MCP (ChatGPT / Claude) · Packaging supplier

Packaging supplier → food and beverage producers in Czechia: what one real Prosp run returned

By the Prosp team · run on 2026-10-03 in production

Run date

2026-10-03

Job duration

9 min 31 s

Results came from

Live web search (1 of 8 from the knowledge base)

Delivered

8 of 8 requested

Rejected candidates

91

Discarded after target

20

The scenario

A packaging supplier wants Czech food and beverage producers that run their own production and bottling lines, because those are the companies that buy packaging. The buyer is a profile we set up to test this market, not a customer. The run is real, and every company on this page came back from it.

Buyer

A supplier of packaging materials selling to food and drink producers. An illustrative profile, not a named customer.

Target

Small and medium food and beverage producers in Czechia with their own production or bottling lines.

The exact request

Sent by an AI agent through the MCP server as the tool call find_prospects with these arguments:

{
  "criteria": {
    "query": "Find food and beverage producers in Czechia that run their own production and bottling lines and buy packaging",
    "industries": [
      "manufacturing",
      "agriculture_food_production"
    ],
    "locations": [
      "Czech Republic"
    ],
    "company_sizes": [
      "small",
      "medium"
    ],
    "keywords": [
      "food production",
      "bottling",
      "packaging line"
    ],
    "maxResults": 8
  }
}

How this run was actually started: through Prosp's internal job endpoint, with exactly these criteria and a target of 8 companies, under an internal exploration account whose results never reach a CRM. That path does not charge credits. Through MCP, an agent sends these arguments to find_prospects, then calls get_discovery_status and get_discovery_results with the job id it gets back.

What came back

Checking the registry ids

Why candidates were rejected

Not counted above: 20 more candidates were still being processed when the target was reached. They were discarded without being evaluated (terminal_job_discarded), so they say nothing about quality either way.

What went wrong

• Madeta's own description says it is the largest milk processor in the Czech Republic, but its company_size is empty, so the small/medium preference could not exclude it. Size is a preference, not a hard filter, and rows without a size pass it.
• Vinařství Spielberg is a Moravian winery stored with the city Praha, which is very likely wrong. Its register match was low-confidence, so no IČO was attached.
• 3 of 8 rows have no registry id (Madeta, Avena, Vinařství Spielberg), although all three are Czech companies that ARES should hold.
• 2 of 8 rows have no company size (Madeta, Avena).

How to read this run

The eight cover dairy (Madeta), spices and dried vegetables (B E N K O R), edible oils, pasta and sugar (Avena), beer (Pivovar Ferdinand), juices and cider (VITAMINÁTOR, Moštárna Louny) and wine (Vinařství Spielberg, SONBERK). All of them produce their own goods rather than resell them. Avena's delivered description says it bottles edible oils for private labels and industrial partners, the most direct match for a packaging supplier.

Size is where the request and the result part ways. The request asked for small and medium companies. Where Prosp knew a company's size it held to that: the largest rejection reason was enterprise_too_large (11). Where it did not know the size, the row passed, and that is how Madeta, by its own account the country's largest milk processor, was delivered.

Five rows carry an employee count from the register (employee_count_source: registry); the other counts come from web sources or a language-model estimate, as the table shows.

What this market can and can't verify

• Registers: yes, free and fast. The Czech business register ARES has a public API. On 2026-10-03 we looked up each of the five IČO numbers Prosp delivered. All five matched on company name and seat, and none had an end date (all active). Each lookup took under a second; the links in the table above go straight to the register record.
• Gaps: three rows came back without an IČO. For Vinařství Spielberg the register match was low-confidence and was withheld rather than guessed; for Madeta and Avena the lookup found nothing, although both are registered Czech companies.

Cost and time

• The job took 9 min 31 s. At submission the API estimated a median of 560 s.
• A discovery costs 100 credits through the REST API, the MCP server or the Telegram bot. This run was started on an internal account and was not charged.

How to reproduce this run

• ChatGPT or Claude: connect the MCP server and ask your assistant for the same companies, or have it call find_prospects with the arguments above.
• REST API: send the same criteria to POST /api/v1/discoveries. See the API reference.
• Telegram: run /find in the bot. See the Telegram guide.

A repeat will not return exactly these eight. The web changes, and each run adds to Prosp's knowledge base, so a later run may be served partly from companies found by earlier ones.