MCP (ChatGPT / Claude) · Wood-processing equipment dealer

Wood-processing equipment dealer → sawmills in Finland: what one real Prosp run returned

By the Prosp team · run on 2026-10-04 in production

Run date

2026-10-04

Job duration

47 min 39 s

Results came from

Live web search

Delivered

8 of 8 requested

Screened at completion

229

Rejected at completion

218

Discarded after target

49

The scenario

A dealer of wood-processing equipment wants Finnish sawmills and wood product manufacturers that run their own processing lines, because those are the companies that buy saws, planers and kilns. 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 dealer of saws, planers and drying equipment for the timber industry. An illustrative profile, not a named customer.

Target

Small and medium Finnish sawmills and wood product manufacturers that run their own processing 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 Finnish sawmills and wood product manufacturers that run their own timber processing lines",
    "industries": [
      "manufacturing",
      "furniture_manufacturing"
    ],
    "locations": [
      "Finland"
    ],
    "company_sizes": [
      "small",
      "medium"
    ],
    "keywords": [
      "sawmill",
      "saha",
      "timber products"
    ],
    "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

Also not counted above: 152 candidates were never judged at all. Our own pipeline ran out of time on them (an infrastructure stall on our side), so they say nothing about the companies; a re-run would judge them.

Not counted above: 49 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

• Kieppi Sawmill Oy (kieppisawmill.com) came back as MICRO although the request asked for small, medium companies.
• Uniwood Oy (uniwood.fi) came back as MICRO although the request asked for small, medium companies.
• Pohjan Timber-hirsi Oy (pohjantimber.fi) came back as MICRO although the request asked for small, medium companies.
• Puumanni Ky (puumanni.fi) came back as MICRO although the request asked for small, medium companies.
• 3 of 8 rows have no company size (Isojoen Saha Oy (isojoensaha.fi), Akonkoskensaha (akonkoskensaha.fi), Westas Group Oy (westas.fi)), so the size preference could not filter them.
• 2 of 8 rows carry no register match (Akonkoskensaha (akonkoskensaha.fi), Timberwise (timberwise.fi)).
• Pohjan Timber-hirsi Oy (pohjantimber.fi): The industry is 'agriculture_food_production' but the brief asked for manufacturing or furniture manufacturing.
• Timberwise (timberwise.fi): The description says 'Timberwise manufactures high-quality, durable plank parquet flooring for residential customers' but the brief asked for companies that run their own timber processing lines.

How to read this run

The table lists the 8 companies exactly as the run returned them, company-level fields only (name, domain, city, country, industry, business model, company size, employee count, employee count source, registry source). Where a cell is empty, the run did not have the value.

Company size is a band (micro, small, medium, large, enterprise); the employee count source says whether the head count came from an official register or is an unverified estimate. A registry source means Prosp matched the row to a government register.

Below the table, the rejection histogram gives the reasons candidates were screened out. It keeps counting candidates that finish after the target is reached, so its total can be larger than the number rejected when the job completed. Candidates still in progress at that point are listed separately as discarded: they were never evaluated.

Checking the registry

On 2026-10-04, 5 rows that carry a register match were looked up on the public register by the delivered name, and the register's seat and status were compared with the row. The id shown is the register's own: the exploration account this run used does not return the ids Prosp delivers.

• isojoensaha.fi — verified (PRH open data (Finnish Trade Register)): register: Isojoen Saha Oy (id 0405653-8), seat ISOJOKI, active.
• kieppisawmill.com — verified (PRH open data (Finnish Trade Register)): register: Kieppi Sawmill Oy (id 2803778-8), seat MÄNTYHARJU, active.
• westas.fi — verified (PRH open data (Finnish Trade Register)): register: Westas Group Oy (id 2523631-9), seat KOSKI TL, active.
• uniwood.fi — verified (PRH open data (Finnish Trade Register)): register: Uniwood Oy (id 1549920-3), seat LAHTI, active.
• pohjantimber.fi — verified (PRH open data (Finnish Trade Register)): register: Pohjan Timber-hirsi Oy (id 0982516-4), seat HAAPAVESI, active.

Cost and time

• The job took 47 min 39 s. At submission the median estimate was 9 min 47 s.
• This run was started on an internal exploration 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.

A repeat will not return exactly these rows. 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.