I kept thinking about what the perfect automation client looks like — built this as a result

Been thinking a lot lately about what makes a great automation client.

Public sector kept coming up. Slow to change, massive budgets, predictable

procurement cycles, and they’re legally required to publish every contract

publicly. The barrier to get in is high, but once you’re in, you’re in.

The problem is finding the right opportunities at the right time. Public

tenders get published daily across multiple portals. By the time a relevant

one gets noticed manually, the deadline is already close or someone else

got there first.

So I built a monitor.

Every morning it automatically pulls all new Austrian public tenders from

official EU sources, downloads each PDF, and uses easybits to extract the

key data, title, contracting authority, federal state, contract value,

deadline, CPV code, contact email. Anything above €100k lands in Slack

within minutes of publication. Everything else gets stored quietly.

The Slack message looks like this:

:bell: Neue Ausschreibung – Österreich

Titel: Website Redesign

Auftraggeber: Institute of Science and Technology Austria

Bundesland: Niederösterreich

Auftragswert: 350.000 €

Frist: 04/05/2026 12:00:00

CPV: 72413000

Kontakt: Email

PDF: Link

No more manual searching. No more missed deadlines.

What I want to build on top of this next:

- Vector database storing every extracted tender for semantic search

- Slack slash command powered by OpenAI — type a question in natural

language, get back matching tenders with citations from the original PDFs

- Automatic bid drafting using the extracted tender fields

- Same setup for other EU countries — the data is all publicly available

Free template JSON in the comments. For questions on the data sources or

if you want help setting this up for your country or industry —

[email protected], happy to talk.

this is really thoughtful work on the public sector angle — the observation about tenders being legally required to publish is a genuine moat. few questions: are you pulling from TED (Tenders Electronic Daily) or from national procurement portals directly? and curious how you’re handling multi-language tender PDFs since the Austrian official ones tend to be German while EU-wide tenders mix languages. also — what does the match logic look like, is it keyword-based or do you have an LLM scoring the CPV codes against a client profile?