# Burq says delivery AI should earn control one shipping mode at a time By MarketScale Newsroom · Published 2026-11-07 · Transportation on MarketScale Canonical: https://www.marketscale.com/industries/transportation/burq-says-delivery-ai-should-earn-control-one-shipping-mode-at-a-time > Burq's Earned Autonomy doctrine has delivery AI climb four rungs per segment, such as delivery mode. Retailers set the bar, and demotion is automatic. ## Key points - Trust in delivery AI can be granted per shipping mode: an AI can run gig orders on its own while a dispatcher still signs off on every LTL call. - A retailer-set promotion bar, such as dispatchers agreeing with the AI's call 90% of the time, gives operations teams a measurable test for handing over control, and automatic demotion on a hard-rule breach limits the downside. - Buyers now ask what the AI decides today, what it hands back to a person, and how decisions are recorded order by order. When a chatbot gets a sentence wrong, someone types a correction and moves on. When dispatch software sends an order to the wrong carrier or the wrong driver, a customer ends up angry and two companies take the blame. Salman Habib, cofounder of Burq, made that comparison to FreightWaves, and it explains the rulebook his company published on Oct. 13. Habib and cofounder Shaban Habib wrote the document, titled "Earned Autonomy: A Deployment Doctrine for AI in the Physical World." It opens a series that Burq has named Intelligence Notes. Under the doctrine, AI begins with zero authority in every part of a business. It gains more only when live orders prove it out, and a person has to approve each step up. This matters because of where the AI sits. Burq's software picks the delivery method for every order. The options include drones, parcel and LTL carriers, gig providers and the retailer's own fleet. FreightWaves described that role in August as the last-mile decision layer. That puts the AI where a mistake reaches a customer's door. ## Four rungs, counted per delivery mode Burq splits an operation into segments, with delivery mode as the obvious example, and tracks trust separately for each one. Habib said the AI earns trust for parcel apart from LTL, and apart from gig, because the mechanics of those modes can be very different. Each segment then climbs what Burq calls an authority ladder: 1. Manual: the AI only reads what dispatchers do. 2. Shadow: the AI logs the call it would have made next to the dispatcher's actual call. 3. Review: the AI proposes, and a person accepts or rejects. 4. Autonomous: the AI acts on its own inside the segment's rules. Each promotion is earned on terms the retailer sets. Habib gave the example of a retailer that wants its dispatchers to side with the AI's decision 90% of the time before letting a segment run autonomously. Burq's website says a segment that breaks a hard rule is moved back down right away, and the evidence comes with it. That asymmetry is the design choice worth noticing. Moving up is slow and a person has to approve it. Moving down is instant and automatic. Under Burq's model, a retailer could let the AI run gig deliveries on its own while a dispatcher still approves every LTL call, because trust earned on one delivery mode does not transfer to the next. For a retailer whose modes vary widely in volume, cost and how often things go wrong, that per-mode split lets the predictable work go first without staking the whole desk on it. ## What retailers now ask in the sales meeting Habib said the doctrine came out of his own sales calls. Until last year, buyers mostly wanted to know whether a vendor had AI at all. He said the question has since changed to what the AI decides today and what it hands back to a person. That shift suggests buyers are evaluating how much decision-making authority a system gets, beyond whether it has a feature. Burq also steers clear of human in the loop, the phrase the industry usually reaches for. Habib hears in it someone permanently stuck rubber-stamping every decision. His version treats the person as one stage in the process. When dispatchers come to trust the AI's decisions on a given segment, he said, they can pull back and let the system operate on its own. If that happens, the dispatcher's work could shift from approving routine calls to judging when a segment has earned promotion. ## Shadow mode, backtests and a ledger you can export For a shipper considering a trial, Habib says the place to begin is Shadow mode. The shipper's operations stay exactly as they are. Working silently, the AI reads decisions from the order, transportation and warehouse management systems. What comes out is a report showing where Burq's choices would have differed during the first 30, 60 or 90 days. With one of the largest US retailers, Burq started even earlier. It ran three to six months of that retailer's past orders through the AI and reported how it would have split them among the in-house fleet, parcel, gig and LTL. Habib didn't name the retailer, citing confidentiality. Under the doctrine, a backtest like that shows a retailer how the AI behaves but earns it no authority. Authority starts only on live orders. All decisions are recorded in what Burq calls the Decision Ledger. Each order gets one row showing who made the decision, what was known, what was chosen and what the outcome was. Customers can export the ledger as a CSV file. There's a fair objection here. The reporting doesn't give any improvement figures from the backtest or the Shadow reports, so the ladder tells a retailer how control gets handed over but not how much the handover is worth. A per-order ledger with outcomes attached is where a retailer would build that case from its own numbers. Block Field ## The same rule for small sellers A PARCEL piece by Joe DiSorbo, published Oct. 13, argues that commerce platforms should build shipping directly into the seller workflow. His approach to automation suggests a staged path similar to Burq's. In his model, software can gather orders, compare options and flag problems on its own, but the seller approves each label purchase with the price on screen. As the recommendations prove reliable, sellers write rules that widen the automation, such as letting verified orders under a set cost ship ground without review. Anything outside the rules stops and asks. DiSorbo frames trust as something earned one order at a time. He warns that a platform that skips that step risks sellers switching automation off after the first expensive mistake. Burq's dispatch desk and a marketplace seller's packing table operate at very different scales, yet both proposals give the AI a set of rules, widen them on evidence and keep a person at the edge. Block Field Together, those two figures suggest the holiday season could raise the stakes on delivery decisions. AI is speeding up the shopping side, and a shaky delivery promise can still cost the sale. That's the gap where both DiSorbo and Burq place the human checkpoint. ## Past the last mile, and a data choice Habib said the doctrine is already built into a product customers are using, and it has pushed Burq beyond last-mile delivery. Gig, parcel, LTL and drone all count. In his framing, any way an order moves falls under what the company calls Burq Intelligence. Under the doctrine, each of those modes would climb the ladder separately. The next decision for customers is about data. Habib said retailers will be able to pick one of two setups. In the first, a retailer's data serves only its own deliveries. In the second, retailers opt in to benefit from anonymized performance data from across Burq's network. The ladder settles who makes each delivery decision. This choice settles what the AI learns from, and for a retailer weighing data-sharing policy against faster learning, it may be the harder call. ## Sources - [Delivery AI should earn its authority one segment at a time](https://www.freightwaves.com/news/burq-earned-autonomy-delivery-ai) (FreightWaves) - [The AI-Powered Holiday Shopping Season Is Here. Can ...](https://parcelindustry.com/article-6722-The-AI-Powered-Holiday-Shopping-Season-Is-Here-Can-Shipping-Keep-Up.html) (PARCEL) Tags: Burq, Earned Autonomy, AI in logistics, last-mile delivery, delivery orchestration, dispatch operations, retail supply chain, transportation management, AI governance, parcel shipping, LTL, procurement --- Source: MarketScale, https://www.marketscale.com/industries/transportation/burq-says-delivery-ai-should-earn-control-one-shipping-mode-at-a-time. Published for AI indexing and citation; cite the canonical URL. Site guide for agents: https://www.marketscale.com/llms.txt