Bluefly commercial doctrine (binding on every definition)¶
Current authority: the "2026 GTM" tab of "Bluefly Organizational GTM" (Google Doc
1ED288ChytgUrrTK301X1japW--b4cCj-rYjq5vHuydA), period Q4 2026 into 2027, audience Bluefly
leadership. Read that tab before writing any definition. It is not in BluCity-Docs or GitLab,
so a repo-only search will miss it. Where it conflicts with BluCity-Docs, it is newer and more
specific about commercial intent; record the conflict rather than silently picking a side.
The hierarchy (Section 32, stated explicitly)¶
Digital Estate Operations = recurring product
Migration Factory = land motion
ContextControl = customer control room
Factory = execution and margin engine
Drupal = first market wedge
The stated purpose of that hierarchy is to stop treating three offers plus ContextControl as four competing strategies. Any new definition either fits one of those five slots or argues against the hierarchy directly. It does not quietly add a sixth.
The company being built¶
An autonomous digital operations company that gets paid to continuously operate important parts of a customer's technology estate. Not a larger Drupal agency. Drupal is the wedge, not the ceiling, chosen because the estates are large, fragmented and expensive, the spend already recurs, and the domain is bounded enough to prove the operating model.
Commercial model, stated as "service as software": SaaS says here is software, your team operates it; consulting says here are people, pay for their time; Bluefly says give us the responsibility, we deliver and prove the result.
The product: Bluefly Digital Estate Operations¶
Bluefly continuously monitors, maintains, remediates, verifies and improves complex Drupal estates under defined customer policies and approval boundaries. It does not stop at "we found a problem." The operating loop:
OBSERVE -> IDENTIFY WORK -> CLASSIFY RISK -> FIX AUTHORIZED CLASSES ->
PREPARE HIGHER-RISK CHANGES -> REQUEST APPROVAL WHEN REQUIRED -> TEST ->
DEPLOY -> VERIFY -> PRODUCE EVIDENCE -> REPEAT
Five initial capabilities, and nothing beyond them is primary: security and update operations, accessibility operations, configuration and code drift, release readiness, content operations. Content operations is aimed at maintaining trustworthy digital estates at scale, explicitly not at AI-generated content volume.
Three tiers to test as commercial hypotheses: Observe (Bluefly detects, customer remediates, lowest value, not the end state), Assist (Bluefly prepares and tests changes, customer approves or deploys), Operate (Bluefly remediates approved classes, deploys within defined authority, verifies, produces evidence; humans move to exceptions and policy). Operate is the stated strategic destination.
The Q4 2026 catalyst (verified 2026-09-21, re-verify before reuse)¶
| Fact | Status | Source |
|---|---|---|
| Drupal 10 reaches end of life December 9, 2026 | Confirmed | drupal.org core release schedule |
| Drupal 12.0.0 and 11.5.0 release the week of December 7, 2026 | Confirmed | same |
| A major version cannot be skipped; Drupal 10 must move through Drupal 11 | Confirmed, stated in those words | drupal.org upgrade process overview |
| Drupal 10 sites must reach 11.3.0 or higher before upgrading to Drupal 12 | Confirmed | same |
| Drupal 10 to 11 requires Drupal 10.3.0 or later | Confirmed | drupal.org how to upgrade from 10 to 11 |
| The Drupal 11 to 12 minimum version | Listed as TBA as of 2026-09-21; the only concrete statement is that core updates prior to 11.3.0 are removed in Drupal 12 | drupal.org how to upgrade from 11 to 12 |
| "11.4 or later required, 11.5 recommended" | NOT CONFIRMED. Appears to originate in a change record for the Short Scale Formatter contributed module, not core | drupal.org node 3607608, 1 July 2026 |
| Drupal 11.5 as an LTS line into end of 2028 | Not found on the release schedule page | - |
| ADA Title II: April 26 2027 for public entities of 50,000 or more; April 26 2028 for under 50,000 and special districts | Confirmed, after a one-year extension | Federal Register 2026-07663 |
The near-miss source above is the instructive part: a change record with an authoritative-looking version number that belongs to a different project. Confirm the project, not just the number.
Sequencing: Drupal 10 EOL is the Q4 2026 catalyst and organizes this quarter. Title II is a real 2027 market and does not. Accessibility is therefore the second remediation class, not a simultaneous launch, and the claim is never that automation produces WCAG conformance.
The wedge this supports is a paid Drupal 10 to Drupal 11 readiness and execution pilot whose purpose is to demonstrate the operating model and leave the customer on a supported Drupal 11 baseline with a defined path into recurring Security and Update Operations. The message is not "move to Drupal 12 now"; it is get safely off Drupal 10, establish a clean Drupal 11 baseline, and stop treating updates as emergency projects.
Pricing¶
Price around the operational responsibility Bluefly assumes. Variables: number of sites, environments, estate complexity, monthly execution volume, remediation authority, protected workflows, accessibility scope, evidence requirements, response requirements, support commitments. Do not price primarily by agent count. Do not price primarily by hours; hours are an internal COGS measure. Final pricing is explicitly not known yet and is a Q4 test.
Economics¶
Agent economics are product economics. Measure per recurring operation: execution cost, model cost, infrastructure cost, human review cost, automated actions, human escalations, success rate, time to resolution, estimated manual work replaced, cost per accepted outcome. Use the least expensive mechanism that can safely complete the work: deterministic check to code, simple classification to a small or local model, normal reasoning to an appropriate model, high-risk judgment to a strong model plus a human. The objective is not to maximize AI usage; it is to maximize profitable verified outcomes.
The compounding test: the first time Bluefly solves a difficult class of problem may take significant senior involvement, the tenth time should not. If every new customer requires the same senior labor as the last, the Factory is not working.
Buyer¶
Stop selling only to Drupal engineers. Champions: Drupal lead, web platform architect, Director of Web, technical delivery leader. Economic buyers: CIO, CTO, VP Digital, VP Engineering, agency executive, public-sector digital leader. The economic buyer needs current cost, current risk, what Bluefly takes over, what stays human, what it costs, what result they receive, how it is proven.
Target customers in priority order: large Drupal estates, higher education, government and public sector, agencies and systems integrators (both a segment and a channel; an agency running 50 sites may be a better first customer than 50 site owners).
The pitch¶
Do not begin with Gas City. Do not begin with ContextControl. Do not begin with agents. Do not begin with AI governance. Begin with: Bluefly continuously operates complex Drupal estates, finds the work, handles what you authorized, brings people into the decisions that need judgment, and verifies the result; instead of paying a team to repeatedly discover and fix the same classes of problem, you pay Bluefly to keep the estate healthy. Then show one example.
GTM motion and channel¶
TRUSTED RELATIONSHIP -> ESTATE OR MIGRATION PROBLEM -> OPERATIONS ASSESSMENT ->
BOUNDED PILOT -> PROVE OUTCOME -> RECURRING OPERATIONS CONTRACT -> EXPAND AUTHORITY AND SCOPE
The goal is not hundreds of leads; it is a small number of customers willing to let Bluefly prove the model. Channels: existing and former customers, Acquia relationships, Drupal agencies, systems integrators, higher education, public sector, Drupal community leadership. Every champion gets something concrete and forwardable: not a capabilities deck, not an architecture diagram, a product proposition.
Migration Factory is the land motion: migration discovery, migration execution, estate goes live, Digital Estate Operations. A migration must leave behind structured estate knowledge, validated patterns, automated testing, operational policies, repeatable release processes, evidence, and ongoing operational capability.
Product acceptance (use this as the definition of done)¶
A capability is not complete because an agent exists. It is complete when: customer problem proven, detector working, remediation working, authority boundary working, human gate working, verification working, evidence working, ContextControl visible, execution cost measured, and reusable on a second estate.
What is real versus not yet proven (as stated)¶
Real: senior Drupal expertise, existing customers and partner relationships, a working internal Gas City and BluCity Factory, durable work orchestration, reusable agents, orders and formulas, GitLab delivery, Drupal automation, active ContextControl development, active AMCS experimentation, migration opportunities, Bluefly.io being rebuilt around the new positioning.
Not yet proven: that customers will buy Digital Estate Operations as a recurring product; the right pricing model; the right initial bundle of operations; the gross-margin profile; customer willingness to delegate remediation authority; repeatable installation across multiple estates; ContextControl as a mature customer product; measurable reduction in senior labor across repeated operations.
Stop doing (each is a failure mode this skill enforces)¶
Defining the company around billable hours. Treating every project as unique. Building technical capability without a product or delivery consumer. Inventing another product every time we build a component. Leading customer conversations with agents, Gas City, Beads or Formulas. Calling internal Factory infrastructure a customer product. Treating ContextControl as a generic AI platform. Building dashboards instead of completing operational loops. Publishing theoretical ROI without measured evidence. Accepting technical activity as proof of commercial progress.
Leadership decisions already made¶
- Digital Estate Operations is the primary recurring product being proven. Not generic governed AI. Not generic consulting.
- Drupal is the initial operating domain, the wedge and not the permanent ceiling.
- Migration Factory is the primary project-based land motion; its preferred expansion path is recurring operations.
- ContextControl is the customer control room. It is not a separate unrelated product strategy.
- Factory investment must trace to customer value or reusable operational capability. Every major investment answers: what customer problem does this help us operate, who will use it, how does it produce revenue or lower COGS. No answer means it is not a current priority.
Supersession chain (three generations in one document)¶
| Generation | Where | Shape |
|---|---|---|
| 2026 pillars | "2026" tab | Seven pillars: AI advisory, open-source leadership, Drupal.org partner, thought leadership, partnerships, product exploration, federal |
| Q3 2026 | "Tab 13" | Four parallel offers: Migration Factory, Drupal Estate Assurance ("most promising candidate for recurring revenue"), AI-Ready Modernization, Agency Delivery Factory |
| Q4 2026 into 2027 | "2026 GTM" | One recurring product (Digital Estate Operations, the promotion of Drupal Estate Assurance), one land motion, one control room, one factory, one wedge |
Also superseded: the 2026-08-24 founder-locked line in BluCity-Docs
Engineering-Standard/glossary/platform-glossary.md selling "governed modernization outcomes"
as the motion. Modernization is now an enabling offer, not the company strategy. The glossary
still carries the older wording; treat the GTM tab as current and flag the drift.
Retained from the older doctrine because the GTM tab does not contradict it: the Service-to-Product ladder, the Commercial Work Gate (buyer, problem, offer, outcome, proof, reuse, revenue horizon, next commercial action), the no-new-repos constraint, the solo-estate ownership reality, and the rule that pricing and compliance claims are draft until operator approval.
Operating loop the product must fit¶
Drupal sees it, Beads records it, Gas City works it, GitLab delivers it, OtterMon proves it, and the customer sees it in ContextControl. A product whose delivery cannot be described that way needs an explanation in Open Questions.