Where do I start?
Describe where you start and find the row that fits best: each route shows which steps to follow, in what order and where you decide. Not every project uses every step. Pick your kit’s tab; if the new process needs software, the process route ends where the software route begins.
Describe your situation and find the closest row. Each route gives the mode, the artifacts in order (with their prompt) and the phase gates, which are the human decisions. A route is a proposal: conditional artifacts (⚠) depend on the roadmap’s applicability matrix: each step states, in its fit (encaje), in which modes it is mandatory, conditional (and on what condition) or not applicable.
1. Pick the mode
Section titled “1. Pick the mode”| Your situation | Mode |
|---|---|
| “We have a process that works badly (slow, costly, error-prone or complained about) and we want to redesign it” | A — Redesign |
| “We need to create a process that doesn’t exist yet” or “the current process can’t be observed” | B — Greenfield |
| “There’s a specific problem in one part of the process (a KPI, a step, a case type)” | C — Focused improvement |
| “We don’t know what processes we have or which one to tackle first” | D — Process architecture |
| “The process is already defined and we want to automate it (BPMS, RPA, AI, integration)” | E — Automation |
| “The same process runs differently in each site, country or group company” | F — Standardization |
| “We need to check whether we comply with a regulation or control framework, and fix it” | G — Compliance |
Torn between two modes? Pick C if the problem and its scope are contained; A if it spans several teams or the cause is unclear. If a root cause in C falls outside the scope, propose moving to A with a PDR.
2. Follow your mode’s route
Section titled “2. Follow your mode’s route”Every route starts the same way: fill in PROCESO.md (template) and Phase 1.
Common Phase 1: charter → SIPOC → stakeholders → glossary. 🚪 Gate 1: the sponsor and process owner approve the charter.
A — Redesign
Section titled “A — Redesign”- Common Phase 1.
- Discovery plan → interview guides ⚠ → document analysis ⚠ → event log ⚠ → synthesis → As-Is model → baseline. 🚪 Gate 2: participants recognize the As-Is as their real process.
- Value and waste → handoffs → risks and controls → requirements ⚠ → root cause. 🚪 Gate 3: the process owner validates problems and root causes.
- To-Be options → PDR → DMN rules ⚠ → automation → To-Be model → gap → business case. 🚪 Gate 4: sponsor (option and business case), process owner (operability) and Risk (controls).
- Model review → scenarios → simulation ⚠ → traceability. 🚪 Gate 5.
- Implementation plan → change management → SOPs → automation spec ⚠ → SDLC briefing ⚠ (only if the team decides to build software). 🚪 Gate 6: go-live.
- KPI dashboard → control plan → runbook → post-implementation review.
B — Greenfield
Section titled “B — Greenfield”Like A, with these changes: in Phase 2 the discovery plan becomes a context-gathering plan and interviews are about requirements; there is no event log, As-Is, baseline or gap. Phase 3 focuses on requirements and risks. Target KPIs are set from benchmarks.
C — Focused improvement
Section titled “C — Focused improvement”Like A, but with the charter limited to one sub-process and one KPI. The process map, context requirements and, for small changes, the To-Be options are usually skipped (the solution is recorded in a PDR). Handoffs ⚠.
D — Process architecture
Section titled “D — Process architecture”Phase 1 with the process map as the main artifact → high-level discovery ⚠ → prioritization PDR. Each prioritized process then starts its own route (A, B, C, E, F or G).
E — Automation
Section titled “E — Automation”Common Phase 1 → if the To-Be didn’t come from this methodology, at least value and waste (simplify before automating) → risks → automation → technology PDR → gap → business case → scenarios (they become the acceptance tests) → automation spec → Phase 7.
F — Standardization
Section titled “F — Standardization”Like A, with discovery and the As-Is model done per variant (site, country, team) and a variant comparison in the synthesis. To-Be options include at least: adopting the best existing variant, a hybrid and a redesign. Legitimate differences are recorded as controlled variants in a PDR.
G — Compliance
Section titled “G — Compliance”Common Phase 1 with {NORMATIVA} as the reference → discovery of the real process (document analysis is mandatory, event log if available) → risks, controls and regulatory requirements → gap between what happens and what’s required → remediation options and PDR → validation → control plan.
3. Between phases: how to use context
Section titled “3. Between phases: how to use context”- One conversation per phase. When a phase starts, bring
PROCESO.mdand the approved artifacts from earlier phases, not the whole conversation. - Within a phase, stay in the same conversation while the artifacts feed each other (for example, synthesis → As-Is model → baseline).
- Don’t move on with a gate still open. If it’s approved with conditions, record them as assumptions or risks.
4. If software comes out of it
Section titled “4. If software comes out of it”If in Phase 4 or 6 the team decides software is needed, the SDLC briefing is the input to the Spec-First methodology (Is2Be Software), Phase 1, Mode B.
Describe your situation and find the closest row. Each route gives the mode, the artifacts in order (with their prompt) and the human decisions. A route is a proposal: conditional artifacts (⚠) depend on the roadmap’s applicability matrix: each step states, in its fit (encaje), in which modes it is mandatory, conditional (and on what condition) or not applicable.
1. Pick the mode
Section titled “1. Pick the mode”| Your situation | Mode |
|---|---|
| “We’re building a product or system from scratch” | New project |
| “We have an old system that needs rebuilding or migrating” | Legacy modernization |
| “We need to add a module or feature to an existing system” | New module in an existing system |
| “We need a standalone service that exposes or consumes APIs or events” | Microservice |
| “We need a frontend piece that plugs into an existing shell” | Micro-frontend |
| “A brief has come in from the process methodology (Evidence-First)” | New project or new module, with the PRD in Mode B from the brief |
2. Follow your mode’s route
Section titled “2. Follow your mode’s route”Every route starts with PROYECTO.md (project variables) and keeps this order: Phase 1 complete for the whole scope → Phase 2 in order → no code without an approved plan and test plan → tests with every module → CI from day one.
New project (and microservice, micro-frontend)
Section titled “New project (and microservice, micro-frontend)”- PRD per module (Mode B, from a brief) → user stories → acceptance scenarios → flow diagrams ⚠ (only flows crossing ≥2 actors or systems). 🚪 The business validates the PRDs, the stories and the diagrams.
- Architecture options → ADR-001 (architecture) → technical sequence diagrams ⚠ (elevates Phase 1’s) + technical stories (Architecture origin) → testing and observability ADRs → build plan (variant A) → test plan → technical stories (Testing origin, for whatever the plan flags as pending). 🚪 A human decides each ADR and approves the plan.
- CI pipeline before the first business module.
- Authentication/authorization module ⚠ (once, with the first module that needs it — the prompt makes explicit whether it applies). For each module: backend (any interface type: HTTP, GraphQL, events, CLI…) → frontend ⚠ → audit trail ⚠ → tests (unit, backend, frontend) → logging → code review.
- E2E tests when each complete flow is done (or without a UI if the flow has no user interface) → CD pipeline → technical stories (DevOps origin) → runbooks before production.
Microservice: the PRD defines the contract (what it exposes and consumes); integration tests check the contract, and generar-e2e-sin-ui.md covers its critical flows end to end against real infrastructure, without waiting for the CD pipeline to find out. Micro-frontend: the architecture decides how it composes with the shell; E2E tests cover the shell integration.
Legacy modernization
Section titled “Legacy modernization”- PRD per module in Mode A (from existing code) → user stories → acceptance scenarios → flow diagrams ⚠. 🚪 The business confirms which rules are preserved.
- As-is architecture → to-be options → ADR-001 → technical sequence diagrams ⚠ + technical stories (Architecture origin) → migration plan (variant B: incremental or big-bang, decided in an ADR) → test plan → technical stories (Testing origin).
- Build as in a new project, plus characterization cases for the legacy system to check parity before cutover — technical stories (DevOps origin) after the CD pipeline.
New module in an existing system
Section titled “New module in an existing system”- Module PRD → user stories → acceptance scenarios → flow diagrams ⚠.
- As-is architecture ⚠ (if the host system isn’t documented) → ADRs only for the module’s new decisions, respecting the system’s existing ADRs → technical sequence diagrams ⚠ + technical stories (Architecture origin) ⚠ → build plan → test plan.
- Authentication/authorization module ⚠ only if the system doesn’t have one yet; backend module (
{CONTEXTO}= existing system) when adding an entity that follows an existing pattern; retroactive audit ⚠. Pipeline only if the existing one needs changes — technical stories (DevOps origin) if it applies.
3. Between phases: how to use context
Section titled “3. Between phases: how to use context”- One conversation per phase or per module. Bring
PROYECTO.md, the approved PRDs and ADRs, not the whole conversation. - To generate a lot of code, run Phase 3 in a coding assistant inside the IDE, with access to the repository.
- Decisions aren’t made in chat: they go into an ADR.