Briefback or Intent Confirmation¶
Protocol — instantiates Agentic Control Loop Design
Before acting, the actor restates the goal, constraints, and plan back to the tasker to confirm shared understanding and surface conflicts early.
The Briefback or Intent Confirmation is a pre-action handshake: after being tasked, the actor says back in their own words what they understand the goal to be, what constraints bind them, and how they intend to proceed — and the tasker confirms or corrects before anything is done. Its defining move is that it verifies the actor's representation of the goal, not the tasker's instructions: the failure it catches is a confident actor about to execute a subtly wrong objective, competently. Because restating the goal against real conditions is exactly where hidden conflicts surface — "you want it fast and under budget, but the fast option costs more" — the briefback is also the moment those conflicts get named and resolved rather than discovered mid-action. It is a check on understanding before execution; it neither enumerates the permitted actions nor runs any of them.
Example¶
A field-service technician is dispatched to a manufacturing plant with a work order: "restore the packaging line to full speed by end of shift." Before touching anything, she runs an intent briefback with the dispatcher over the radio. Her restatement: "Goal as I read it — line running at rated speed before 3pm, and I'm assuming a temporary fix is acceptable if a permanent one needs a part we don't stock. Constraint: I can't take the line fully down during the active production window without your sign-off."
That restatement immediately surfaces a conflict the work order hid: the real priority is today's shipment, so the dispatcher confirms a temporary fix is fine and, more importantly, that a two-hour full shutdown is not — better a slow line than a stopped one. Had she skipped the briefback, she would have competently done the "right" thing (a proper repair requiring a shutdown) and blown the shipment. The protocol changed nothing about her skill; it corrected her representation of the goal and resolved the speed-versus-uptime conflict before the wrench moved.
How it works¶
- Restate, don't repeat. The actor paraphrases the goal in their own words rather than reading the order back; a paraphrase reveals the understanding, a recital only proves the words arrived.[n1]
- Confirm goal, then constraints, then plan. Work from the objective outward to the bounds and the intended approach, so a misread goal is caught before effort is spent detailing a plan for the wrong target.
- Surface conflicts out loud. The restatement deliberately names tensions the tasking glossed — competing priorities, an assumption the actor is making — so they are resolved by the tasker now, not improvised later.
- Close with an explicit confirm or correct. The protocol ends with the tasker either confirming the understanding or correcting it; an unanswered briefback has confirmed nothing.
Tuning parameters¶
- Depth — a one-line confirm versus a full restatement of goal, constraints, and plan. Deeper catches subtler misreads and more conflicts but costs time; shallower is fast but only catches gross misunderstanding.
- Trigger scope — every task versus only novel, ambiguous, or high-stakes ones. Briefing everything guarantees coverage but breeds ritual box-ticking; briefing selectively saves time but relies on correctly spotting which tasks are risky.
- Conflict-surfacing pressure — how actively the actor is expected to raise tensions rather than smooth them over. More pressure catches buried conflicts but can feel like pushback; less keeps things quick but lets conflicts detonate mid-action.
- Formality — a verbal exchange versus a written confirmation. Written leaves a record and forces precision but slows the tempo; verbal is fast and fluid but leaves the shared understanding unrecorded.
When it helps, and when it misleads¶
Its strength is catching the most expensive error cheaply: a competent actor executing the wrong goal wastes everything downstream, and a sixty-second restatement catches it before any of that is spent. It shines exactly when instructions are terse, the environment is fluid, and the tasker cannot specify every contingency — the situation the mission-command tradition built the backbrief for. As a byproduct it drags goal conflicts into the open at the one moment they are still cheap to resolve.
Its failure mode is degrading into ritual compliance: the actor recites the order back verbatim, both parties nod, and nothing is actually verified — a briefback that confirms the words while leaving the misunderstanding intact.[n1] It can also become a bottleneck if every trivial task demands a full confirmation, training people to rubber-stamp. And an intimidated actor will confirm understanding they do not have rather than admit confusion, which quietly defeats the whole point. The guarding discipline is to insist on paraphrase over recital, reserve depth for tasks that warrant it, and make it safe to say "I don't actually understand the goal yet."
How it implements the components¶
The briefback realizes the archetype's goal-alignment slice — confirming the actor's objective and resolving conflicts before execution:
represented_goal— it verifies and corrects the actor's own representation of the goal, ensuring the objective driving the action matches the one intended.goal_conflict_resolver— the restatement is where competing priorities and hidden tensions surface, and the tasker's confirmation is where they are resolved, before the conflict is discovered mid-action.
It confirms the goal but does not fix what actions are in bounds — the action_repertoire and legitimate_action_boundary the actor may draw on are Safe Action Menu's, and the world_model assumptions behind the plan are Model Assumption Register's. The briefback is the pre-action handshake on the objective; the menu is the standing list of permitted moves.
Related¶
- Instantiates: Agentic Control Loop Design — it verifies the represented goal and resolves conflicts before the loop runs.
- Sibling mechanisms: Agency Loop Map · Agency Health Dashboard · Model Assumption Register · After-Action Learning Cycle · Safe Action Menu · Graduated Autonomy Ramp · Decision-Rights Matrix
Editorial Notes¶
Form Classification¶
Form family: Communication, Facilitation & Learning
Rationale: Before action, the actor paraphrases the goal, constraints, and plan and surfaces tensions so the tasker can detect and correct misunderstanding, making its operative form a two-way confirmation exchange.
Nearest alternative: Protocol, Workflow & Routine — The briefback follows a repeatable order, but mutual understanding produced through restatement and correction is the defining mechanism.
Review outcome: Adjudicated after independent review; high confidence.
Origin Attribution¶
Primary origin: Military & Strategic Studies
Origin pattern: Single lineage
Present-day reach: Multi-domain
Rationale: Restating commander's intent, constraints, and planned action before execution is characteristic military mission-command practice.
Related originating lineages:
- Aviation & Aeronautics — Readback/hearback closes the loop on operational clearances and catches mismatches before action.
- Medicine & Healthcare — Teach-back confirms understanding of intended actions and repeats correction until the account is accurate.
Review resolution: Military and strategic studies is the agreed primary lineage because backbriefs require subordinates to restate mission and commander's intent before release to act. Aviation and healthcare contribute convergent closed-loop verification practices; the generalized protocol remains an established lineage rather than a new synthesis.
Review outcome: Reconciled after independent review; high confidence.
Notes¶
[n1] In mission command, the backbrief has a subordinate restate the commander's intent and their own plan for accomplishing it, so misunderstanding is caught before execution. Its value depends on genuine paraphrase — a verbatim read-back confirms transmission, not comprehension. ↩a ↩b