A research digest that runs on your own hardware.

Stop paying per-token for summaries. This n8n workflow searches your SearXNG, asks a local Ollama model to summarize the results, and sends you a polished digest by email or Telegram, on a schedule you set.

Download the template See how it works

What is this?

A single importable JSON file you drop into any n8n instance. It gives you an automated research pipeline:

Schedule Trigger fires on your cron schedule: daily, weekly, or weekdays only.
Parallel SearXNG search hits your self-hosted instance for every topic you configure.
Ollama summarization sends all raw results to a local model (Qwen, Gemma, Llama) and returns structured summaries ranked by relevance.
Dual delivery formats the digest as HTML email and Telegram message, sent automatically.

No OpenAI API bills. No third-party summarization services. Everything runs on your server.

Who is this for?

Researchers & analysts who track multiple topics daily and want a morning briefing without manual searching.
Small business owners in Israel or elsewhere who want automated market intelligence without recurring API costs.
Self-hosting enthusiasts who already run SearXNG + Ollama and want to put them to work together.
n8n users looking for a production-ready workflow template they can customize and extend.

How it works

The workflow is a linear pipeline with parallel search. Here are three concrete scenarios:

Scenario 1: Daily tech briefing

Set 5 topics: "AI regulation EU", "Open-source LLM releases", "Rust security advisories". The workflow runs every morning at 7 AM, searches SearXNG for each topic, summarizes the top results with your local model, and sends a single email digest to your inbox. You start the day already informed.

Email preview

Morning Digest — Mon, Nov 3

AI regulation EU
1. New AI Act enforcement timeline extended to 2026...
2. France proposes stricter rules on generative AI training data...
3. EU member states disagree on enforcement mechanisms...

Open-source LLM releases
1. Mistral announces next-gen 8B model...
2. Llama 3.1 community fine-tunes hit 50k on HuggingFace...
3. Qwen 2.5 coder update adds 128K context...

Rust security advisories
1. CVE-2024-XXXX in serde crate...
2. Tokio runtime memory leak fixed in 1.37...
3. RustSec adds three new advisories...

Scenario 2: Weekly market scan for a small business

A Tel Aviv startup owner configures topics like "competitor pricing changes", "new funding rounds Israel tech", "regulatory updates fintech". Every Monday, the digest arrives on Telegram with concise summaries and links. No manual Googling, no subscription fatigue.

Telegram preview

Weekly Market Scan

Competitor pricing changes
- Competitor A raised prices 15% for enterprise tier...
- Competitor B announced free tier removal...
- New player X enters market with freemium model...

New funding rounds Israel tech
- Tel Aviv AI startup raises $12M Series A...
- Cybersecurity firm Y secures $8M seed...

Regulatory updates fintech
- Bank of Israel publishes draft guidelines on crypto...
- New compliance deadline moved to Q2 2026...

Scenario 3: Academic literature watch

A PhD student sets topics for their research area (e.g., "large language model alignment", "reinforcement learning from human feedback"). The workflow runs twice weekly, searches SearXNG across arXiv and news sources, summarizes the most relevant papers, and delivers them to both email and Telegram. They never miss a key publication.

Academic digest

Literature Watch — Wed, Nov 5

LLM alignment
Paper: "Constitutional AI: Harmlessness from AI Feedback"
Summary: Proposes a method for training AI to self-critique using a set of principles...
Link: arxiv.org/abs/2212.08073

RLHF updates
Paper: "Direct Preference Optimization Beyond PPO"
Summary: Introduces DPO as a simpler alternative to PPO-based RLHF with comparable performance...
Link: arxiv.org/abs/2305.18290

The pipeline at a glance

Schedule Trigger
       |
       v
  Set Topics (JSON array)
       |
       v
  Parallel HTTP Requests --> SearXNG /search?q=...&format=json
       |
       v
  Collect & Group Results by Topic
       |
       v
  Build Prompt for Ollama /api/chat
       |
       v
  Call Ollama (local model) --> get summaries
       |
       v
  Format Digest (Code node)
       |
       +---> Send Email (SMTP)
       +---> Send Telegram (Bot API)

Requirements

Item	Details
n8n	v1.0+ (self-hosted recommended)
SearXNG	Self-hosted with JSON API enabled (`/search?q=...&format=json`)
Ollama	Running locally, model pulled (`ollama pull qwen2.5:7b`)

All three can run on the same machine or in Docker. Minimum 8 GB RAM (16 GB recommended for larger models). GPU accelerates Ollama but is optional.

Pricing

Option	Price	Details
Template only	$19	workflow.json + config.example.json + README.md
Template + Setup Guide	$39	Everything above + detailed video walkthrough
Bundle (3 templates)	$49	This digest + 2 other local-AI workflows

First two customers get the template at cost — email me for a discount code.

Frequently asked questions

Do I need to pay for the LLM?: No. The workflow calls your local Ollama instance (/api/chat). Zero API costs.
Can I use any model?: Yes. Pull any model via ollama pull <name> and set it in the workflow config. Works with Qwen 2.5, Gemma 4, Mistral, Llama 3, Phi 3, and more.
Does it work without SearXNG?: The search step requires a SearXNG-compatible JSON API. If you don't have one set up yet, I can help — just reach out.
Can I send to WhatsApp instead of Telegram?: Not directly — n8n's WhatsApp integration requires a business account. Telegram is the easiest free option. You could also route through any HTTP webhook.
Is this beginner-friendly?: If you can install Docker and run npm install n8n, yes. The README walks through every step. I also offer setup assistance for small businesses in Israel (INR 2,500-8,000 depending on scope).
Can I modify the workflow?: Absolutely. It's MIT licensed. Change topics, add nodes, re-route connections — it's your instance.

Get the template

Instant download. Import into n8n in under 10 minutes.

Buy on Gumroad — $19

Questions or need help setting it up? Email me directly:

tevel.mantsur@gmail.com