AI Enablement & Workflow Innovation Lead

This role exists to make Canopy an AI-native company in how we build, operate, support customers, and make decisions. This is a hybrid role in the New York City metro area.
About the Role

We're hiring an AI Adoption Accelerator: someone who lives at the frontier of agentic AI and can pull the rest of the company up to that frontier with them.

You do not need to be a career software engineer. You do need to be the person who's already built a dozen agents, knows how the current frontier models differ in practice, has strong opinions about context engineering, can stand up an MCP server in an afternoon, and won't ship an agent without an eval behind it. And you need to be able to sit next to someone who has never written a prompt and leave them able to build the next workflow themselves.

The job has two halves, and they matter equally:

• Build. Embed with teams across Canopy (CX, Sales, Marketing, Product, Engineering, Operations, Finance, People), map their work, and turn the highest-leverage workflows into real AI tooling. Sometimes that's a Claude Project with the right skills and connectors. Sometimes it's a multi-step agent wired into our systems. Sometimes it's a sharp prompt that ends a problem someone has been dragging through their day for a year.
• Teach. Every build is also a tutoring session. By the time you ship something with a teammate, they should understand enough to build the next version without you. The goal is not to become the bottleneck, it's to leave behind people who can pattern-match on their own, and to compound Canopy's capability with every workflow.

What you will own

• Building & shipping AI workflows. Design, build, and deploy AI workflows that range from lightweight Claude Projects with custom skills to full multi-step agentic systems. Build and integrate MCP servers and connectors so our agents reach into the tools people already use: Slack, Salesforce/HubSpot, Zendesk, Gong, Google Workspace, internal APIs, our own products. Reach for the right tool for the job, including no-code tools when that's genuinely all a problem needs, without over-engineering.
• Operating rhythm & experimentation. Establish a lightweight model for how AI experiments happen at Canopy: intake, prioritization, concise experiment briefs, demos, documented learnings, and clear scale-or-kill decisions. Partner with the early-adopter group to pick 2-3 high-impact workflows per quarter. Run a regular demo cadence where teams show what they built, what worked, and what others can reuse.
• Enablement & teaching. Build workflow-based learning experiences that help people use AI in their actual day-to-day work, tailored by role. Run office hours, workshops, and "build with me" sessions. Produce reusable assets (prompt libraries, workflow templates, playbooks, starter kits, demo recordings) that let teammates extend AI work without you. Teach not just how to use AI, but when to, when not to, and where human judgment has to stay central.
• Governance, evals & responsible use. Develop eval practices so we know our agents actually work, and keep working as models change underneath us. Partner with Security, Legal, and People to keep AI usage safe and compliant, with real guardrails around PHI, customer data, and regulated workflows. Build the habits that prevent overreliance, quality drift, and unclear ownership.

What Success Looks Like

• In the first 90 days, you'll understand Canopy's priorities, systems, and team pain points; stand up the experimentation operating rhythm with the early-adopter group; build an initial cross-functional use-case inventory; ship several real AI workflows with named owners and eval-backed success criteria; and launch the first version of Canopy's internal AI hub.
• In the first 6 months, you'll have shipped and scaled meaningful workflows across multiple departments, created a measurable lift in how confidently people use AI, reduced manual work in priority workflows, and established a repeatable path from idea 10 experiment 10 demo 10 adoption 10 scale with the evals to prove the agents hold up.
• In the first year, you'll have built a durable internal capability for AI building, enablement, and adoption; demonstrated measurable business impact through faster execution, better quality, and less manual work; and helped create a culture where teams don't just "use AI" but continuously rethink how the work should be done.

What You Bring

• Frontier LLM fluency. Deep, current, hands-on experience with Claude, GPT, Gemini, and open models. You can explain how they differ in practice, not just on benchmarks.
• Agentic systems experience. You've built real things with tool use, planning, memory, and multi-step orchestration, and you've debugged them when they went sideways.
• MCP fluency. You've written MCP servers, integrated clients, and built connectors into the systems teams actually use.
• Context engineering instinct. Strong, defensible opinions about what goes in the window, what gets retrieved, summarized, or evicted, and when to break your own rules.
• Evals discipline. You don't ship on vibes. You build eval harnesses and can articulate why your agent works, or why it doesn't yet.
• Enough code to be dangerous. You write TypeScript or Python to glue real systems together, even if you'd never call yourself a software engineer.
• A teaching gene. Patient, able to meet people where they are, and genuinely invested in their learning. You can explain a hard concept to someone twice your seniority without making them feel small.
• Operator judgment. You turn ambiguous business problems into scoped experiments, you drive adoption through trust and usefulness rather than authority, and you document well.
• Clear-eyed on risk. Strong instincts around data privacy, quality control, and responsible use. Excited by the potential, honest about the failure modes.

Nice to Have

• Healthcare, health tech, or HIPAA experience
• Built or contributed to public MCP servers, agent frameworks, skills, or open source
• A public presence in the AI / agentic community (writing, talks, OSS)
• Experience standing up an internal AI adoption or responsible-AI program
• Familiarity with our stack: TypeScript, MongoDB, BigQuery, GCP, Kubernetes

You May Be a Great Fit If

• You're hands-on enough to build the first version yourself, and disciplined enough to make it teachable
• You believe demos are one of the fastest ways to create momentum
• You know adoption is a behavior-change problem, not a technology rollout
• You like turning messy problems into structured experiments
• You can move between strategy and execution without getting stuck in either
• You have high curiosity, low ego, and a bias toward learning by building

Why This Role Matters

Most companies trying to "do AI" are stuck running pilots that never reach production or buying tools their teams never use. We're betting on a different path: put serious capability in the hands of the people closest to each problem, and make sure the agents we ship actually work and keep working. Your job is to make that real, one teammate, one workflow, one eval at a time and to build the muscle that lets Canopy keep rethinking how work gets done.

The base salary range for this position is $200k - $225k per year, depending on location, experience and qualifications.