Step 2: Connect Implisense API Access and Run Settings

Define API access and run parameters that control the size and explanation behavior for the lookalike request.

1. Open Auth Headers and set x-api-key to your API key (replace [CONFIGURE_YOUR_API_KEY]).
2. Open Run Settings and set explain to false.
3. In Run Settings, set size to 3.
4. Verify the connection order: Manual Launch → Auth Headers → Run Settings.

Step 3: Set Up Company ID Serialization

Prepare the company identifiers so they can be sent in the lookalike API request.

1. In Sample Company IDs, keep the provided jsCode to output a list of id values for testing.
2. In Serialize Identifiers, keep the function that builds serialization from each item.json.id.
3. In Format Serialization, set serialization to {{ $json.serialization.replace(/[,]/g,'","').trim() }} and enable keepOnlySet.
4. Confirm the flow: Run Settings → Sample Company IDs → Serialize Identifiers → Format Serialization.

Step 4: Configure the Lookalike Request and Routing

Send the lookalike request and route outputs based on the explain flag.

1. In Request Lookalikes, set URL to https://api.implisense.com/recommend?explain={{ $('Run Settings').item.json.explain }}&size={{ $('Run Settings').item.json.size }}.
2. In Request Lookalikes, set Request Method to POST and enable JSON Parameters.
3. Set Body Parameters JSON to { "baseCompanies": [ "{{ $json["serialization"].slice(0,-3) }}" ], "locationsFilter": [ { "type": "state", "code": "de-be" } ] }.
4. In Route ICP vs Lookalikes, set the rule to equals with leftValue {{ $('Run Settings').item.json.explain }} and rightValue true.
5. Confirm flow: Format Serialization → Request Lookalikes → Route ICP vs Lookalikes.

Step 5: Set Up Term Processing and the AI Narrative

Extract term features, build language digests, and generate the ICP narrative with OpenAI.

1. In Extract Term Features, keep the function that iterates items[0].json.targetProfile.features and outputs new items; note it continues on error.
2. Extract Term Features outputs to both Filter English Terms and Filter German Terms in parallel.
3. In Filter English Terms, keep the filter for item.json.key === 'term.en', then use Sort Terms Desc and Build EN Digest to generate the English digest string.
4. In Filter German Terms, keep the filter for item.json.key === 'term.de', then use Sort Terms Desc 2 and Build DE Digest to generate the German digest string.
5. Use Combine Term Digests with mode combine and combinationMode multiplex, then map the combined output in Prepare Prompt Text with text set to {{ $json["digest"] }}.
6. In Generate ICP Narrative, select model gpt-5-mini and set the user content to {{ $json.text }}; keep the system content as provided in the node.
7. Credential Required: Connect your openAiApi credentials in Generate ICP Narrative.
8. In Store ICP Report, set icp_report to {{ $json.output[0].content[0].text }}.

Step 6: Configure Output Mapping for Lookalike Results

Expand the lookalike response and map fields into CRM-ready output.

1. In Expand Company List, set fieldToSplitOut to companies.
2. In Map CRM Fields, map fields exactly as follows: crmLeadScore to {{Math.round($json.score * 100)}}, crmAccountName to {{$json.name}}, crmDomain to {{$json.url}}, crmCity to {{$json.city}}, crmZip to {{$json.zip}}, crmStreet to {{$json.street}}, crmImplisenseId to {{$json.id}}, crmProfileUrl to {{$json.profile}}, crmExplanation to {{$json.explanation}}, and crmIsActiveCompany to {{$json.active}}.
3. Confirm the branch flow: Route ICP vs Lookalikes → Expand Company List → Map CRM Fields.

Step 7: Test and Activate Your Workflow

Validate both the lookalike and ICP narrative branches before turning the workflow on.

1. Click Execute Workflow to run from Manual Launch.
2. Verify a successful lookalike run returns data into Map CRM Fields when explain is false.
3. Set Run Settings → explain to true and re-run; confirm Store ICP Report receives icp_report.
4. Once confirmed, switch the workflow to Active for production use.