About us
We build AI agents for manufacturing. Allie integrates intelligence into factory operations: learning from machines, sensors and systems to detect problems, recommend actions and coordinate responses in real time. Our platform connects to PLCs, MES and ERPs through secure edge gateways. It transforms raw data into operational decisions that improve availability, quality and throughput. We work with large manufacturing groups in the food, beverage and CPG industries deploying AI agents that embed into production lines and scale across sites. Allie is built to act, not just report. From real-time monitoring to autonomous control, we help your factory think and self-optimize.
Sales, rebuilt as an engineering discipline. One operator, one agent stack, the entire revenue machine — for an AI company selling into the factory floor.
FULL-CYCLE · AI-NATIVE · BUILD THE MACHINE, OWN THE NUMBER
01 · THE THESIS
This role didn't exist three years ago
The old playbook — hire SDRs, buy twelve tools, pray the CRM stays clean — is being replaced. In its place: one or two operators who build the machine instead of being the machine.
They map the ICP and TAM programmatically. They watch every named account for signals and react the same day. They run outbound, inbound, events, and deal support as maintained systems, not repeated tasks. They spawn agents that verify their own output and loop until the outcome is achieved. And they do it for a fraction of what the equivalent sales-tool stack costs.
The profile is scarce: full-cycle sales fluency, technical instinct, and AI-native operation rarely live in the same person. This charter exists because Allie runs GTM this way — and is looking for the operator who wants to own it.
02 · THE ROLE
One operator, the entire top-of-funnel
Allie sells an operational copilot to Tier-1 manufacturers. The buyers are plant directors, VPs of operations, and CTOs at global brewers, food producers, and CPG companies — people who don't respond to spray-and-pray. They respond to precision.
This is not an SDR job and not a marketing job. It is one operator running the entire acquisition machine as a system: every list, every signal, every sequence, every call, every event, every piece of reporting — designed, automated, and maintained by you, with Claude Code and a small set of external APIs doing the work a six-person growth team used to do.
You own the number that matters: qualified hand-offs to technical sales. Everything below exists to feed it.
03 · WHAT YOU OWN
Six systems, one owner
SYS 01 · ICP & data foundation
- Map the full ICP and TAM: the named-account universe for brewing, food & bev, and CPG manufacturing across US, Mexico, and LATAM
- Build signal-based and lookalike lists off closed-won accounts; refresh continuously, not quarterly
- Track every champion who changes jobs and route them as a new account
- Own CRM architecture and reporting — pipeline truth lives here, and it is your truth to keep clean
SYS 02 · Signals & real-time reaction
- Watch named accounts for tech-stack changes, funding, layoffs, headcount swings, new plant announcements, and expansion CapEx — and react the same day
- De-anonymize website traffic, score it against closed-won, and push qualified visitors into sequence
SYS 03 · Outbound engine
- Set up and maintain the mail infrastructure — domains, warm-up, deliverability
- Build and run multi-touch sequences across email and LinkedIn
- Build the cold-calling engine: parallel dialing at scale — calls running back-to-back with no dead time, the next account queued while the current call is live, every outcome written straight into CRM. Calling is a throughput problem, and you engineer the throughput
- Score every inbound against past closed-won and auto-route qualified leads to technical sales
- Re-engage closed-lost on a systematic cadence
SYS 04 · Inbound & content
- Build the inbound system end to end: content filtered for the ICP, distributed across LinkedIn, amplified through a creator/operator network
- Own content management and the publishing cadence — a visible calendar, a pipeline from draft to published, a rhythm that doesn't skip weeks
- Run focused AEO efforts so Allie is the answer when a plant director asks an AI "how do I cut unplanned downtime"
SYS 05 · Events — an owned motion, not a side project
- Own the event calendar for the ICP — several shows are already locked for this year; you run those end to end and build the calendar forward from what they produce
- Pre-event: named-account target list per show, booking sequences 4–6 weeks out, meetings landed before the doors open — a booth with no calendar is a cost center
- On-site: every conversation captured into CRM same-day with context, not badge scans
- Post-event: 100% of captured contacts in sequence within 48 hours; qualified ones handed off within a week
- Report pipeline generated per event and kill the events that don't pay for themselves
SYS 06 · Deal support & feedback loops
- Auto-build a pre-call brief for every meeting on the calendar
- Generate the one-pager, ROI model, and proposal inputs per deal
- Analyze sales calls and feed patterns back into sequences, content, and qualification criteria
- Manage the sales collateral library — decks, one-pagers, case studies — versioned, current, and findable, so no one ever ships a stale asset
04 · HOW YOU WORK
Local-first. Agent-native. Human judgment on every send.
You build local-first. Every workflow starts on your own machine with Claude Code — no platform dependency, no waiting on anyone else's roadmap. If a workflow needs a platform to exist, it doesn't exist yet. As workflows prove out, they migrate into Rooster, Allie's internal GTM agent, so the engine becomes a company asset, not a laptop. Local Claude Code is where you build; Rooster is where it lands.
Every workflow is a system you build once and maintain, not a task you repeat. You spawn subagents that verify their own output, you loop until the outcome is achieved, and you treat "I did it manually" as a bug report against your own tooling.
Nothing client-facing auto-sends. You review and approve every external touch. Automation buys you volume; judgment is still the job.
05 · WHO YOU ARE
Three non-negotiables
TRAITWHAT IT MEANS HEREFull-cycle sales fluencyYou understand the whole cycle — prospecting through close through expansion. You've carried a number, or sat close enough to one to feel it.Technical by instinctYou can read an API doc, reason about a data model, and debug your own workflow at 11pm before a launch. You may not call yourself an engineer; you build like one.Al-native operationYou run AI as an operating discipline — agents that verify their output, loops that run until the outcome is achieved, workflows that cost a fraction of the equivalent SaaS stack.
Strong signals: you've built a growth system solo that a team later inherited; you've sold into manufacturing, industrial, or another long-cycle technical buyer; you've worked an event circuit and can show pipeline per show, not photos of the booth.
06 · THE FIRST 30 DAYS
The engine gets built. All of it.
Not phased, not sequenced, not "foundations first." By day 30, every system above is live: ICP universe in CRM, signal watching running, mail infrastructure warm and sequences shipping, the calling engine dialing, inbound pipeline producing, event calendar locked with meetings booked, pre-call briefs generating for every calendar meeting. Built locally with Claude Code, migrating into Rooster as each workflow proves out.
From day 31, the conversation is results. The measurement framework already exists — a weekly growth scorecard, built and run by the CRO, with qualified hand-offs and closes as the north star. You don't design the measurement system; you feed it.
07 · WHERE THIS GOES
The machine you build becomes the org you run
This role grows into technical sales leadership: managing both the growth function and the technical-sales AEs who carry deals from demo to close. The path is direct — first you build the acquisition machine, then you own the people and the machine together.
Most companies split this work across six tools and four people. Here it is one operator, one agent stack, and full ownership — and the ceiling on the role is whatever you make the machine capable of.