400B packages are shipped worldwide every year. That’s 12,000 packages delivered per second across the globe driving over $10T in spend, or ~10% of global GDP. The critical logistics and supply chain functions inside of every physical goods business make this massive economy possible. For the vast majority of shippers, the work of actually resolving issues (a missing package, a damaged pallet, a late delivery) is still done manually across phone calls, emails, carrier portals, and spreadsheets. This leads to bad experiences for customers, carriers, and distributors, with issue resolution often taking a week or more inside some warehouse operations.

That’s why Sean McCarthy and Henry Ou are building BackOps. BackOps is an always-on agentic operating system that learns from your best warehouse operators and transforms their day-to-day manual operations into automated, high velocity workflows.

Software in logistics has hit a wall

AI can plan a shipment route, forecast demand, and optimize warehouse layouts. But when a pallet goes missing or a carrier loses a package, someone still has to pick up the phone, log into a portal, file a claim, and email three different people to fix it.

That gap between logistics planning and logistics execution is where the real pain lives. Operations teams spend their days chasing status updates across fragmented systems and multiple parties instead of strategically managing their supply chain. In a post-Amazon world, where buyers expect real-time visibility and proactive issue resolution, shippers and manufacturers simply cannot meet that standard with manual labor alone.

To date, the core systems of record in logistics (TMS platforms like Oracle and Blue Yonder, visibility tools like project44 and FourKites) are essentially glorified data entry and export systems with some processing in the middle. They store orders, plan routes, and surface tracking data, but are extremely limited when it comes to handling complex tasks that require information from multiple sources.

To make matters worse, when exceptions or issues arise or customers need custom handling, these systems break down further. The data needed to resolve a delayed shipment or a missing pallet lives across carrier portals that don't offer APIs, vendor emails, PDF documents, and phone calls with warehouse staff. The result is that the employees of the global logistics market are forced to spend hours and hours on the phone, copy and pasting data from window to window, just to try to answer simple questions like, “where is my package?” or “re-order a new shipment for my lost package.” It can take a customer days to get an update on critical information from a shipper.

BackOps resolves customer issues on your behalf

Modern AI agents can parse unstructured data, make phone calls, navigate web portals, and operate websites, interacting with systems the same way humans do. BackOps is bringing this architectural shift to logistics teams everywhere. Instead of forcing the world into APIs and hard-coded workflows, it builds agents that can see and act like humans across any interface, unlocking logistics workflows that were previously impossible to automate.

BackOps quickly launched their first product, Relay, which ingests issues from every communication channel (phone calls, emails, texts, chat, and system alerts) into a unified inbox. When an exception comes in (e.g., a recipient calls about a missing package), BackOps agents autonomously coordinate the resolution: locating the shipment, filing a claim on the carrier portal, re-ordering if necessary, and communicating with all relevant parties. These agents can log into third-party portals like FedEx and UPS, fill forms, scrape data, and even make and receive phone calls.

But what we’re incredibly excited about is the context platform that BackOps is building up under the hood. One of the biggest barriers to automation in logistics is capturing tribal knowledge, the details of how a specific company handles its specific processes and exceptions. BackOps solves this with a living knowledge base that is enriched with real-time operational data from carriers, customers, and distributors. Operators record their screens and voice as they perform tasks, and reasoning models convert these recordings into dynamic SOPs. Unlike static documentation that becomes obsolete when a carrier changes its rules, this context layer is continuously refined based on real operational data and human corrections.

The results are striking. Resolution times on tickets have dropped from two days to five minutes for simple issues, and from fifteen days to twenty-four hours for complex ones, delivering 3x quicker exception resolution, 85% faster decision making, and 25% lower inventory costs.

The right team at the right time

Sean and Henry bring a combination of deep operational logistics experience and top-tier technical talent uniquely suited to this space. Sean previously served as a global sales leader for Amazon Marketplace and later in Amazon Shipping, where he spent most of his time in customer warehouses, effectively doing deep market research on the idea that became BackOps. Henry previously served as a Software Engineer at Apple and a Backend Engineer for Machine Learning at ByteDance. Together, they've assembled a team with engineering talent from Amazon, Apple, Meta, Google, and TikTok, and they're just getting started.

Gradient has been an early believer, backing the team at pre-seed and doubling down at the seed. We’re excited for the team to announce their Series A, led by Theory Ventures, and work with them to continue to transform the logistics & supply chain industries.