Product Manager, Applied AI (Staff

About the role

You will own the design and execution of AI workflow agents that automate the entire permitting lifecycle from document comprehension to submission and inspection tracking. This role requires you to translate complex construction workflows into reliable agentic behaviors while maintaining a sharp focus on user needs in the field. You will define the thresholds for reliable automation and build the guardrails that ensure submissions meet jurisdiction-specific standards every time. You will partner closely with operations to instrument real-world performance and convert errors into structured learning signals. This position gives you direct influence over how AI reshapes the daily reality of builders across North America. You will work at the intersection of product, engineering, and domain expertise to ensure agents reduce compliance time and de-risk project execution. Your decisions will shape not only internal processes but also the speed and reliability of housing, clean-energy, and infrastructure delivery.

Key facts

What you'll do

• Deliver our core set of workflow agents for permitting, enabling rapid comprehension of construction projects and accurate compliance submissions for permits and inspections.
• Define what "reliable enough to submit" means at both the field and submission level, and own the quality bar that gates release.
• Own our path to full automation, segmenting use cases and driving to single-shot submission where accuracy allows.
• Design the architecture for assistive use cases, building the agents and tools that empower permit coordinators to move faster where humans-in-the-loop is the best solution.
• Work with our operations org to build an evaluation and data flywheel through early testing, golden data set generation, and structured error analysis ahead of submission.
• Build a durable design system around our data flywheel, ensuring that users can easily provide feedback on our Agents & AI products.
• Build the internal tooling that allows our product team to iterate on agent behavior quickly, measure impact in production, and coordinate with engineering on releases.
• Partner with our go-to-market teams to surface agent capabilities that become key differentiators in customer conversations and adoption stories.
• Maintain a living map of jurisdiction-specific rules and workflows so that our agents can adapt to new requirements without extensive rework.
• Translate field feedback into product requirements that sharpen agent accuracy and reduce the cognitive load on our users.
• Define and track core metrics for agent reliability, cycle time, and user trust to guide prioritization and roadmap decisions.
• Collaborate with data science and engineering to align model capabilities with real-world deployment constraints and regulatory expectations.
• Champion best practices for prompt architecture, tool use, and deterministic checks that keep high-stakes submissions safe.
• Act as the primary liaison between deep domain expertise and the product team, ensuring that nuance is preserved in automated flows.

Requirements

• 8+ years of experience in product management, with a track record of delivering complex software products in regulated environments.
• 3+ years of hands-on experience building or shipping AI products, including agentic workflows or tooling.
• Demonstrated ability to work with messy, real-world data and turn unstructured inputs into deterministic product logic.
• Experience operating in construction, permitting, or another heavily regulated physical workflow domain is mandatory.
• Strong written and verbal communication skills to translate technical constraints into clear product requirements.
• Comfort working directly with customers and stakeholders to validate assumptions and refine solutions.
• Proven experience designing evaluation frameworks and quality gates for high-risk automated systems.
• Deep understanding of how AI models behave in production, including limitations around hallucination and edge cases.

Nice to have

• Experience with large language models, tool use patterns, and agent orchestration frameworks.
• Familiarity with construction-specific software such as Procore, Buildertrend, or similar platforms.
• Background in data flywheel design and continuous evaluation systems.

Practical notes

• Work arrangement is full_time with a hybrid schedule, requiring presence in the New York City office on Monday, Wednesday, and Friday.
• Candidates must be based in New York City or be open to relocation to support the hybrid model.
• No specific visa sponsorship details or travel requirements are outlined in the source material.
• No compensation figures, deadlines, or formal application pages are provided in the source material.