Starko

Empty fields, filled in.Research on every company and contact, with sources.

Enrichment reads a company's own website, searches the web and fills the empty fields on the record. It cites where each fact came from and never overwrites anything your team typed.

v2.starko.one/dashboard/companies/c2a9…
Companies/Kaltbach Energie
72%
OverviewTimelineClientsTicketsProjectsDealsAnalytics
Kaltbach Energie
Notes

Kaltbach Energie is a regional energy supplier based in Freiburg, Germany. It sells electricity and gas to households and small businesses in Baden-Württemberg and runs a growing rooftop solar installation service. The company operates its own customer portal and a mobile app for meter readings and invoices.

Services:

  • Household electricity and gas
  • Business energy contracts
  • Rooftop solar installation
  • EV charging tariffs

Target audience: Households and small businesses in southern Germany

Sources:

  • Kaltbach Energie, Über uns
  • Kaltbach Energie GmbH on LinkedIn
  • Handelsregister entry, Amtsgericht Freiburg
At a glance

Enrichment in short.

Companies get
A written profile with services, target audience and sources, plus industry, employees, founding year, LinkedIn, city, country and a logo.
Contacts get
Job title, LinkedIn profile, a short bio, city and country.
The one rule
Only empty fields are filled. Values your team entered are never overwritten.
Your questions
Up to 8 per list, for companies and for clients, answered as text, number or yes or no.
Confidence
Low-confidence answers stay in the run record and are never written to the record.
Freshness
Automatic and bulk runs skip records enriched in the last 24 hours. The Enrich button always runs.
Bulk
Up to 100 selected records at a time.
Permissions
Anyone can press Enrich. The workspace owner manages the questions and the blocklist.
Companies

How a company gets researched

Company enrichment runs in phases, and each phase sharpens the next. Starko reads the company’s own site first, works out who they are, then sends a research agent to the wider web with that context before it writes anything.
  • It reads the homepage from the company’s domain and the /about page when there is one. With no usable domain or a page too thin to learn from, the run stops and says why.
  • A quick pass pulls the canonical name, a one-line summary and a likely industry, so the searches that follow look for the right company.
  • The research agent picks its own queries, reads the pages it finds, follows business registries and LinkedIn, and searches in the company’s local language when that is where the answers are.
  • The company’s own site counts for what they do. Third-party sources count for numbers like employee count and founding year. Companies with similar names are filtered out.
  • The profile lands in the record’s Notes: a 3 to 6 sentence description, a list of services, a target audience line and a Sources list linking the pages used. If the company has no logo, its favicon becomes one.
Contacts

People, only when the match is certain

Person data is easy to get wrong, so contact enrichment is deliberately cautious. Writing the wrong person’s job title on a record is worse than writing nothing, and the task is built around that.
  • It only runs with a real name and company context: a linked company or a work email domain. A first name at a free-mail address is not enough to search on.
  • The evidence has to identify this person at this company. If it does not, nothing is written and the run is marked person not confidently identified.
  • A confident match fills the job title, the person’s own LinkedIn URL, a one to three sentence bio in the description, city and country.
  • Your client questions run only after the person is identified, with the found job title folded into the searches.
Your questions

Ask your own research questions

Settings, Enrichment holds custom research questions the AI answers for every enriched record, one list for companies and one for clients. Each answer is saved as an attribute under the field key, so it shows up as a table column, a board grouping and a filter in the CRM.
  • Each question has a label, a field key, the question in plain language and a type: text, number or yes or no.
  • Up to 8 questions per list. Keys the built-in fields already use, like industry or employees, cannot be reused.
  • Every answer carries a confidence (high, medium or low) and up to four source URLs. Low-confidence answers are kept in the run record and never written to the record.
  • Only the workspace owner can edit the questions.
v2.starko.one/dashboard/settings/enrichment
Settings/Enrichment

Enrichment fields

Custom research questions the AI answers for every enriched record. Answers land as attributes under the field key, usable in tables, boards, and filters.

Company questionsUp to 8 questions per list.
CRM in usecrm_in_useWhat CRM do they currently use?Text
Annual revenueannual_revenueWhat is their annual revenue in EUR?Number
Has mobile apphas_mobile_appDo they offer their own mobile app to customers?Yes / no
Add questionSave
Client questionsUp to 8 questions per list.
Decision makerdecision_makerDoes this person approve software purchases?Yes / no
When it runs

Automatically, on demand, in bulk or in a workflow

Most records fill themselves in. The rest are one click away, and an automation can run enrichment wherever it fits in your process.
  • When a new sender writes from a business email address, Starko creates or links their company from the domain, with a name and logo, and enriches it. These companies carry an Auto-created chip.
  • Free-mail domains and no-reply addresses never create companies, and the Blocklist in Settings rules out specific domains or addresses. Those senders still become clients.
  • A client created from an inbound email is enriched as a person when the name and company context allow it.
  • The Enrich button on any client or company page runs right away, even if the record was enriched recently. It spins until the run finishes.
  • Select up to 100 records in a list and press Enrich. The toast says how many runs were queued and how many were skipped, and the list shows an “enriching” count until they finish.
  • The Enrich company node puts the same pipeline in a ticket workflow, for example right after a company is created.
Provenance

Every run leaves a record

Each enrichment run is stored with every fact it found, the confidence and the sources, including the facts it did not write. The record’s Timeline shows each run as finished, skipped with the reason, or failed.
  • Skip reasons are plain, for example page too thin, no usable domain or person not confidently identified.
  • Writes check the current values at the moment of saving, so an edit a teammate makes while a run is in progress is never overwritten.
  • Web pages are read as data, never as instructions. The agent can only open pages that came up in its own searches or on the company’s domain.
  • When you merge duplicate records, their enrichment runs move to the record that stays.

FAQAbout Enrichment.

Keep exploring

Connect a channel and a repository. Turn the support queue into a build queue.

Feedback starts flowing in minutes. The first recommendation shows up as soon as a pattern repeats, and the first pull request as soon as you accept one.

Start for freeGet a demo