Beam Protocol Docs

Appearance

First Production Partner Workflow Contract ​

This page defines the exact workflow Beam 1.0.0 treats as the first production-grade partner handoff.

When the landing page, guided evaluation, onboarding pack, or operator queue says "one workflow," this is the workflow they mean.

Workflow Name ​

Quote Approval Partner Handoff

Plain-Language Summary ​

A buyer-side procurement agent asks a partner-side operations agent for stock, delivery timing, and a quote. The partner-side agent responds on the same Beam thread. A buyer-side operator and finance owner can inspect the request, the reply, and any delayed follow-up without losing the paper trail.

Why This Workflow ​

This workflow is the right 1.0.0 anchor because it is:

  • cross-company,
  • async by nature,
  • operationally sensitive,
  • easy to explain to a non-technical stakeholder,
  • good at exposing whether Beam keeps proof and ownership intact when work slows down.

Named Roles ​

Sender ​

  • buyer-side procurement agent
  • example: procurement@acme.beam.directory

Recipient ​

  • partner-side operations or supplier desk agent
  • example: partner-desk@northwind.beam.directory

Buyer-Side Owner ​

  • the business owner who cares whether the quote and delivery answer arrives on time

Operator Owner ​

  • the person who will inspect traces, signal state, dead letters, and recovery steps if the workflow stalls

Expected Latency ​

The production contract for this workflow is:

  1. Trace visibility
    • the request should become visible in Beam within seconds
  2. Operator visibility
    • an operator should be able to identify the current stage and owner within one minute
  3. Workflow response
    • the partner side should either reply or record an explicit async follow-up state inside the same thread within the agreed workflow window
  4. Escalation clarity
    • if the expected business reply window is missed, the operator signal and next action should already be visible

The exact commercial SLA can vary by partner. The product contract does not: Beam must make the state legible, attached, and recoverable.

Acceptable Failure Handling ​

The workflow is still acceptable if one of these happens, as long as Beam makes it explicit:

  • the recipient is temporarily unavailable and Beam retries,
  • policy or trust controls reject the request and the reason is visible,
  • the reply needs human or system follow-up later and the same thread carries the next action,
  • the request reaches dead-letter and the operator can point at the failure state without guessing.

The workflow is not acceptable if:

  • a request disappears without a visible state,
  • the operator cannot say who owns the next step,
  • the proof depends on memory or ad hoc chat recap,
  • a buyer has to trust that the message moved without seeing the evidence.

Operator Proof Points ​

For this workflow to count as production-ready, Beam must show:

  • a healthy operator baseline before the request starts,
  • one traceable request from arrival to reply or explicit async follow-up,
  • one owner and next action attached to the request,
  • visible retry, delay, or dead-letter state if the workflow degrades,
  • a proof package that can be shared without exposing operator-only detail.

Production-Ready Exit Criteria ​

Beam 1.0.0 should treat this workflow as production-ready only when all of these are true:

  • the workflow can be explained in one sentence by a normal buyer,
  • the sender and recipient are fixed and visible,
  • the onboarding pack and go-live checklist both point at this same workflow,
  • the operator dashboard can show the current state, owner, and recovery path,
  • blocked go-live prerequisites can be recorded explicitly in the request,
  • the proof pack can be exported from live evidence,
  • backup, restore, and fire-drill paths have been rehearsed on the current architecture,
  • the final dry runs pass for buyer, operator, and production-partner views.

What This Contract Does Not Try To Cover ​

This contract does not try to solve every Beam use case at once. It is intentionally narrow.

It does not define:

  • a generic internal automation story,
  • a marketplace of many workflows,
  • broad enterprise rollout packaging,
  • a self-serve "bring all your agents" motion.

If the first production partner workflow is not stable, Beam should not pretend the broader story is production-ready.