CZ-01 — System maturity · Brand pivot
Build velocity
From positioning decision to live B2B brand. One day.
Source material: AXIOM operating docs, agent infrastructure guide, and local website repo. Verification target: readable static pages, clean internal links, customer-readable setup path, and explicit QA gates before public launch.
11Static pages in repo
1Customer-readable guide
48hBuild sprint protocol
Approach. Atlas mapped the pivot from a prior creative-studio positioning to the autonomous-growth-firm thesis. Lyric drafted page copy on the new B2B brand. Stricture fact-checked agent descriptions against the agent-os scaffolding and rejected unsubstantiated claims. Forge ported the visual system to static HTML. Sentinel reviewed every section for tone and claim integrity before commit.
Result. Repo pages include homepage, /work, /about, /services, /methodology, /local, /blog, /privacy, /terms, /qa-protocol, and /first-30-days. Internal links are checked before launch. No merge or production deploy is implied until explicitly approved.
CZ-02 — Specification depth · Agent stack
Architecture
Twelve agents. Five components each.
Every agent specified as: (1) system prompt · (2) curated tool kit · (3) scoped memory · (4) evaluation rubric · (5) deterministic handoff protocol. JSON-typed inputs and outputs. Schemas versioned.
12System prompts
4Handoff schemas
47-pointEditorial rubric
Approach. Atlas wrote the strategist prompt and the cross-agent handoff contracts. Stricture wrote the editorial rubric and the source-claim mapping protocol. Sentinel formalized the seven-gate quality model: source-backed, scope-safe, brand-safe, buyer-useful, legally-cautious, privacy-safe, handoff-complete. Quill specced the operations boundary — what each agent can draft vs what requires operator approval.
Result. The public architecture now documents 12 agent roles, required components, handoff contracts, QA gates, and customer setup expectations. Concrete prompts/profiles stay internal unless intentionally published as architectural examples.
CZ-03 — Open methodology · Published
Transparency
We don’t hide the architecture. We sell the architecture.
Architectural-level prompts, rubrics, and workflow maps publishable quarterly. Methodology page covers the 5-component agent object, the canonical workflows, the quality gates, and the operating boundaries. Field Notes archive defined.
7Canonical workflows
5Operator runbooks
6Security phases
Approach. Atlas wrote the quarterly Field Note cadence. Lyric drafted /methodology covering the 5-component agent object, four canonical workflows (programmatic SEO, lifecycle email, signal outbound, weekly client report), boundaries, and open commitments. Concord defined which agents own which parts of each workflow.
Result. /methodology live. /blog live with six content pillars defined. First quarterly Field Note publishes at the end of Q1 with the architecture-level prompts and rubric updates from this quarter. Methodology will be expanded with full security-roadmap and permissions-matrix details in the next iteration.
CZ-04 — Economics · Cost stack public
Unit economics
Software-business margin. On services-business revenue.
The economic thesis is simple: lower marginal delivery cost than a traditional agency, while keeping human accountability at the approval and trust layers. Targets remain targets until client data exists.
lowerMarginal cost target
ownedClient assets/data
gatedHuman approvals
Approach. Atlas and Ledger modeled the unit economics from the build-guide’s reverse-engineering of the $1M/quarter target. Quill mapped fixed costs: LLM API spend, vector DB, orchestration, hosting, CRM, operator, E&O insurance, part-time contractor. Sentinel verified the model holds at the bootstrap floor (1–3 clients) as well as at scale.
Result. Cost stack documented openly. Three-tier pricing public on the homepage with monthly equivalents. First external engagement will produce real per-client cost data; published anonymized at quarter close.
