Saddler examples

Lead Enrichment Playbook

Turns a target-account CSV into enriched company records, a confidence brief, and a reviewable CRM write plan.

Run teardown

One run produced these files, costs, limits, and follow-up choices.

InputTarget-account CSV with 42 rows
Actions4 recorded highlights
Outputs3 produced files
Limit4 missing-domain records stayed draft-only
Source
42-row target CSV plus approved public sources
Output
Enriched CSV, summary, source log
Approval
CRM write plan paused before writeback
Cost
$8.82 in 23m 48s

Input materials

  • Target-account CSV with 42 rows
  • Campaign focus: regional finance operators
  • CRM field policy with writeable fields
  • Approved source list and low-confidence threshold

Action log highlights

  1. Validated 42 input rows and separated 4 records with missing company domains.
  2. Read CRM account fields in view-only mode before proposing any writeback.
  3. Searched public sources and attached citations to high-confidence enrichment fields.
  4. Generated a CRM write plan and paused for approval before touching the external system.

Produced artifacts

CSV

enriched_leads.csv

38 write-ready rows and 4 draft-only rows.

Produced by run_lead_demo_042 - Lead Enrichment Playbook v2

Markdown

enrichment_summary.md

Campaign-level findings, assumptions, and recommended next steps.

Produced by run_lead_demo_042 - Lead Enrichment Playbook v2

JSON

source_log.json

Per-field source trail used by the run result.

Produced by run_lead_demo_042 - Lead Enrichment Playbook v2

Why this run matters

Sales teams repeat the same research before every outbound campaign. The work is usually trapped in chat history, and CRM write decisions are hard to audit after the campaign launches.

How it was guided

The run used a short operating rule set. Keep the proof first, then open the prompt excerpt when you need to inspect the instructions.

Show prompt excerpt
Enrich each lead using public sources. Never invent missing fields. Cite every material claim. Separate low-confidence rows from write-ready rows. Ask before changing external systems.

What the agent could use

Connection

CRM read access

Reads account fields and proposed write targets. It cannot write without approval.

Skill

Lead rubric

Scores company size, buying signal, source quality, and missing-field risk.

Context

Campaign focus

Keeps the run aligned to the outbound segment and excluded industries.

Memory

Prior campaign notes

Recalls only approved account notes from the workspace memory store.

Cost, limits, next step

Run cost
$8.82
Duration
23m 48s
Human equivalent
About 3 hours of research and CRM prep
Known limit
4 missing-domain records stayed draft-only
Next action
Approve the write plan, request changes, or rerun with a stricter source rule.

What we'd change

We would add a stricter low-confidence rubric before letting this playbook handle enterprise accounts. The first run showed that funding data is noisy for private regional firms.

Try it yourself

Start from this example in your workspace, launch it on the sample CSV, then open the run result before changing the prompt.

Start from this example