# Bmc Antibody And Sandbox Design > Reads a new business model for organizational-rejection triggers — margin conflict, sales-process conflict, cannibalization fear — inside the client's own organization, and designs a protected sandbox so internal 'antibodies' don't kill the experiment before it's tested. - Skill: `pilot2service/bmc-antibody-and-sandbox-design` (Agent Skill) - Install (CLI): `npx skillmds@latest add pilot2service/bmc-antibody-and-sandbox-design` - Raw SKILL.md: https://api.skillmd.com/api/skills/pilot2service/bmc-antibody-and-sandbox-design/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: AI & ML - Author: Pilot2Service (https://skillmd.com/u/pilot2service) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/pilot2service/bmc-antibody-and-sandbox-design --- # BMC Antibody and Sandbox Design ## Purpose Every other skill in this pack judges whether a business model is GOOD. This skill judges something different: whether the client's OWN organization will let it survive long enough to find out. A new model often threatens existing incentives, processes, or identities inside the company that's trying to build it — the biological metaphor practitioners use is "corporate antibodies": the organization reacts to something unfamiliar the way a body reacts to a mismatched transplant, attacking it regardless of its merit. Use this once a candidate model exists (after `bmc-innovation-pattern-matching` or `bmc-canvas-clarity-and-iteration`) and before it's handed to the parts of the organization that didn't design it. ## Anchored in research The "corporate antibodies" concept is independently documented under this exact name across multiple sources: Harvard Business Review, ["Get the Corporate Antibodies on Your Side"](https://hbr.org/2012/05/get-the-corporate-antibodies-o) (2012); and a substantial, independently-arrived-at body of corporate-innovation writing using the same biological metaphor (Forbes, Wellspring, Idea to Value, CIO.com). The taxonomy of antibody types in Step 2 below is this pack's own synthesis of the recurring patterns across that literature, not a direct reproduction of any single source's list — treat it as a working checklist, not a citation. ## Method 1. **Compare the new model against the existing one, block by block, looking for conflict — not fit.** For each canvas block, ask: does this new model ask anyone in the existing organization to act against their own current incentives, metrics, or identity? The four most common conflict points: - **Margin conflict** — the new model has structurally lower margins than the core business, so anyone measured on blended margin has a reason to starve it of resources or deprioritize it. - **Sales-process conflict** — the new model requires a different sales motion (self-serve vs. enterprise, subscription vs. one-time), and the existing sales team is compensated on the old motion — they will not sell what doesn't pay their commission structure, regardless of stated company priority. - **Cannibalization fear** — the new model could shrink revenue from an existing, currently-successful line, and whoever owns that line's numbers has a direct reason to slow-walk or undermine it. - **Identity/brand conflict** — the new model doesn't match "who we are" in the organization's own self-story (a premium brand entering a low-cost segment, a B2B company going direct-to-consumer) — this resistance is often the least visible because it's expressed as "that's not really us" rather than a stated business objection. 2. **Read for early antibody activity, not just structural risk.** Signals that antibodies are already active, before the model has even launched: sudden compliance or process objections that were never raised for similar past initiatives; a "yes, but" response from every functional lead consulted; resourcing requests that get acknowledged verbally and then quietly deprioritized; the project being reframed internally as a "pilot" or "experiment" by people who weren't asked whether it should be one. 3. **Design the sandbox — don't just document the risk.** Once conflict points are named, choose protective structures matched to the specific conflict: - **Margin conflict** → a separate cost center or P&L line, so the new model isn't blended into a metric that makes it look like a failure by the old business's standard. - **Sales-process conflict** → a dedicated sales channel or team not carrying the old quota/commission structure — not the existing sales team "also" selling it on the side. - **Cannibalization fear** → an explicit executive mandate that cannibalization of the old line, if it happens, is an accepted and expected outcome, stated before launch, not litigated after the first quarter it happens. - **Identity conflict** → a distinct name, brand, or team identity for the new initiative, so it isn't forced to inherit expectations ("but that's not how we do things") that only apply to the core business. - **General, any conflict type** → a named, senior executive sponsor whose own performance isn't measured by the core business's existing metrics, and a time-boxed exemption from the standard approval/procurement/compliance gates that assume core-business norms. 4. **Set an explicit integration trigger.** A sandbox that never ends is its own failure mode — decide in advance what evidence (revenue threshold, validated demand, operational maturity) triggers folding the initiative back into normal organizational structure, and who makes that call. 5. **Name the antibody risk out loud to the sponsor, not just internally to the design team.** The point of this skill isn't to quietly work around resistance — it's to make the risk visible enough that the people with authority to build a sandbox actually do, before the antibodies get there first. ## What this skill does NOT do - Doesn't guarantee the sandbox will hold — organizational politics can overrule any structural protection; this skill reduces predictable risk, it doesn't eliminate it. - Doesn't replace `bmc-canvas-diagnostic-reading` or `bmc-innovation-pattern-matching` for judging whether the model itself is sound — a model can be well-designed and still get killed by antibodies, or poorly designed and survive because the sandbox is strong; this skill only addresses the second axis. - Doesn't apply meaningfully to a solo founder building something new with no existing organization to conflict with — this is specifically for a new model being introduced INTO an existing company (including a growth-stage client adding a new line, not just large enterprises). ## Refinement notes - Which of the four conflict types have you actually seen kill an otherwise good model in your own client work — and which sandbox structure actually held? - What's the earliest antibody signal you personally watch for, before it's obvious? - Have you ever seen a sandbox stay open too long and become its own problem? What triggered closing it? ## Continue from here - Use after: `bmc-innovation-pattern-matching/SKILL.md` or `bmc-canvas-clarity-and-iteration/SKILL.md` — once a candidate model exists. - Use before: handing the model to functions outside the design team (sales, finance, operations) for execution. - This pack's shared guardrails: `../../CLAUDE.md` ## References - `../../references/bmc-source-material-notes.md` — source material background - `../../references/bmc-resilience-heuristics-research.md` — selection and grounding notes for this skill and its siblings - `../../CLAUDE.md` — this pack's shared guardrails