builder” agent uses Gemini to generate three optimized queries so you cover different angles instead of repeating the same results.
Tavily gathers sources and summaries. The “Research Orchestrator” routes those queries into Tavily’s research tool, pulling back relevant pages and AI-assisted answers that are easier to synthesize than raw links.
A Japanese HTML report is composed and emailed. The “Report Composer” uses Gemini again to dedupe, structure, and format the brief, then the Gmail step sends it as an HTML email to your chosen address.
You can easily modify the report structure to emphasize competitors, pricing, feature comparison, or risks based on your needs. See the full implementation guide below for customization options.
Step-by-Step Implementation Guide
Step 1: Configure the Manual Trigger
This workflow starts on demand, so you’ll configure the manual trigger to run the research report generator when needed.
- Add the Manual Execution Start node as the trigger.
- Connect Manual Execution Start to Input Prompt Setup to match the execution flow.
Step 2: Connect the Input Prompt
Define the research topic in the setup node so downstream AI agents have a clear query to work from.
- Open Input Prompt Setup and add a string field named query.
- Set the query value to
n8nとdifyの違い(or replace with your target topic). - Ensure Input Prompt Setup connects to Search Query Builder.
Step 3: Set Up the Query Generation Agent
The query builder uses Gemini to turn the input into three optimized search queries and parses the output into a structured object.
- Configure Search Query Builder with the prompt text
=質問に対して有力なクエリを3つ作成してください。and keep Prompt Type set to define. - Attach Gemini Chat Core as the language model for Search Query Builder and set Model Name to
models/gemini-2.5-flash. - Attach Structured Result Parser as the output parser with Schema Type set to manual and the schema set to
{"type":"object","properties":{"query1":"string","query2":"string","query3":"string"}}.
Credential Required: Connect your googlePalmApi credentials in Gemini Chat Core. The parser is a sub-node of Search Query Builder, so credentials are added to the language model, not the parser.
Step 4: Configure Research and Report Composition
This phase runs the web research using Tavily and then composes the final HTML report using Gemini.
- Open Research Orchestrator and keep the prompt text set to
受けとった3つのqueryを一つずつTavily_Search_Toolにqueryとして渡して3つのqueryに関して調査を実行し検索クエリと調査結果を省略せずにアウトプットしてください。。. - Attach Gemini Chat Assistant as the language model for Research Orchestrator with Model Name
models/gemini-2.5-flash. - Attach Tavily Research Tool as the AI tool with Query set to
={{ /*n8n-auto-generated-fromAI-override*/ $fromAI('Query', ``, 'string') }}and optionsmax_results: 10,search_depth: advanced,include_answer: advanced. - Configure Report Composer to output HTML using the prompt text
調査結果を統合し、ユーザーの質問に対してわかりやすい構造化された包括的な最終レポートをHTML形式でアウトプットしてください。. - Attach Gemini Report Model as the language model for Report Composer with Model Name
models/gemini-2.5-proand Max Output Tokens set to6000.
Credential Required: Connect your googlePalmApi credentials in Gemini Chat Assistant and Gemini Report Model.
Credential Required: Connect your Tavily credentials in the parent node Research Orchestrator (the Tavily Research Tool is an AI tool sub-node).
Step 5: Configure Report Delivery
The report is emailed using Gmail with the report HTML and the original query as the subject.
- Open Email Report Dispatch and set Message to
={{ $json.output }}. - Set Subject to
={{ $('Input Prompt Setup').item.json.query }}. - Ensure Report Composer is connected to Email Report Dispatch to finalize the workflow.
Credential Required: Connect your Gmail credentials in Email Report Dispatch.
Step 6: Test and Activate Your Workflow
Run a manual test to verify the full research and report pipeline works end-to-end before turning it on for production use.
- Click Execute Workflow to run Manual Execution Start.
- Confirm that Search Query Builder outputs structured
query1,query2, andquery3fields. - Verify Report Composer produces an HTML report in output.
- Check your inbox to confirm Email Report Dispatch sends the report with the subject set to the input query.
- Toggle the workflow to Active when you’re ready to use it regularly.
Watch Out For
- Google Gemini credentials can expire or be tied to the wrong GCP project. If things break, check the Gemini API key and billing status in Google Cloud Console first.
- If you’re using Wait nodes or external rendering, processing times vary. Bump up the wait duration if downstream nodes fail on empty responses.
- Default prompts in AI nodes are generic. Add your brand voice early or you’ll be editing outputs forever.
Common Questions
Usually in about 30 minutes if you already have the API keys.
Yes. You won’t write code, but you will connect accounts and paste API keys into the right credential fields.
Yes. n8n has a free self-hosted option and a free trial on n8n Cloud. Cloud plans start at $20/month for higher volume. You’ll also need to factor in Gemini API usage plus Tavily search credits (cost depends on how deep you search).
Two options: n8n Cloud (managed, easiest setup) or self-hosting on a VPS. For self-hosting, Hostinger VPS is affordable and handles n8n well. Self-hosting gives you unlimited executions but requires basic server management.
Start by editing the prompt in “Input Prompt Setup” so it asks for the exact structure you need (for example: competitors, pricing, positioning, risks, and a recommendation). Then adjust the “Search Query Builder” instructions to generate queries that match your industry language. Finally, tune the “Report Composer” system message to change tone, headings, or how aggressively it removes duplicates. If you want the report to land somewhere else, you can also swap the Gmail send step for a Google Docs or Google Sheets destination.
Most of the time it’s an auth issue: the Gmail account was disconnected, permissions weren’t granted, or the wrong Google account is selected in n8n credentials. Reconnect Gmail in n8n and confirm the “Send a message” step is using that credential. If the email sends but looks broken, it’s usually HTML formatting in the report output, not Gmail itself.
On n8n Cloud Starter, you’re generally fine for low-to-mid volume reporting, and you can upgrade if you need more executions. If you self-host, there’s no execution cap from n8n, but your server and API rate limits will be the real bottlenecks. Practically, most teams run a handful of briefs per day without issues. If you push high volume, you’ll want to watch Tavily quotas and Gmail sending limits.
Often, yes, because this flow needs multi-step AI orchestration (query generation, research, synthesis, formatting) and n8n handles branching and agent-style logic more comfortably. Zapier and Make can work, but the moment you add “run three searches, merge results, dedupe, then format HTML,” you’ll feel the friction and cost. Another factor is hosting: self-hosted n8n is appealing if you don’t want every run billed as a premium task. That said, if all you need is “take a prompt and email it,” Zapier may be faster to set up. Talk to an automation expert if you want a quick recommendation.
You end up with research you can actually send, not research you have to rescue. Set it up once, then let the workflow do the busywork.
Need Help Setting This Up?
Our automation experts can build and customize this workflow for your specific needs. Free 15-minute consultation—no commitment required.
