How to Automate Enterprise Prospecting With a 9-Agent AI Pipeline
By Cece Renick, Strategic Account Executive at Chili Piper ·
What if typing one command could turn an account name into a complete prospecting package: a structured account plan, verified and tiered contacts, personalized outreach sequences for every rep, and LinkedIn ABM ads ready for Campaign Manager? That is what Cece Renick walked through in this GTM Clawllective session, presenting em-gine, a 9-agent Claude pipeline built by Emma Carey that Chili Piper's Strategic SDR team is actively using for enterprise prospecting.
This post covers how the system works, what each of the nine agents does, what connectors it requires, and how you can get your hands on it.
What is em-gine?
em-gine (Strategic Allbound Machine) is a 9-agent AI pipeline built on Claude Desktop. It ships as three installable plugin files and runs on a shared Supabase database that all agents read from and write to. The entry point is a single command: /run-pipeline [account name]. Nine sequential agents fire from there, each completing a distinct step in the prospecting workflow and writing its output back to Supabase for the next agent to consume.
The system was architected by Emma Carey and built in collaboration with Michael Nettleton and the Strategic SDR team at Chili Piper. It was the highest-rated submission at Chili Palooza 2026.
What problem does em-gine solve?
Enterprise prospecting is slow, inconsistent, and usually built on guesswork. Reps spend hours pulling CRM data, researching accounts, finding contacts, and writing outreach, and the quality varies depending on who does the work.
em-gine collapses that into a single automated run, with a quality gate built in before anything reaches an SDR. Critically, the system design started from actual CRM data rather than assumptions. The team analyzed closed won and lost deals, contact roles, and titles, then clustered ICPs before writing a single agent prompt. That data-driven foundation is what separates em-gine from pipelines built on gut feel.
Cece also noted that the team has built different configurations for mid-market, enterprise, and individual AE workflows, so the pipeline scales in both directions.
How is the system structured?
em-gine ships as three plugins, each handling a distinct phase of the prospecting workflow.
| Plugin | Version | What it does | Requires |
|---|---|---|---|
| em-gine-core | v3.0.4 | Pipeline orchestration, public research, contact discovery, Salesforce validation, account plan synthesis | Salesforce MCP, LeadIQ MCP, Supabase MCP |
| em-gine-intel | v3.0.1 | CRM data pull, Gong call intelligence, quality gate before plan reaches SDR | Salesforce MCP, Gong MCP, Supabase MCP |
| em-gine-outreach | v3.0.2 | Priority-based SDR sequences and LinkedIn ABM ad copy, pushed to SalesLoft | Supabase MCP, n8n, SalesLoft |
What are the 9 agents and what does each one do?
| Step | Agent | What it does |
|---|---|---|
| 1 | Pipeline Orchestrator | Creates the Supabase entry, sequences all 9 agents, manages approval gates |
| 2 | Internal Expert Agent | Pulls Salesforce CRM data and Gong call transcripts with per-contact relationship intel |
| 3 | Research Agent | Gathers public intel: tech stack, hiring signals, funding, and Why Now signals |
| 4 | LeadIQ Contact Agent | Finds and tiers contacts with verified email, phone, and LinkedIn via LeadIQ |
| 5 | SFDC Contact Validator | Checks 5 outreach guardrails, computes SalesLoft readiness, prevents duplicates |
| 6 | Value Agent | Synthesizes all data into a 7-tab account plan with a Lead Flow Map |
| 7 | KANSA | Quality gate: returns APPROVE, REVIEW, or HOLD with a confidence score |
| 8 | Mikey | Writes full omni-channel sequences for all tiers, pushes to SalesLoft via n8n |
| 9 | ABM Expert | Generates 3 LinkedIn ad variations per account ready for Campaign Manager |
What does the system actually produce?
Running /run-pipeline [account name] outputs four things:
- A 7-tab structured account plan, surfaced in a Lovable signal app
- Verified, tiered contacts with Salesforce guardrail checks
- Personalized SDR outreach sequences pushed directly to SalesLoft (Tier 1: 11 touches, Tier 2: 10 touches, Tier 3: 7 touches)
- 3 LinkedIn ABM ad variations per account, ready for Campaign Manager
Every outreach message is personalized per contact. This level of customization is scoped to enterprise accounts where the deal size justifies it.
What connectors does em-gine require?
| Connector | Used by | Swap options |
|---|---|---|
| Salesforce MCP | Internal Expert, SFDC Validator | HubSpot MCP, Pipedrive MCP |
| Gong MCP | Internal Expert | Chorus, Clari, or skip |
| LeadIQ MCP | LeadIQ Contact Agent | ZoomInfo, Clay |
| Supabase MCP | All agents | Any MCP-connected SQL store |
| n8n + SalesLoft | Mikey Outreach | HubSpot Sequences |
The team uses Salesforce in read-only mode for the CRM connection. Swap options exist for most connectors, so the pipeline is not locked to any single tech stack.
What is the quality gate and why does it matter?
Step 7 is KANSA, a dedicated QA agent. Before any account plan reaches an SDR, KANSA evaluates it and returns one of three verdicts: APPROVE, REVIEW, or HOLD, with a confidence score attached. A manager can process HOLD decisions using the approve-override skill, at which point Mikey's sequences fire.
Baking a hard quality gate into the pipeline rather than treating review as an afterthought is one of the more deliberate design choices in em-gine. It keeps reps from acting on incomplete or low-quality account plans.
How does the outreach sequencing work?
Mikey, the outreach agent, generates priority-based omni-channel sequences for all tiered contacts simultaneously. Touch counts by tier:
| Tier | Touches |
|---|---|
| Tier 1 (P1) | 11 |
| Tier 2 (P2) | 10 |
| Tier 3 (P3) | 7 |
Sequences push directly to SalesLoft via n8n webhook. The first step is set as manual, giving reps a review gate inside SalesLoft's native UI before the automated cadence fires.
Where can you get em-gine?
All three plugin files are available for download on the GTM Clawllective workflows page. Skills are also being shared through the GTM Clawllective library and GitHub repo. The system is modular: you can run the full pipeline or use individual skills in isolation, and you can customize the connector layer to match your stack.
View and download em-gine on the GTM Clawllective workflows page
Key Takeaways for GTM Teams
Start with your CRM data, not assumptions. Clustering ICPs from closed won/lost deals before writing agent prompts produces sharper targeting than starting from a blank brief.
Build a quality gate into the pipeline. KANSA's APPROVE/REVIEW/HOLD logic prevents low-quality account plans from reaching reps. If you are building your own pipeline, treat quality review as a step in the workflow, not a check that happens afterward.
Keep the first outreach step manual. Letting SDRs review and send step 1 inside SalesLoft before automation fires gives reps a meaningful checkpoint without breaking the cadence.
Modular design lowers the barrier to adoption. Shipping as three separate plugins means teams can start with just em-gine-core and add the intel and outreach layers when ready.
The visualization layer drives adoption. Surfacing the 7-tab account plan in a Lovable signal app makes the output immediately usable, not just an export to parse.
Want to Attend the Next Session?
GTM Clawllective is a practitioner community where GTM operators share the systems they are actually building and using. Sessions are not recorded, so attending live is the only way to catch the full discussion and Q&A. Join the community to get access to future sessions, the workflows library, and the Slack channel where members share prompts and hacks between sessions.
About the Speaker
Cece Renick is a Strategic Account Executive at Chili Piper, where she has spent over 7 years across enterprise account management, customer team leadership, and new business growth. Her work sits at the intersection of RevOps, marketing, and sales, and she has helped enterprise customers including ZoomInfo, Twilio, and Intuit drive conversion across the customer journey.
In this session, Cece presented em-gine, a pipeline architected by Emma Carey and built with the Strategic SDR team at Chili Piper. The system was the highest-rated submission at Chili Palooza 2026.