Nordic Tender Matcher (AI)

Answers the question "which Nordic public tenders should my company bid on?" — not "here are 10,000 rows."

Give it a short description of your company. It fetches the open public tenders from Norway, Finland and Sweden, uses OpenAI to pre-filter the obviously irrelevant ones, then scores each remaining tender against your profile and returns a ranked shortlist with a reason, an English summary and the key requirements for every match.

It's built for the moment a salesperson, bid manager or founder asks "is there anything worth bidding on this week?" — and for AI agents that need that answer as a tool call.

Quick start

1. Fill in the input — set the fields in the input tab (prefilled with a working example).
2. Click Start — the Actor runs and writes one record per item to the dataset.
3. Get your data — export to JSON, CSV or Excel, or pull it via the API.

---

How it works

1. Fetch — pulls open tenders from the selected sources (Norway = Doffin, Finland = Hilma, Sweden + current Finland = EU/TED), reusing the same official-API logic as the Nordic Public Tenders Actor. No scraping.
2. Cheap pre-filter — OpenAI turns your companyProfile into relevant CPV codes and keywords (one small call), and obviously irrelevant notices are dropped before any expensive analysis.
3. AI scoring — OpenAI scores each surviving tender and returns matchScore (0–100), matchReason, summaryEn and keyRequirements.
4. Rank & deliver — matches at or above minScore are sorted by score and returned, best first.

Real OpenAI token usage (and an estimated cost) is logged for every run and stored in the run's SUMMARY.

---

Why it's different

A raw tender feed makes you read every notice in three languages and decide what fits. This Actor does the reading and the judgement call:

• Ranked, not raw — you see the best opportunities first, with a score.
• Explained — every match says why it fits your company (and where the gaps are).
• English everywhere — summaryEn, matchReason and keyRequirements are always English, regardless of the tender's language.
• Agent-ready — see "Use it as an AI-agent tool" below.

---

Input

Field	Type	Default	Description
companyProfile	string	— (required)	Free text: what you sell/do, size, regions, references. The more specific, the better the matching.
countries	string[]	all	Sources to search: ted (current SE + FI), doffin (NO), hilma (historical FI).
publishedFrom	string (ISO date)	7 days ago	Only tenders published on/after this date.
publishedTo	string (ISO date)	today	Only tenders published on/before this date.
minScore	integer	60	Only deliver matches scored at or above this (0–100).
maxResults	integer	25	Max delivered matches (one tender-matched charge each).

---

Output

One dataset item per match, highest score first. It is the normalised tender schema plus the AI analysis:

Field	Type	Description
matchScore	number	0–100 fit score for your company.
matchReason	string	2–3 sentences: why it fits (and gaps).
summaryEn	string	What's being bought, scope, key requirements — in English.
keyRequirements	string[]	The most important requirements/qualifications.
id, source, country	string	Stable id, source (doffin/hilma/ted), buyer country (NO/FI/SE).
title, titleEn	string	Original and English title.
buyerName, buyerOrgNumber	string | null	Contracting authority and its org number.
cpvCodes	string[]	CPV classification codes.
estimatedValue, currency	number | null, string | null	Estimated value, if published.
publishedDate, deadline	string | null	ISO-8601 dates.
noticeType, procedureType	string | null	Notice and procedure type.
description	string | null	Original-language description.
url	string | null	Official public notice page.

---

Examples

1. IT consultancy (Sweden + Finland)

companyProfile

Mid-sized Swedish IT consultancy (40 employees) specialising in cloud migration,
custom software and system integration for the public sector. Operates in Sweden
and southern Finland. References in regional healthcare and municipal e-services.

Example output item

{
  "matchScore": 86,
  "country": "SE",
  "titleEn": "Sweden – IT consulting services – Framework agreement for system development",
  "buyerName": "Västra Götalandsregionen",
  "matchReason": "Strong fit: the buyer wants public-sector system development and integration, exactly your core offering, and your regional healthcare references are directly relevant. The required team size is within reach for a 40-person firm.",
  "summaryEn": "Framework agreement for custom system development and integration services for a regional authority. Multi-year, call-off based, with options for maintenance.",
  "keyRequirements": ["Public-sector references", "Agile delivery capability", "GDPR/security compliance", "Swedish-speaking team"],
  "deadline": "2026-11-03T23:59:00.000Z",
  "estimatedValue": 40000000,
  "currency": "SEK",
  "url": "https://ted.europa.eu/en/notice/-/detail/677811-2026"
}

2. Facility cleaning company (Norway)

companyProfile

Norwegian facility-services company, 120 employees, providing commercial cleaning,
window cleaning and light grounds maintenance across Oslo and Viken. ISO 14001
certified. No construction work.

Returns high scores on municipal/regional cleaning and facility-maintenance framework agreements in Oslo/Viken; low scores on construction, catering or IT notices (dropped at pre-filter or scored below minScore).

3. Medical equipment distributor (all three countries)

companyProfile

Distributor of medical imaging equipment and consumables (MRI, CT, ultrasound)
with service contracts. Sells to hospitals across the Nordics. Not a software vendor.

Returns imaging-equipment and service-contract tenders from hospitals/health regions in NO/FI/SE, with keyRequirements surfacing things like CE marking, service-level guarantees and installation scope — and clearly lower scores where a notice is only teleradiology software rather than hardware.

---

Pricing — pay per event

Two events, so you only pay for work actually done:

Event	When it's charged	Notes
tender-scanned	Per notice pre-filtered (the cheap scan step)	Covers fetching + cheap filtering.
tender-matched	Per match delivered after AI analysis	Covers the OpenAI scoring + margin.

How to price it (developer note). The main cost is the OpenAI scoring of the tenders that survive pre-filtering. Set tender-matched high enough to cover the OpenAI tokens for the analysed set (including analysed-but-below-threshold notices) with margin, and keep tender-scanned low. Every run logs real token usage and an estimated AI cost (in the log and in SUMMARY.ai), so you can calibrate prices against actual spend. Note Apify enforces a $0.10 minimum price per event.

---

Environment variables / secrets

Variable	Required	Purpose
OPENAI_API_KEY	Yes	OpenAI API key — powers the pre-filter and scoring.
OPENAI_MODEL	No	Override the model (default gpt-4o-mini).
DOFFIN_API_KEY	For Norway (doffin)	Doffin Public API key (Ocp-Apim-Subscription-Key).
HILMA_API_KEY	For historical FI (hilma)	Hilma AVP-Read key (Ocp-Apim-Subscription-Key).

Sweden + current Finland (via ted) need no key.

---

Use it as an AI-agent tool (Apify MCP)

Once published, this Actor is callable through the Apify MCP server (https://mcp.apify.com). Any MCP client — Claude, Claude Code, Cursor, ChatGPT — can invoke it as a tool. Because the input is a single natural-language companyProfile and the output is a ranked, explained shortlist, an agent can be told:

"We're a Finnish security-services firm — find public tenders worth bidding on this month."

…and call this Actor directly, then reason over the returned matches. The flat, English, scored output is designed to be consumed by an LLM without extra glue code.

---

Run it locally

npm install
export OPENAI_API_KEY=...     # required
export DOFFIN_API_KEY=...        # for Norway (optional)
export HILMA_API_KEY=...         # for Finnish history (optional)

# Put your input in storage/key_value_stores/default/INPUT.json, then:
apify run
# or: npm run build && node dist/main.js

---

Related

This Actor reuses the data-collection layer of Nordic Public Tenders (the raw-feed Actor). Use that one if you want the unfiltered notices; use this one if you want the answer to "what should we bid on?".

License

ISC.

Related actors

• Nordic Public Tenders (Norway, Finland, Sweden) — Collects public procurement notices from Norway (Doffin), Finland (Hilma) and Sweden
• Nordic Company Registry (Norway, Finland) — Official company-registry data from Norway (Brønnøysund) and Finland (PRH) in one
• Signal Radar — AI Account Intelligence & ICP Scoring — Turn company domains into ICP-scored, signal-tagged account intelligence from public