SAN FRANCISCO--(BUSINESS WIRE)--#AI--Burq, the delivery decisioning platform, today published Earned Autonomy: A Deployment Doctrine for AI in the Physical World, the first release in its new Intelligence Notes series. Authored by co-founders Salman Habib and Shaban Habib, the note lays out the thinking behind Burq Intelligence: how Burq decides, dispatches, and defends the best way to deliver every order in real time, and how an AI system earns the right to make that call on its own instead of just suggesting it.
The note argues that the way AI is sold to commerce operations today has the order of operations backward. Most systems are deployed with full authority from day one: connect the account, set the objective, and trust the model, with evaluation deferred to a benchmark score. That pattern may be tolerable when a model's mistakes are limited to a bad sentence. It is not tolerable when a model's mistakes commit real dollars to a provider and attach a broken promise to a real customer. A wrong dispatch is a refund, a missed window, and a customer who remembers.
Earned autonomy proposes granting authority step by step instead. Under the doctrine, an AI system starts with zero authority in every part of the business and has to earn trust separately for each one, local delivery, scheduled installs, and so on, since being reliable in one doesn't prove it's reliable in another. It proves itself by working alongside the existing human process on real, live orders, not on historical replays, and building a track record. Once it clears a performance bar that's agreed on in advance, a person still has to sign off before it's handed more authority, and before it's trusted to act fully on its own, the evidence must show the orders it decided don't fall behind the team's by more than a small margin agreed in advance. If a breach of a hard rule is detected, its authority is pulled on the spot. Every decision, and the reasoning behind it, is logged, so it can be reviewed later. The approach mirrors how other industries already handle high-stakes automation: airline autopilot systems earn wider authority only after monitored flight hours, and self-driving cars are trusted with fewer humans in the loop only after millions of supervised miles.
"A year ago the first question in our meetings was 'do you have AI?' Now it's 'what does it decide on its own, and what does it hand to a person?' That is the right question," said Burq founders, Salman and Shaban. "Earned autonomy is our answer to it. We're not asking retailers to trust an AI because a vendor says it's good. We're asking them to watch it earn the right to act, one part of the business at a time, with a record that proves it."
The note ships with an illustrative worked example of the doctrine running inside Burq Intelligence, Burq's decision layer, which is built on the Decision Graph: a live record of how delivery providers, zones, hours, and order types have actually performed across every operation Burq runs. The example shows the evidence board an operations team reads for each part of the business: how often the AI agrees with the human team, how often its disagreements turn out to be the better call, what those calls are worth, and how often its authority has been pulled back. The note treats that last number as a feature. A safety switch that never trips, the Habibs note, is one to worry about. Earned Autonomy: A Deployment Doctrine for AI in the Physical World is the first in Burq's Intelligence Notes series, where Burq shares how it builds and deploys AI for commerce and logistics.
About Burq
Burq is the delivery decisioning platform. For every order, Burq decides, dispatches, and defends the best way to deliver it, in real time, across a network of hundreds of delivery providers. Burq Intelligence is the decision layer behind that call, built on a live Decision Graph of how providers, zones, hours, and order types actually perform. Burq was recently named to FreightWaves FreightTech 100, Fast Company's Best Workplaces for Innovators list in 2025 and Inc. Power Partner in 2025.
Contacts
Media contact:
Brielle Patterson
marketing@burqup.com
