AI Implementation Specialist

The 30-second version

LeadrPro is a B2B marketplace where software sellers meet verified, paid buyers. Our AI agent, LP, now lives inside our customers’ Slack workspaces — surfacing matched buyers, drafting outreach, triaging demo requests, and pushing meeting prep into the channels where their teams already work.

Every customer’s Slack agent is only as good as the prompts behind it. Right now, a human has to sit with each account, learn their ICP, their voice, and their sales motion, and tune the agent until it earns its place in the channel.

That human is you.

You’ll be embedded with real revenue teams, writing and testing the prompts that make their agents useful — then turning what works into templates so the next 100 customers get it in an hour instead of a week.

This is not a research role. You will talk to customers. You will ship on day three.

What you’ll actually do

Onboard customer agents. Run configuration sessions with new accounts, extract their ICP, tone, product positioning, and edge cases, and translate all of it into working system prompts and tool configurations for their Slack agent.

Write, break, and fix prompts. Draft prompts, build small eval sets from real customer scenarios, run them, find where the agent hallucinates, over-hedges, or sounds like a robot, and fix it. Version everything. Document why each change worked.

Own the feedback loop. Read agent transcripts. Find the recurring failure modes across accounts. Bring the top three to the product team every Friday with a proposed fix, not just a complaint.

Turn one-offs into product. The job is to make yourself unnecessary. Every bespoke prompt you write should end up as a reusable template, a variable, or an onboarding question. We measure you on how much of your work stops being manual.

Be the customer’s person. When an account says “the agent isn’t getting it,” you’re the one who responds, diagnoses, and closes the loop — usually the same day.

What good looks like

Day 30 — You’ve onboarded 5+ customer agents end to end without a hand on your shoulder. You’ve built our first shared eval set.

Day 60 — Median time-to-first-value for a new Slack agent is under 24 hours, down from where we started. You’ve published a prompt library the rest of the team uses.

Day 90 — At least 60% of new agent configurations are handled by templates you built, not by a person. You’ve shipped at least one prompt change that measurably lifted a customer metric — reply rate, meetings booked, or agent messages actually acted on.

Who this is for

• You’ve built something real with an LLM API — a side project, a tool, an automation. Not a certificate. A thing that ran.
• You write clearly. Prompt engineering is 80% writing, and vague writers produce vague agents.
• You can read and edit JSON, understand what a tool call is, and debug why one failed. You don’t need to be an engineer, but you can’t be afraid of a terminal.
• You’re comfortable on a Zoom with a VP of Sales who is skeptical of AI and short on time.
• You’re allergic to leaving a problem half-diagnosed.

Who this isn’t for

• Anyone who wants to “explore AI strategy.” We need output, not frameworks.
• Anyone who needs a spec before starting. You’ll get a customer, a goal, and a deadline.
• Anyone who would rather not talk to customers. Half this job is listening.

Bonus points

Experience with Slack’s API or bot framework · evals or LLM observability tooling (Braintrust, LangSmith, or homegrown) · B2B SaaS or sales ops background · you’ve worked a quota-carrying or SDR seat and know why sellers ignore tools

Pay: $15.00 - $25.00 per hour

Expected hours: 20.0 – 30.0 per week

Work Location: Remote