<![CDATA[Delavo: AI & SaaS Tools Tested]]>https://delavo.hashnode.devRSS for NodeWed, 04 Nov 2026 17:34:04 GMT60<![CDATA[No-code scraping vs. brittle scripts: where Browse AI helps and where it stops]]>https://delavo.hashnode.dev/no-code-scraping-vs-brittle-scripts-where-browse-ai-helps-and-where-it-stopshttps://delavo.hashnode.dev/no-code-scraping-vs-brittle-scripts-where-browse-ai-helps-and-where-it-stopsFri, 30 Oct 2026 09:31:34 GMTAnyone who has maintained a scraper knows the pattern: it runs fine for weeks, then a class name changes and the data quietly stops arriving. Browse AI is a no-code tool aimed at exactly that maintenance problem. This is a summary of what the vendor documents and what users report. I have not run a hands-on test of Browse AI myself.

How it works

You train a "robot" in a browser extension by clicking the data you want, for example product name, price and URL on a listing page. The robot repeats those steps on a schedule, from hourly to monthly. For common sites such as Amazon, Google Maps, LinkedIn or Glassdoor, the vendor offers 250+ prebuilt robots that it maintains itself.

Results go to Google Sheets, Airtable, Zapier, Make, webhooks, a REST API, or CSV and JSON exports.

The part that matters for maintenance

Selector-based scrapers break silently when a layout changes. According to the provider, Browse AI adapts a robot after many layout changes and sends an alert. That is the main reason to choose it for continuous monitoring rather than one-off extractions. Fundamental structural changes still require retraining the robot.

Where it stops

  • Strong anti-bot protection and login walls: on sites with Cloudflare Enterprise or reCAPTCHA v3, the AI adaptation does not help and manual retraining remains necessary.
  • Pagination and dynamic content: configuration effort rises noticeably compared with simple listing pages.
  • Credit billing: with sporadic or seasonal use, part of the monthly allowance can go unused.
  • Data location: according to the provider, data is hosted on AWS in the US (as of October 2026). A data processing agreement is available on request. Inside the EU, that deserves a review.
  • Legal side: robots.txt and website terms remain the user's responsibility, not the tool's.

Practical tips

  1. Start with a monitoring use case. For a single extraction without monitoring, a simpler tool such as Thunderbit is the more direct fit.
  2. Use bulk runs and robot chaining. Feed a list of URLs in one step, or chain two robots: the first collects product links, the second scrapes each detail page.
  3. Treat alerts as part of the pipeline. Route change notifications to Slack or email through Zapier so a broken robot does not go unnoticed.
  4. Test on the free plan first. According to the provider, it includes 50 credits per month, 2 websites and unlimited robots, which is enough for a small monitor.
  5. Know when to look elsewhere. If you need full control over the scraping infrastructure or very high volume, Apify is the better direction, at the cost of programming effort.

Takeaway

Browse AI trades control for maintenance savings. For marketing, SEO and e-commerce teams without developers who need regular price, ranking or listing data, that is a good deal. For protected sites, enterprise volume or strict data residency requirements, it is not.

I run delavo.de, where this article was first published as part of a longer review with a rating breakdown and FAQ: Browse AI review on delavo.de. The original draws on vendor information and user reviews, not on a hands-on test. This summary was written with an AI agent based on that original.

]]><![CDATA[Privacy-first newsletters in the EU: what I learned testing rapidmail]]>https://delavo.hashnode.dev/privacy-first-newsletters-in-the-eu-what-i-learned-testing-rapidmailhttps://delavo.hashnode.dev/privacy-first-newsletters-in-the-eu-what-i-learned-testing-rapidmailTue, 27 Oct 2026 13:56:22 GMTIf your side project or small business needs a newsletter and your readers are in the EU, the hard part is rarely the editor. It is consent, tracking and deliverability. I tested rapidmail, a German newsletter tool, on a free account in October 2026 and focused on exactly those points. Here is what matters if you care about privacy and want mails that actually arrive.

Tracking can be switched off, not just anonymised

Most newsletter tools let you anonymise open and click tracking. rapidmail lets you deactivate it completely in the privacy settings. That is useful if you only want to track with explicit consent.

The trade-off is real: without tracking, A/B tests, follow-up mailings ("send again to everyone who did not open") and activity-based segments stop working. Decide this before you build automations around it.

Deliverability checks before you hit send

Before every mailing, a spam check looks at:

  • Authentication (DKIM, SPF, DMARC) of your sender domain
  • Email weight, i.e. size and image load
  • Mobile-friendliness
  • List quality

If you have never set up DKIM or DMARC for your domain, this check is a good nudge to do it. In my test the mail arrived immediately and rendered correctly in Outlook.

Imports force you to think about consent

Recipients can be imported by file (xls, csv, txt), by copy and paste or one at a time. Before each import you have to confirm that consent exists and state whether the addresses come from third parties. It slows you down for a few seconds, but it keeps messy, bought or scraped lists out.

Sign-up forms come with a built-in captcha and double opt-in. One catch for English audiences: the form texts were pre-filled in German even though I used the English interface, so check them before you embed the form.

The AI is opt-in and transparent

The only AI feature in my test was subject line autocomplete with a quality score. It is off by default. When you enable it, rapidmail states that your website data and the subject lines you type are sent to OpenAI. I like that this is explicit.

The results were mixed: a German subject line start got five English continuations, one of them off-topic. The suggestions seem to follow the interface language, not the language of your text.

Automations without a CRM

There are six templates (welcome series, birthday, event reminder, reactivation, email course, latest issue for new subscribers) and a visual builder with three step types: email, wait and condition (received, opened, clicked). That covers typical newsletter flows. There is no CRM and no sales pipeline, and transactional emails via API are only available on request.

Where it does not fit

  • You need transactional and marketing mail from one API: look at Brevo, which also has a permanent free plan.
  • You need multi-step customer journeys with a CRM: ActiveCampaign goes much deeper.
  • You expect AI writing help in the editor: not available in my test.

There is no permanent free plan, but there is a pay-per-send option without a subscription and, according to the provider, a free first mailing. Servers are in Germany and a data processing agreement can be concluded in the account, also according to the provider. Check the contract and your settings yourself before going live; this is not legal advice.

Bottom line

rapidmail is a straightforward newsletter tool for small teams in the EU that want strict tracking control and a pre-send deliverability check. It is not the right choice if newsletters are only one part of a larger email and CRM stack.

I run delavo.de, where I test AI and SaaS tools hands-on. The full rapidmail review with screenshots of every step and a detailed rating was first published there. This summary was written with an AI agent based on that original hands-on review.

]]>