Engineering · EP050

Distribution Engine.

A 37-node demand-to-outcome pipeline, built as seven executable process boundaries that each run deterministically end to end, from discovering a real customer problem to a measured outcome and the feedback that improves the next cycle. The project identifier, EP050, is engineering metadata; what it builds is the engine behind the Content & Distribution and Lead Generation platforms.

The problem

Finding real demand, turning it into content that answers it, and getting that content in front of the people who had the question is normally three separate workstreams run by three different tools with no shared record between them. Nothing stops a system from generating content about demand that doesn't exist, or distributing an asset nobody asked for, without a plausibility gate and a closed feedback loop, volume gets mistaken for progress.

Objective

Close the loop end to end, deterministically, with no manual data entry in the path: discover a real, evidenced demand signal; generate an asset grounded in verified facts; distribute it to where the signal came from; capture and attribute the outcome; and feed that outcome back into what gets discovered and built next.

Design principles

Fail closed.
No fabricated data.

Every accepted node executes deterministically with no human manually keying data in the loop. Where a node lacks a real, verified data source, a postcode mapping, a cost rate, a business claim, it fails closed rather than guessing. Two real quality gates already caught this in practice: a live search for a Mars-related test query returned real, non-local results that a naive non-zero-results check would have accepted, and a coincidental brand-name collision nearly let an unrelated local business through as a service match. Both are now blocked by dedicated gates, not a generic filter.

Architecture

Seven phases,
37 nodes.

Status reflects genuine node acceptance as of 2026-08-25, not aspirational readiness, sourced directly from the project's own live workflow map.

01

Target ingestion

Nodes 01–04. Target and conversion definition.
Complete, plus a candidate-clustering and bulk-import extension shipped 2026-08-18.

02

Demand discovery

Nodes 05–10. Demand intelligence.
Automated, code complete for all 6 nodes; 2 of 6 run live today, 4 need user-supplied API credentials.

03

Strategy & path

Nodes 11–15. Opportunity and path strategy.
Complete, all 5 nodes automated, including live-signal chain re-derivation.

04

Knowledge & assets

Nodes 16–19. Canonical knowledge and asset factory.
Complete, plus a winner-replication loop-back extension shipped 2026-08-18.

05

Distribution & conversion

Nodes 20–27. Placement, destination and capture.
Mixed, nodes 20/21/26 accepted and console-wired; node 27 working, pending acceptance; nodes 22–25 deferred (MVP).

06

Lead lifecycle

Nodes 28–31. Lineage, qualification and routing.
Pending acceptance, real and tested, no formal sign-off yet.

07

Learning loop

Nodes 32–37. Outcome learning and allocation.
Pending acceptance, real and tested; a winner-replication and scale-out extension shipped 2026-08-18.

Inputs

  • A target definition — service and geography — and its conversion definition
  • Live search, video-platform and community-discussion data for demand signal collection
  • Verified facts and claims a business can actually cite
  • Curated geo and service adjacency data for candidate expansion

Processing

Each of the seven phases above is its own deterministic process boundary with a documented node contract, automated steps and a completion test, not a single monolithic script. A candidate only proceeds from one phase to the next once its own phase's quality gates are satisfied.

Outputs

  • Generated content and video assets, each traceable to a real demand signal and verified fact
  • Distributed, published assets across the configured channels
  • Captured and attributed leads with full lineage back to the originating signal
  • A cost and performance ledger once real paid actions are wired in

Intelligence

Every classification carries a commercial-intent score and a service-relevance check computed directly against live fetched results, not a static keyword list, so a candidate is excluded by evidence, not assumption.

Measurement

  • Node acceptance status per phase, tracked on the project's own live workflow map
  • Test-suite pass rate on every change (over 650 tests passing as of the last full run)
  • Real quality-gate exclusions, logged per stopped candidate with a reason

Feedback loop

When the learning loop (nodes 32–37) flags a winner, that result can trigger winner-replication (re-running the proven chain with a new format) and one-hop candidate clustering into adjacent geographies or services, each new candidate still has to earn its own real signal and facts before proceeding, not inherit the winner's.

Current status

Honest status,
not a highlight reel.

4 of 7 phases (ingestion, demand discovery, strategy, knowledge & assets) are complete or fully automated. Distribution & conversion is mixed: 3 nodes accepted, 1 working but pending acceptance, 4 deferred as MVP scope. Lead lifecycle and the learning loop are real and tested but await formal acceptance. Several real campaigns are currently blocked at the knowledge/fact-verification gate specifically because the only fact source ever registered was a fixture, purged once identified as unreal, that gap, not a missing feature, is the project's current highest-priority blocker.

What remains

  • A real, owned source of verified business facts and claims — currently blocking every live campaign at the knowledge gate
  • Formal board acceptance for nodes 27–37
  • External dispatch adapters for the four deferred distribution nodes (22–25)
  • Owned master data for postcode/locality mapping and real provider cost rates

Related

Platform &
evidence.

This engine is what the Content & Distribution and Lead Generation platforms run on. The detailed, node-by-node status map is the primary evidence for everything on this page.

Distribution Engine workflow map

The full 37-node, 7-phase architecture with per-node purpose, automated steps, completion tests, evidence and dependencies, the source for every status figure on this page.

View workflow map

Content & Distribution

The reusable platform this engine's demand-to-content and distribution stages are packaged as.

View platform

Next step

Have a process
worth automating?

A short conversation, an honest read on whether we're the right studio, and a clear picture of what it would take. No pitch deck.