The opportunity

Autopilot is project44's no-code platform for deploying purpose-built workflows with AI agents into the mission-critical workflows supply chain teams run every day - calling carriers, collecting missing data, reconciling documents, resolving exceptions, and running mini-bids within guardrails. It is the product layer that sits on top of our agent portfolio (Freight Procurement, Disruption Management, Network Operations, Execution Recovery, Stockout Risk, and more) and gives customers the steering wheel: configurable triggers, transparent logic, audit history, and human checkpoints at every critical step.

This role owns three connected surfaces:

1. AI Workflows - the growing library of packaged, customer-deployable agent workflows (e.g., validate early ETAs, late-shipment carrier outreach, collect missing milestones, stale-position investigation). You define which jobs we automate next, in what sequence, and to what standard.

1. AI Agent Workflow Manager (fka Autopilot) - the no-code configurator itself: the trigger / condition / action canvas, workflow variants, multi-agent orchestration, human-in-the-loop controls, and the build-and-deploy experience that lets customers (and our own teams) ship workflows without engineering.

1. AI Agent Analytics & Reporting - the measurement layer (AI Agent Analytics, LunaIntel and LunaVoice dashboards, collaboration and carrier-performance reporting) that proves outcomes by use case and persona, exposes agent performance to customers, and closes the loop back into the roadmap.

The arc you'll drive: from support, to augment, to automate. The first generation of AI workflows supported users - surfacing the right signal at the right moment. The current generation augments them - taking a discrete action inside a human-run workflow (make the call, collect the milestone, draft the response). The mandate for this role is to push to the next stage: multi-agent workflows that automate complete work tasks end to end - coordinating several agents across a full job (e.g., detect 14 reach out 14 confirm 14 reconcile 14 write back 14 close) so an entire operational task runs without a human in the loop, while staying transparent, auditable, and reversible. You'll own that progression use case by use case, deciding when a workflow is ready to move from augmenting a person to automating the task outright.

Natural-language workflow authoring with Mo. Mo is project44's AI Supply Chain Analyst - a conversational agent embedded in Movement that lets users ask questions of their own data in natural language (NL14SQL / NL14API), grounded in their shipments, business rules, and carrier history. Mo's roadmap runs from search to analyze to act, and Autopilot is the 22act.22 A core part of this role is making Autopilot workflows authorable and executable through Mo in plain language - 22set up a workflow that calls the carrier whenever an FTL shipment is running two hours late and update the ETA22 - and surfacing Autopilot task history and run status back inside Mo's answers. You'll own the Autopilot side of that integration and partner closely with the Mo product management team on the shared experience, the trigger/condition/action vocabulary Mo maps natural language onto, and the guardrails for letting a conversational agent stand up an autonomous workflow.

We're moving fast - from shipping roughly one workflow a week to one a day - and the bar is high: very little vibe-coded software is production-ready, and customers only adopt agents they trust. You'll set the throttle, use case by use case, so speed never outruns trust.

What You'll Do:

Lead with customers and research

• Own the customer problem before the solution. Every workflow starts from a clearly stated customer problem, who is impacted (planners, logistics managers, appointment and yard managers, carrier dispatch, drivers), and when it occurs - not from a feature idea.

• Run primary research continuously: customer interviews, ride-alongs with operations teams, design-partner pilots, Customer Advisory Board (CAB) validation sessions, win/loss and churn intake reviews, and direct analysis of platform behavior.

• Recruit and manage design partners for shadow-mode pilots - where the agent logs what it would do before it acts - to establish honest baselines and earn trust ahead of live deployment.

• Be the domain and product expert in customer-facing settings: demos, executive briefings, CAB, and conferences. Translate what you hear into a prioritized, defensible roadmap.

Drive AI innovation

• Push the frontier of what agents can safely do in production: autonomous voice and email outreach, document parsing and reconciliation, reason-code classification and write-back, and multi-agent workflows that coordinate several agents across a single business outcome.

• Move workflows up the maturity curve - from supporting a user (surfacing a signal), to augmenting them (taking one action in a human-run flow), to automating a complete work task (a multi-agent workflow that runs the whole job end to end). Define, for each use case, the bar a workflow must clear to graduate to the next stage.

• Make workflows authorable and executable in natural language through Mo, project44's AI Supply Chain Analyst - so a user can describe a workflow conversationally and have Autopilot stand it up, run it, and report back. Own the Autopilot side of the integration and the mapping from natural language to triggers, conditions, and actions, partnering with the Mo product management team.

• Design for trust: configurable controls, transparent logic, audit trails, intervention points, hallucination guards, and throttles tuned per use case. Decide where humans stay in the loop and where agents can act autonomously.

• Partner with engineering, applied AI, and design on workflow architecture - triggers, conditions, actions, contact-resolution strategy, retry and cadence logic, and closure semantics - and on the tooling that lets us scale workflow production toward one per day.

• Stay ahead of a fast-moving competitive field of agentic logistics startups; know precisely why project44's network and context are the durable advantage and build the product to exploit it.

Write outcomes-based requirements

• Author crisp PRDs along with rapid prototypes. framed around goals and non-goals, explicit success/failure metrics, and leading and lagging indicators - not feature checklists. (A workflow marked 22completed22 is not the same as a workflow that succeeded; you'll define success by the outcome it produced.)

• Specify configurability deliberately: what is a sensible pilot default versus what each tenant must be able to tune (thresholds, conditions, allow/block lists, cadence, channels).

• Maintain a prioritized backlog across the three surfaces and sequence it against customer value, trust gating, and business results. Synthesize complex, multi-mode use cases (FTL, LTL, ocean, drayage, intermodal) into an actionable roadmap.

• Hold the gating bar: data availability, provider readiness, legal/compliance review (e.g., TCPA and calling-hours guards for outbound contact), and human-QA thresholds before write-back or autonomous action is unlocked.

Measure and report outcomes

• Define the metric model for every workflow before it ships, and instrument it: validation/completion rates, outcome classification confidence, reduction in manual coordination and exception handling, accuracy improvements (e.g., ETA MAPE/MAE), freight-spend and disruption-cost impact, response and reach rates, and adoption.

• Own AI Agent Analytics and the LunaIntel / LunaVoice reporting experience so customers can see agent performance and outcomes by use case and persona - and so we can prove ROI in renewals, QBRs, and executive reviews.

• Run the outcome loop: turn what the dashboards reveal (non-response patterns, ambiguous outcomes, value by segment) back into roadmap decisions, throttle changes, and the next workflows to build.

• Produce high-quality, executive-ready deliverables - investment memos, roadmap reviews, launch readouts, and enablement - with the same attention to detail you bring to the product.

What success looks like in the first year

• A steadily expanding, high-trust AI Workflow library shipping at an increasing cadence, with each workflow tied to a measured customer outcome.

• At least one flagship multi-agent workflow that fully automates a complete operational task end to end - moving a meaningful job from human-run to agent-run without eroding trust - with a clear, repeatable bar for graduating future workflows from support, to augment, to automate.

• Autopilot workflows authorable and executable through Mo in natural language, with adoption measured by workflows created and triggered via Mo - shipped in partnership with the Mo product management team.

• Measurable business impact from deployed agents - in the range project44 already demonstrates today: meaningful reductions in freight spend and manual coordination, double-digit reductions in manual exception handling, materially improved ETA accuracy, and faster sourcing cycles.

• AI Agent Analytics adopted as the system of record for agent performance - used by customers, CX, and the executive team alike.

• Rising adoption and trust among Autopilot customers, reflected in NPS and renewal/expansion, with no trust-eroding incidents from agents acting beyond their guardrails.

What we're looking for

• 5+ years in product management (more for Principal level), including hands-on ownership of a technical, data-rich, or AI/ML product through the full lifecycle - discovery, definition, GTM, and iteration.

• Demonstrated customer obsession: a track record of grounding product decisions in direct research and of representing the customer credibly to engineering and to executives.

• Fluency with AI / agentic systems - LLM-powered agents, orchestration, evaluation, human-in-the-loop design, and the practical realities of getting non-deterministic software production-ready and trusted.

• An outcomes-first operating style: you write requirements as goals, non-goals, and success metrics, and you instrument and report on impact rather than output.

• Strong analytical skills - comfort defining metrics, working in dashboards and warehouse data (e.g., Snowflake-backed analysis), and reasoning quantitatively about agent performance and ROI.

• Excellent written and verbal communication; able to influence engineers, designers, sales, partners, customers, and corporate leadership, and to produce polished executive deliverables.

• Comfort with ambiguity and speed; able to set the right throttle so the team ships fast without outrunning customer trust.

Nice to have

• Logistics, supply chain, or transportation domain experience (visibility, TMS, YMS, procurement, carrier networks).

• Experience building no-code / workflow-builder, automation, or analytics-and-reporting products.

• Experience with conversational / natural-language interfaces (LLM chat, NL14SQL / NL14API) and shipping a shared experience across two product teams.

• Familiarity with voice/communications platforms or outbound-contact compliance (TCPA, calling-hours rules).

• Prior work running design-partner programs, CABs, or beta/pilot motions for net-new product categories.

• • In-office Commitment: Employees are expected to contribute to our collaborative culture by working in the office FOUR days weekly