Workflow Enrich

Adds company and person attribute fields to a matched dataset from the Landbase graph. This is the attribute enrichment step — it adds fields like industry, size, description, and tech stack. For email and phone enrichment, use contact-enrich instead.

Run

landbase-cli workflow enrich ds-MATCHED \
  --company-fields=industry,size_range,description,keywords,technologies_used \
  --wait

An unrecognized field name is not an error: the run SUCCEEDs and that column comes back empty on every row. Use the names below verbatim, and datasets peek the output before feeding it to qualify or research — those read a blank column as “no information” and answer anyway.

Interactive runs: enrich on a large dataset is long. In an interactive session, trigger without --wait, capture the workflow_run_id, and follow the workflow-monitor skill to narrate live progress — a --wait subprocess can’t stream its progress line into the chat. Keep --wait for scripted/CI runs or when the user just wants the final result.

Flags

Standard field pack

Common company fields for most enrichment tasks:

industry,
size_range, revenue_range, employees_count,
hq_country, hq_state, hq_city,
founded_year, description, keywords, technologies_used,
last_funding_round_name, last_amount_raised_usd, total_amount_raised_usd

Identity columns (company_name, website, linkedin_url) are already on the matched row — no need to request them again.

Names that look plausible but return an empty column: business_description (use description), sub_industry, growth_signals, funding_stage (use last_funding_round_name), employee_count (use employees_count), last_funding_amount_usd / total_funding_usd (use the _amount_raised_usd spellings above). These are enrich-path names only — the same word may be valid in a search query, which resolves against a different vocabulary.

Getting the enriched dataset

Enrich writes a new dataset and leaves ds-MATCHED untouched — the finished run names the output in output_dataset_id:

landbase-cli workflow enrich ds-MATCHED --company-fields=industry,size_range --wait \
  | jq -r '.poll.run.output_dataset_id'
# → ds-ENRICHED

Without --wait: workflow status ds-MATCHED --run=<run-id> → .[0].output_dataset_id once SUCCEEDED.

Enrich output downloads directly — datasets download <ds-ENRICHED> out.jsonl.gz, no publish step, unlike qualify and research. Those are the native bytes; name the file .jsonl.gz because that is what enrich writes, not because the extension converts anything. publish only if you need CSV.

In the pipeline

workflow enrich fits both the upload pipeline and the search-refinement path:

# Upload pipeline
upload → onboard → match → enrich → qualify / research → publish

# Search-created pool refinement (Path B in the landbase-search skill)
search --mode=default → enrich → transform → publish

Datasets from search --mode=default are already matched, so they go straight to enrich (no onboard/match). Uploaded CSVs must be onboarded and matched first.

See dataset-pipeline for the full upload pipeline with ID extraction, and workflow-transform for the refinement step.

When to use this instead of –mode=research

To filter or rank a search --mode=default list on a catalog attribute it doesn’t carry yet, workflow enrich is the self-serve alternative to a fresh --mode=research run — and it doesn’t charge credits, which --mode=research does. Append the field here, then filter/rank with workflow transform.

Works when the attribute is in the Landbase company/person catalog — e.g. technologies_used, description, employees_count, revenue_range, founded_year, keywords.

Does NOT work — use --mode=research instead:

• The attribute is non-catalog — job-posting signals, headcount trend (enrich returns an empty column, and a WHERE on it drops every row). Confirm with datasets peek on the enrich output.
• The rows you need were already filtered out of the pool by the original search — enrich only appends to rows already present, it can’t recover excluded rows. Widen the original --mode=default search instead.
• The signal needs a phrase-match over the full Landbase database to define the universe — enrich only appends to your existing pool.

Red Flags — STOP

• About to request identity fields (company_name, website) in --company-fields → they’re already on matched rows, skip them
• Running enrich before match → dataset must be matched first; raw uploaded/onboarded datasets don’t have Landbase IDs to look up against
• Confusing this with contact-enrich submit → workflow enrich adds graph attributes; contact-enrich finds email/phone via external providers