Pattern: Sequential Pipeline | Team size: 6 agents
A step-by-step research pipeline that turns messy inputs into a repeatable operating system: channel prioritization, ICP insights, competitor gaps, then packaging into a measurable 90-day plan. This works well for Claude Code in a local workspace because each stage can write structured artifacts (tables, briefs, checklists) that become inputs to the next stage, minimizing rework and keeping evidence traceable.
Design a specialized AI agent research team to help APG MEDIA, a Vilnius-based marketing agency serving B2B clients, research how to raise brand awareness among B2B decision-makers. Desired outcome: a practical research operating system that identifies the highest-leverage awareness channels, competitor positioning gaps, ICP pains, proof assets, partnerships, content angles, measurement plan, and a 90-day action roadmap. The team should be optimized for Claude Code executing research from a local workspace, producing concise but evidence-backed deliverables suitable for agency leadership.
Create an agent team to design a specialized AI agent research team “Awareness OS Assembly Line” for APG MEDIA (Vilnius-based marketing agency serving B2B clients) to research how to raise brand awareness among B2B decision-makers and produce a practical, evidence-backed Research Operating System (OS) with channel priorities, competitor gaps, ICP pains, proof assets, partnerships, content angles, measurement plan, and a 90-day action roadmap.
PROJECT CONTEXT (READ CAREFULLY)
- Organization: APG MEDIA, marketing agency in Vilnius, Lithuania. Primary work: B2B marketing for clients (likely EU/Baltics/CEE reach, but don’t assume—validate).
- Objective: Raise brand awareness among B2B decision-makers (for APG MEDIA itself and/or as a repeatable playbook for APG MEDIA to apply to client engagements). Build a reusable Research OS suitable for agency leadership.
- Output style: concise, practical, evidence-backed, with traceable citations and clear assumptions. Use tables and checklists. Avoid fluffy frameworks without actionable steps.
- Workspace: Claude Code runs locally; use the local workspace files. If no prior data exists, create templates and a “data request list” APG can fill.
GLOBAL CONSTRAINTS
- Evidence: Every non-obvious claim must be supported by (a) a citation to a source (URL) or (b) clearly labeled as “Hypothesis” with rationale.
- Region sensitivity: Prefer EU/Lithuania/Baltics/CEE-relevant sources when possible; otherwise include global benchmarks but note applicability.
- Leadership-ready: Each artifact must have an “Executive Summary” (max 120 words) plus a “So What / Decisions Enabled” bullet list (3–7 bullets).
- Traceability: Maintain a master bibliography and a “claim-to-evidence” map for key recommendations.
- Sequential pipeline: Do not start a downstream stage until upstream deliverables are completed and referenced.
OUTPUT DIRECTORY (MUST USE EXACT PATHS)
Write all outputs to: outputs/agent_teams_demo/apg_media_awareness_os/
TEAM DESIGN: SEQUENTIAL PIPELINE (STRICT)
Agents must run in this exact order and respect dependencies:
1) Research Orchestrator (Stage 0 + final synthesis)
2) ICP & Buying Committee Analyst (Stage 1)
3) Channel Intelligence Analyst (Stage 2)
4) Competitive Positioning & Gap Mapper (Stage 3)
5) Proof & Content Strategist (Stage 4)
6) Measurement & Experiment Designer (Stage 5)
7) Research Orchestrator (Stage 6 synthesis/review + final OS + roadmap)
COLLABORATION MECHANICS (MANDATORY)
- After each stage, the agent must:
1) Write its artifact file(s).
2) Append 5–10 “handoff notes” to outputs/agent_teams_demo/apg_media_awareness_os/00_handoff_notes.md including:
- What was done, key findings, key uncertainties, and what the next agent must assume or verify.
3) Add at least 5 entries to outputs/agent_teams_demo/apg_media_awareness_os/99_bibliography.md (or fewer only if truly impossible), including title, publisher, date, and URL.
- Each downstream agent must begin by reading 00_handoff_notes.md and explicitly listing:
- “Inputs consumed” (file paths)
- “Assumptions inherited”
- “Assumptions challenged” (at least 2 challenges per stage; if you agree, say why with evidence)
- Disagreement protocol: If an agent disagrees with upstream conclusions, it must:
- Document the disagreement in 00_handoff_notes.md
- Propose a resolution test (what evidence would settle it)
- Continue with the best-available path, clearly labeled.
FILES TO PRODUCE (REQUIRED INDEX)
Create an index file first and keep it updated at each stage:
- outputs/agent_teams_demo/apg_media_awareness_os/README.md
- Include: purpose, how to use the OS, file map, and a 10-line “How to update this quarterly” procedure.
Also maintain these living files across all stages:
- outputs/agent_teams_demo/apg_media_awareness_os/00_handoff_notes.md
- outputs/agent_teams_demo/apg_media_awareness_os/99_bibliography.md
- outputs/agent_teams_demo/apg_media_awareness_os/98_claims_to_evidence_map.md
STAGE 0 — RESEARCH ORCHESTRATOR (SCOPE + QUALITY GATES)
Goal: Translate the business problem into a concrete research plan, define decision points, and set quality gates for each stage.
Deliverables (must complete before Stage 1 begins):
1) outputs/agent_teams_demo/apg_media_awareness_os/01_research_brief.md (900–1,200 words)
Required sections:
- Background & Objective (APG MEDIA context, 80–120 words)
- Primary Decisions This OS Must Enable (7–10 bullets)
- Research Questions (12–18 numbered questions)
- Definitions (brand awareness, “decision-maker”, “channel”, “proof asset”, “partnership”) (5–8 definitions)
- Scope & Non-Scope (two lists)
- Evidence Standards (what counts as evidence; what doesn’t)
- Risks & Failure Modes (6–10 items)
2) outputs/agent_teams_demo/apg_media_awareness_os/02_data_requests.md (400–700 words)
- A checklist of internal data APG should provide (past campaigns, CRM, client verticals, case studies, budget ranges, geographic focus).
- For each item: why needed, owner suggestion, and “minimum viable substitute.”
3) Initialize/Update:
- README.md (create file map)
- 00_handoff_notes.md (create)
- 99_bibliography.md (create)
- 98_claims_to_evidence_map.md (create with a starter template)
Quality gate checklist (must appear at end of 01_research_brief.md):
- Clear decision list
- Testable research questions
- Explicit assumptions
- Clear boundaries
- Output evaluation criteria
STAGE 1 — ICP & BUYING COMMITTEE ANALYST
Dependency: Do not start until 01_research_brief.md and 02_data_requests.md exist.
Goal: Map ICP segments and buying committee roles; extract pains, triggers, objections, and “moments of awareness” relevant to APG MEDIA’s B2B services.
Deliverables (must complete before Stage 2 begins):
1) outputs/agent_teams_demo/apg_media_awareness_os/10_icp_buying_committee.md (1,200–1,800 words)
Required sections:
- Executive Summary (≤120 words)
- ICP Segmentation (3–5 segments): for each include firmographics, maturity, typical motions, “likely awareness channels,” and why APG can win (table)
- Buying Committee Map: roles (e.g., CEO/Founder, CMO/Head of Marketing, RevOps, Sales leader, Procurement) + influence level + evaluation criteria (table)
- Jobs-to-be-Done & Triggers: 8–12 items (bullet list with “trigger → desired outcome”)
- Pains, Objections, and Misconceptions: 12–18 items; include “what they say” and “what they mean” (table)
- Trust & Proof Requirements: what evidence each role trusts (table)
- Insights Most Relevant to Awareness (5–9 bullets)
2) outputs/agent_teams_demo/apg_media_awareness_os/11_interview_guide.md (700–1,000 words)
- 12–15 interview questions for decision-makers + 6–8 for internal stakeholders
- Include “signal to listen for” under each question
- Include a 10-minute pre-call survey (5–7 questions)
3) Update 98_claims_to_evidence_map.md with at least 8 ICP-related claims and citations.
Research instructions:
- Use credible sources: B2B buying research (e.g., Gartner/Forrester summaries, LinkedIn B2B Institute, Edelman, WARC, industry benchmarks) and EU context where possible.
- Clearly label anything that’s a hypothesis pending APG internal data.
STAGE 2 — CHANNEL INTELLIGENCE ANALYST
Dependency: Do not start until Stage 1 deliverables exist and are read.
Goal: Identify and rank awareness channels for reaching B2B decision-makers; propose a channel mix with rationale, costs/effort, and expected awareness lift proxies.
Deliverables (must complete before Stage 3 begins):
1) outputs/agent_teams_demo/apg_media_awareness_os/20_channel_landscape_prioritization.md (1,400–2,000 words)
Required sections:
- Executive Summary (≤120 words)
- Channel Landscape: Paid / Owned / Earned / Partner / Events (brief definitions)
- Channel Prioritization Matrix (table): 10–16 channels/tactics; columns must include:
- Target roles reached (from Stage 1)
- Reach potential (H/M/L)
- Credibility effect (H/M/L)
- Time-to-signal (weeks)
- Cost level (H/M/L) + notes
- Execution complexity (H/M/L)
- Best-fit ICP segments
- Evidence/citation
- Top 5 “Highest-Leverage” Channels: for each include a 150–220 word mini-brief:
- Why it works for decision-makers
- Minimum viable execution
- Common pitfalls
- What to measure first
- Channel Mix Recommendation: 60/30/10 or similar allocation with rationale (must be tied to constraints/assumptions)
2) outputs/agent_teams_demo/apg_media_awareness_os/21_partner_event_shortlist.md (table, 25–60 rows)
- Identify potential partner types and example organizations (not just categories): associations, communities, podcasts, newsletters, event series, SaaS ecosystems, accelerators, VC/operator communities (include Baltics/EU where possible).
- Columns: partner name, type, audience, estimated fit (H/M/L), activation idea, contact path (if public), notes, source URL.
3) Update 98_claims_to_evidence_map.md with at least 10 channel-related claims and citations.
STAGE 3 — COMPETITIVE POSITIONING & GAP MAPPER
Dependency: Do not start until Stage 2 deliverables exist and are read.
Goal: Audit competitors (direct agencies and “category alternatives” like freelancers, consultancies, platforms) to find positioning whitespace and distribution gaps.
Deliverables (must complete before Stage 4 begins):
1) outputs/agent_teams_demo/apg_media_awareness_os/30_competitive_positioning_audit.md (1,400–2,200 words)
Required sections:
- Executive Summary (≤120 words)
- Competitor Set Definition: 8–12 competitors; include selection criteria and categories
- Positioning & Messaging Claims Table: for each competitor capture:
- Primary promise
- Proof pattern (case studies, logos, awards, metrics)
- Vertical focus (if any)
- Channel/distribution signals (blog cadence, LinkedIn, webinars, events, paid)
- Differentiators and “sameness”
- Source URLs
- Gap Analysis: 6–10 whitespace opportunities (each with “why credible,” “why underserved,” “how to test”)
- Implications for APG MEDIA: 5–8 decisions about message, proof, and channels
2) outputs/agent_teams_demo/apg_media_awareness_os/31_positioning_options.md (3 options, 250–350 words each)
Each option must include:
- One-line positioning statement
- Who it’s for (ICP segment)
- The “enemy” / alternative
- Proof required
- Best channels to express it
- Risk/downsides
3) Update 98_claims_to_evidence_map.md with at least 10 competitor/gap-related claims and citations.
STAGE 4 — PROOF & CONTENT STRATEGIST
Dependency: Do not start until Stage 3 deliverables exist and are read.
Goal: Translate ICP + channel + competitor gaps into a proof asset plan and content/narrative angles designed for B2B decision-maker awareness.
Deliverables (must complete before Stage 5 begins):
1) outputs/agent_teams_demo/apg_media_awareness_os/40_proof_asset_blueprints.md (1,200–1,800 words)
Required sections:
- Executive Summary (≤120 words)
- Proof Asset Inventory Template (table) that APG can fill (columns: asset type, ICP, funnel intent, claim supported, format, status, owner, due date)
- “Proof Gaps” (8–12): what APG must create to compete (benchmarks, teardown studies, case studies, POVs, playbooks)
- 6 Asset Blueprints (each 150–220 words): goal, audience, outline, required data, CTA, distribution channels
2) outputs/agent_teams_demo/apg_media_awareness_os/41_content_angle_bank.md (at least 30 angles)
Format: table with columns:
- Angle/title
- ICP segment + role
- Hook (1 sentence)
- Core insight
- Proof to include
- Best format (post, article, webinar, deck, calculator, email)
- Distribution channel(s)
- Measurement signal
3) outputs/agent_teams_demo/apg_media_awareness_os/42_editorial_calendar_90_days.md (table)
- 13 weeks; 3 posts/week minimum + 1 “hero” asset every 2–3 weeks
- Include owner role, status, dependencies (e.g., “needs case study data”)
4) Update 98_claims_to_evidence_map.md with at least 8 proof/content-related claims and citations (benchmarks about content effectiveness, trust, etc.).
STAGE 5 — MEASUREMENT & EXPERIMENT DESIGNER
Dependency: Do not start until Stage 4 deliverables exist and are read.
Goal: Define awareness measurement, instrumentation, baselines, and an experiment cadence connecting actions to leading indicators (not just vanity metrics).
Deliverables (must complete before Stage 6 begins):
1) outputs/agent_teams_demo/apg_media_awareness_os/50_measurement_plan.md (1,300–1,900 words)
Required sections:
- Executive Summary (≤120 words)
- KPI Stack: North Star + supporting KPIs (awareness, consideration proxies, pipeline assist) (table)
- Instrumentation Checklist: tracking requirements per channel (LinkedIn, website, webinars, podcasts, partnerships, events, email) (table)
- Baseline Plan (first 2 weeks): what to measure, how, and where it lives
- Attribution & Interpretation: how to avoid false certainty; what “good” looks like by week 4/8/12
- Reporting Cadence: weekly and monthly templates (bullets)
2) outputs/agent_teams_demo/apg_media_awareness_os/51_experiment_backlog.md (table, 12–20 experiments)
Columns:
- Hypothesis
- Channel
- ICP/role
- Creative/offer
- Primary metric
- Guardrail metric
- Sample size / duration estimate
- Effort (S/M/L)
- Expected impact (S/M/L)
- Dependencies
- Decision rule (“ship/iterate/kill”)
3) outputs/agent_teams_demo/apg_media_awareness_os/52_dashboard_wireframe.md (400–700 words)
- Describe the dashboard layout and required data sources; include a simple ASCII wireframe.
4) Update 98_claims_to_evidence_map.md with at least 8 measurement-related claims and citations.
STAGE 6 — RESEARCH ORCHESTRATOR (FINAL SYNTHESIS + OS + ROADMAP)
Dependency: Do not start until Stages 1–5 are complete and all artifacts exist.
Goal: Compile the full Research OS and a practical 90-day plan. Ensure internal consistency, resolve conflicts, and produce a leadership-ready package.
Deliverables:
1) outputs/agent_teams_demo/apg_media_awareness_os/60_awareness_research_os.md (2,000–2,800 words)
Required structure:
- Executive Summary (≤150 words)
- The OS in 7 Steps (each step 120–200 words, numbered, with inputs/outputs)
- Decision-Maker Insights (condensed from Stage 1: 8–12 bullets)
- Channel Priorities (Top 5 + “do later” list) (table)
- Positioning Recommendation (choose 1 of the 3 options; justify with evidence and risks)
- Proof & Content System (what to create first, why, and how it compounds)
- Partnerships & Events Motion (3 motions: “lightweight”, “standard”, “big bet”)
- Measurement System (KPI stack + cadence + decision rules)
- Open Questions + How to Answer Them (8–12 items)
2) outputs/agent_teams_demo/apg_media_awareness_os/61_90_day_action_roadmap.md (900–1,300 words + tables)
Must include:
- Week-by-week plan (13 weeks) in table form: initiatives, owner role, dependencies, expected output, success metric
- Resourcing model: minimum viable team (roles + hours/week) and “if we have more budget” upgrades
- Risk register (top 8) with mitigations
3) outputs/agent_teams_demo/apg_media_awareness_os/62_leadership_brief.md (450–650 words)
- A crisp memo for agency leadership: what we’re doing, why now, what success looks like in 90 days, and what decisions/inputs are required from leadership.
4) Finalize:
- 98_claims_to_evidence_map.md: ensure every key recommendation in 60/61 is mapped to evidence or labeled hypothesis.
- 99_bibliography.md: deduplicate, ensure consistent format.
- README.md: update “How to use” and file map.
FINAL SYNTHESIS/REVIEW STEP (MANDATORY AND EXPLICIT)
After writing Stage 6 deliverables:
- Perform a consistency audit across all files:
- No contradictory ICP/channel/positioning guidance
- KPIs align with experiments and roadmap
- Roadmap dependencies are feasible
- Write outputs/agent_teams_demo/apg_media_awareness_os/70_synthesis_review.md (700–1,000 words) with:
- “What we’re confident about” (5–8 bullets with citations)
- “What is still uncertain” (5–8 bullets + proposed tests)
- “Top 10 actions ranked by leverage” (numbered list with rationale)
- “Known limitations of this research” (4–8 bullets)
- End by listing exactly 10 next-step questions to ask APG MEDIA leadership to tailor the OS to their constraints.
EXECUTION NOTE
Use local workspace materials if present; otherwise, rely on web research with citations and clearly label hypotheses. Ensure every stage reads prior outputs, challenges at least two upstream assumptions, and leaves actionable handoff notes.
Create your own AI agent team at Build Agents Store. Describe your business problem and get specialized agent teams with ready-to-use prompts for Claude Code.