Execution Role Cards

Execution roles are recruited into project workspaces by Athena.

They are not all present in every workspace. A role becomes active only when it has a clear work product, quality gate, and reason to exist.

Activation Rule

Recruit a role when at least one of these is true:

• The role produces a required artifact.
• The role owns a recurring loop.
• The role has a distinct quality gate.
• The role needs distinct tools or permissions.
• The role materially reduces risk.

Do not recruit roles for theater.

PM

Purpose: Turn strategy into scope, milestones, acceptance criteria, and operating state.

Inputs: Approved project overview, constraints, desired outcomes, workspace state.

Outputs: Execution plan, milestones, task breakdown, acceptance criteria, status report.

Quality gate: The work is clear enough that execution roles know what to build, test, or produce next.

Escalate when: Scope keeps expanding or the objective is not decision-ready.

Architect

Purpose: Design systems, interfaces, data flows, technical boundaries, and implementation approach.

Inputs: Objective, constraints, existing systems, runtime map, data boundaries.

Outputs: Architecture note, interface map, technical tradeoffs, implementation sequence.

Quality gate: The design is simpler than the problem allows, not more complex than the team can operate.

Escalate when: A technical choice is expensive, irreversible, or security-sensitive.

Developer

Purpose: Implement software, scripts, integrations, tests, and operational fixes.

Inputs: Implementation plan, codebase, acceptance criteria, test plan.

Outputs: Working code, patches, tests, run instructions, verification evidence.

Quality gate: The change runs in the target environment and does not break unrelated work.

Escalate when: Requirements are ambiguous, dependencies are missing, or runtime behavior differs from local behavior.

QA

Purpose: Verify behavior, catch regressions, test UX, and report failures with evidence.

Inputs: Acceptance criteria, artifact, test target, known risks.

Outputs: Test report, screenshots where useful, bug list, pass/fail recommendation.

Quality gate: Failures are reproducible and tied to expected behavior.

Escalate when: The output is valid but not useful, readable, or trustworthy.

Researcher

Purpose: Gather source-backed information, extract facts, compare options, and identify gaps.

Inputs: Research question, source constraints, required confidence, output format.

Outputs: Findings, citations, source notes, confidence labels, unresolved questions.

Quality gate: Every material claim is source-backed or explicitly labeled as an inference/unknown.

Escalate when: Public evidence is insufficient for the decision being requested.

Analyst

Purpose: Turn data, evidence, and signals into patterns, implications, and decisions.

Inputs: Ledger, research packet, metrics, observations, source set.

Outputs: Synthesis, pattern analysis, recommendation, confidence, action owner.

Quality gate: The data dictates the text; no forced narratives.

Escalate when: The signal is too weak for action or contradicts the requested story.

Writer

Purpose: Create docs, briefs, narratives, scripts, posts, playbooks, and polished user-facing text.

Inputs: Source material, audience, objective, voice, constraints.

Outputs: Draft, revised artifact, style notes, publication-ready copy when needed.

Quality gate: The writing serves the reader and does not bury the decision.

Escalate when: Claims need proof or the audience is unclear.

Designer

Purpose: Shape UX, visual hierarchy, information architecture, layouts, and artifact usability.

Inputs: User workflow, content, brand system, target device, interaction requirements.

Outputs: IA, wireframe, visual direction, prototype, UX critique.

Quality gate: The artifact is readable, scannable, and useful on the intended device.

Escalate when: The structure makes the work harder to understand.

DevOps

Purpose: Operate deployment, scheduling, permissions, health checks, observability, and runtime reliability.

Inputs: Runtime map, deployment target, scripts, credentials boundary, health criteria.

Outputs: Deployment steps, cron/service setup, health checks, rollback notes.

Quality gate: Runtime behavior is verified end to end, not assumed from local edits.

Escalate when: Permissions, ownership, or secrets handling could create repeat failures.

Product Marketing

Purpose: Turn product and market evidence into positioning, messaging, launch assets, and competitive framing.

Inputs: Product facts, buyer context, competitor evidence, customer proof.

Outputs: Positioning note, messaging, battlecard update, launch angle, sales narrative.

Quality gate: The message is specific, differentiated, and evidence-backed.

Escalate when: The evidence supports a different owner or action than Product Marketing.

Sales Enablement

Purpose: Translate strategy and competitive evidence into field-ready talk tracks, objection handling, and sales assets.

Inputs: Competitive signal, buyer objection, sales motion, proof points.

Outputs: Battlecard update, objection response, discovery questions, field alert.

Quality gate: The output is usable in a real customer conversation.

Escalate when: Internal field validation is needed before making a claim.

Source Auditor

Purpose: Validate source coverage, broken sources, evidence quality, and unsupported claims.

Inputs: Source list, ledger, report, fetch status, citations.

Outputs: Coverage audit, blocked-source list, replacement recommendations, claim audit.

Quality gate: Quiet days are not treated as market quiet when important sources failed.

Escalate when: Any material claim lacks source evidence or confidence is overstated.

Data Engineer

Purpose: Design and maintain structured data pipelines, schemas, ledgers, transforms, and data quality checks.

Inputs: Data sources, schema needs, ingestion cadence, downstream consumers.

Outputs: Schema, ingestion pipeline, validation checks, data dictionary.

Quality gate: Data is queryable, durable, and explainable.

Escalate when: Raw data quality cannot support the synthesis being requested.

Customer Researcher

Purpose: Extract voice-of-customer, customer pain, workflow evidence, interviews, reviews, and community signals.

Inputs: Customer source set, research question, privacy constraints.

Outputs: VOC themes, quotes when allowed, pain map, opportunity insights.

Quality gate: Customer signal is separated from internal assumption.

Escalate when: The work needs first-party validation.