Skip to main content

​
Deepline SDK

Beta: the SDK docs are published and usable, but the surface is still evolving.
Write GTM workflows that just work: start from a search, CSV, webhook, or schedule; enrich the right records; route each customer through the right path; push the result to Salesforce, a sequencer, a warehouse, or a CSV; then visualize what changed.

​
1. Pick the job

Keep the use case short. The workflow can hold the detail.
JobExample
Weekly account searchFind new companies, enrich them, send qualified accounts to Salesforce
Custom waterfallTry providers in your order, stop at the first verified email
Segment routingRun one path for directors, another for VPs
Inbound routingEnrich demo requests and send the right context to the owner
Data reviewExport a CSV and chart segments, signals, and gaps

​
2. Pilot one record

Start with a prebuilt workflow before writing your own. 
deepline plays describe prebuilt/person-linkedin-to-email --json

deepline plays run prebuilt/person-linkedin-to-email \
  --input '{"linkedin_url":"https://www.linkedin.com/in/real-person/","first_name":"Jane","last_name":"Smith","company_name":"Acme","domain":"acme.com"}' \
  --watch
Use a real lead you are allowed to enrich. A fake LinkedIn URL can return nothing.

​
3. Pilot a small list

head -n 4 leads.csv > /tmp/leads-pilot.csv

deepline plays run lead-email-waterfall.play.ts \
  --csv /tmp/leads-pilot.csv \
  --watch

deepline runs export <run-id> --out /tmp/leads-pilot-enriched.csv
Inspect the output. If the rows look right, run the full file.

​
4. Add your business logic

This is where the SDK matters. 
import { definePlay } from 'deepline';

type Lead = {
  lead_id: string;
  seniority: 'VP' | 'Director';
  company_domain: string;
};

export default definePlay('route-leads-by-seniority', async (ctx, input: { leads: Lead[] }) => {
  const rows = await ctx
    .dataset('routed_leads', input.leads)
    .withColumn('research', async (lead, rowCtx) => {
      if (lead.seniority === 'VP') {
        return await rowCtx.tools.execute({
          id: 'vp_research',
          tool: 'test_rate_limit',
          input: { key: lead.company_domain },
          description: 'Research this enterprise lead.',
        });
      }

      return await rowCtx.tools.execute({
        id: 'director_research',
        tool: 'test_rate_limit',
        input: { key: lead.company_domain },
        description: 'Research this midmarket lead.',
      });
    })
    .run({ key: 'lead_id', description: 'Route each lead to the right research workflow.' });

  return { rows };
});
Use this for custom waterfalls, segment-specific paths, weekly searches, and background jobs.

​
5. Ship the output

deepline plays run weekly-new-accounts.play.ts \
  --input '{"query":"B2B SaaS companies hiring RevOps"}' \
  --watch

deepline runs export <run-id> --out output/new-accounts.csv
Then push to Salesforce, send to a sequencer, write to your warehouse, or chart the result.

​
Examples

Weekly company search to Salesforce

Search once a week, enrich new accounts, and create Salesforce tasks.

Custom email waterfall

Try your preferred providers and stop at the first verified email.

Director vs VP routing

Run different research paths based on buyer type.

Inbound webhook enrichment

Enrich form fills and route them to the right owner.

Nightly account refresh

Refresh account signals on a schedule with a run cap.

Python caller

Trigger a saved workflow from Airflow, notebooks, or backend jobs.

​
Reference

SDK Reference

CLI commands, workflow authoring, config, triggers, and org safety.

API Reference

HTTP request shapes for starting and checking workflow runs.