# ElizaOS Strategic Intelligence (Intel) — 2026-01-28

## 0) Executive Snapshot (What changed + why it matters)
- **Strategic pivot crystallized:** Core-dev consensus to shift from hardware emphasis to **Eliza as an AI workflow automation assistant** powered by **n8n integration** and conversational workflow generation (Eliza **v2.0.0 published to npm**: `@elizaos/core@next`).
- **User demand signal validated:** Team estimate: **~95%** of target users want *workflow building* (connect Gmail/Calendar/Notion/Linear with conditionals) over “auto-coding”.
- **Cost + configuration pain surfaced (high impact):**
  - Knowledge plugin misconfig: `EMBEDDING_PROVIDER=openrouter` rejected (enum allows only `openai | google`).
  - Knowledge plugin cost blow-up: **$0.03–$0.04 per query** and **tens of thousands of tokens** on trivial docs when **contextual embeddings enabled** (`CTX_KNOWLEDGE_ENABLED=true`).
- **Community risk area persists:** Token-price anxiety + airdrop ambiguity and chain liquidity friction; needs clearer comms to prevent churn.

---

## 1) Data Pattern Recognition

### 1.1 Development velocity & trend (quantitative)
**GitHub (Jan-to-date snapshot in provided data):**
- **38 PRs opened / 21 merged**
- **93 issues opened / 52 closed**
- **31 active contributors**
- **Code churn:** ~**25,258 additions / 8,658 deletions** across **262 files** (high churn month; indicates rapid iteration + architectural movement)

**Trend interpretation**
- High issue creation vs closure suggests **discovery + roadmap expansion outpacing stabilization**.
- Large-churn V2 workstreams risk **integration debt** unless MVP is tightly scoped (see Prioritization).

### 1.2 Community engagement patterns (qualitative → operational)
- Engagement clusters around:
  1) **Workflow automation pivot** (core-devs highly active)
  2) **Knowledge plugin cost/config** (coders channel support loops)
  3) **Token/airdrop questions** (discussion channel; repeated uncertainty)
- Support dynamics: a small set of “experts” (e.g., plugin-knowledge, core-dev) repeatedly unblock users → indicates **documentation gaps** and opportunity for **self-serve fixes**.

### 1.3 Feature adoption signals (early indicators)
- **Eliza v2.0.0 “next” release** is available for testing (`npm install @elizaos/core@next`), but adoption appears **developer-only** right now; no funnel metrics yet.
- **n8n cloud account created + invites issued** (operational readiness for a shared workflow instance).
- **Self-hosting interest**: mini-PC + Oracle free tier patterns indicate demand for **“always-on, low-cost”** deployments.

### 1.4 Pain point correlation across channels (highest → lowest impact)
1) **Knowledge plugin cost blow-ups** (coders) → directly threatens trust + usability for “knowledge” features.
2) **Configuration/validation confusion** (coders) → raises setup failure rate; increases support load.
3) **Plugin callback behavior confusion** (partners/tech) → docs vs runtime mismatch perception.
4) **Token/airdrop ambiguity + chain liquidity issues** (discussion) → sentiment + retention risk.
5) **Security/scam bot incident** (support) → trust + user safety risk.

---

## 2) User Experience Intelligence

### 2.1 Feedback categorized by impact & theme
**High impact (blocks adoption / causes cost shock)**
- Knowledge plugin:
  - Cost shock when `CTX_KNOWLEDGE_ENABLED=true` (unexpected token explosion).
  - Validation error for `EMBEDDING_PROVIDER` enum mismatch (`openrouter` not allowed).
- Airdrop eligibility questions (Kraken, Solana DEX trading) unresolved → increases rumor load.

**Medium impact (slows builders / creates confusion)**
- Plugin action handler callbacks perceived as “only first callback works” → likely planner configuration misunderstanding (onestep vs multistep), but perceived as framework bug.

**Low impact (opportunity / future growth)**
- Seeker phone integration idea: potential distribution channel with weak incumbents.
- Desire for specialized single-purpose agents (developer GTM clarity issue).

### 2.2 Usage patterns vs intended design
- **Observed:** Users attempt “OpenRouter everywhere” (chat + embeddings) → hits enum constraints and/or unintended defaults.
- **Intended:** Clear separation between **chat provider** and **embedding provider**; currently under-documented and easy to misconfigure.
- **Observed:** Users enable contextual embeddings without understanding cost profile → mismatch between “simple query” expectation and “expensive RAG pipeline” reality.

### 2.3 Implementation opportunities (product + docs + guardrails)
- **Guardrails:** Make high-cost modes *opt-in* with explicit warnings and runtime cost estimates.
- **UX copy:** Improve error messages (e.g., “`EMBEDDING_PROVIDER` does not accept openrouter; use `openai` and configure `OPENROUTER_EMBEDDING_MODEL` instead”).
- **Golden paths:** Provide two official configs:
  1) **Budget RAG** (no contextual embeddings)
  2) **High-recall RAG** (contextual embeddings + explicit “expected cost” table)

### 2.4 Community sentiment (trackable signals)
- Sentiment bifurcation:
  - **Builders:** optimistic, focused on workflow automation pivot.
  - **Holders/community:** anxious about price + eligibility + liquidity; calls for memetic revival.
- Risk: Without cadence + clarity, discussion channels will continue to be dominated by **price/airdrop uncertainty** rather than product learning.

---

## 3) Strategic Prioritization (Impact × Risk × Dependencies)

### 3.1 Critical path (next 7–14 days)
**Goal stated by core-devs:** End-of-week top-of-funnel: **landing page → poke integration → agent conversation**  
This is the **primary forcing function**. Everything else should be staged behind it.

**Dependencies**
1) Poke integration must reliably hand off to agent runtime (latency + session reliability).
2) Agent personality spec (“funny, spunky, spicy, not mean”) must be testable (prompt + safety boundaries).
3) Basic observability: conversion + drop-off events across funnel.

### 3.2 Initiative scoring (recommendation matrix)

| Initiative | User Impact | Tech Risk | Effort | Recommendation |
|---|---:|---:|---:|---|
| **Funnel MVP (lander → poke → convo)** | Very High | Medium | Medium | **Do now (P0)** |
| **n8n integration MVP (prefab workflows + conversational creation)** | Very High | High | High | **Do next (P0/P1), but scope tightly** |
| **Knowledge plugin cost + config fixes (docs + defaults + warnings)** | High | Low–Med | Low | **Do now (P0)** |
| **OAuth Gateway (multi-tenant credentials)** | High | High | High | **Start design now; implement incrementally (P1)** |
| **Workflow validation/testing system** | High | Medium | Medium | **Start immediately; prevents “generated workflow” failures (P1)** |
| Plugin callback behavior clarification (docs + examples) | Medium | Low | Low | **Quick win (P1)** |
| Babylon airdrop comms (eligibility + timeline) | Medium–High (sentiment) | Low | Low | **Do now (P0 for comms)** |
| Seeker platform integration | Medium (distribution upside) | Medium | Medium | **Defer until funnel instrumented (P2)** |
| PumpFun tracking feature requests | Medium (niche) | Medium | Medium | **Backlog (P3)** |

### 3.3 Resource allocation (based on current parallel tracks)
- **Sam (Comms runtime + chat surfaces):** keep focus on **SMS/iMessage plugins + cloud runtime integration** *only insofar as it supports funnel conversion and retention*. Avoid expanding to new transports until baseline funnel metrics exist.
- **Stan (n8n + OAuth):** concentrate on **n8n plugin + OAuth gateway skeleton** and a minimal **credential passing contract** (plugin creds → n8n creds).
- **Odilitime (PIM/memory + platform coherence):** continue **plugin-pim** but explicitly align memory model to workflow automation needs (e.g., remembering connected accounts, workflow preferences, user goals).

---

## 4) Actionable Recommendations (Concrete, measurable)

### 4.1 Reduce support load + cost incidents (Knowledge plugin)
**Ship within 48–72 hours**
1) **Documentation patch (minimum):**
   - Add “Cost Modes” section:
     - `CTX_KNOWLEDGE_ENABLED=false` (default recommended)
     - Expected token ranges
     - Example shown: **3.1MB PDF → 517 chunks** as a baseline reference point
   - Add “Provider wiring” section:
     - `EMBEDDING_PROVIDER` valid values; clarify OpenRouter usage pattern.
2) **Runtime warnings (low engineering effort, high ROI):**
   - On startup, if `CTX_KNOWLEDGE_ENABLED=true`, log a **yellow/red warning** with approximate cost expectations and a link to docs.
3) **Better validation error text:**
   - When enum mismatch occurs, print “Did you mean: `EMBEDDING_PROVIDER=openai` + set `OPENROUTER_EMBEDDING_MODEL=openai/text-embedding-3-small`?”

**Success metric:** cut repeated “why is this so expensive?” questions by **>50%** week-over-week in coders channel.

### 4.2 n8n workflow automation MVP (avoid overbuilding)
**Define MVP as: “3 workflows + conversational creation + safe execution”**
- Prefab workflows (start library):
  1) “Summarize unread emails and post to Notion”
  2) “Create Linear ticket from message + context”
  3) “Calendar scheduling + follow-up email”
- Conversational builder:
  - “Closest match” retrieval + Opus-based mutation
  - Mark generated flows as **EXPERIMENTAL**
- Execution safety:
  - Credential-based validation gate (“required creds present?”)
  - Dry-run mode (simulate nodes; no side effects)
  - Human confirmation for destructive actions

**Success metrics**
- % of users who create at least one workflow within first session
- Workflow success rate (runs without manual debugging)
- Median time-to-first-automation (TTFA)

### 4.3 Funnel instrumentation (to guide prioritization with data)
Implement event tracking for:
- Landing page view → poke click → conversation started → conversation > N turns → “connected an account” → “first workflow run”
- Tag all drop-offs with client + server timestamps.

**Success metric:** identify top 2 drop-off points within 72 hours of launch and ship 1 fix/day.

### 4.4 Community comms to stabilize sentiment (airdrop + updates)
- Publish a single canonical post:
  - “Babylon airdrop: confirmed in principle, details pending”
  - Explicitly answer: Kraken custody, Solana DEX trading eligibility status (even if “TBD”)
  - Provide an **update cadence** (“weekly on X day”)
- Security PSA: warn about ticket bots asking for wallets; pin message.

**Success metric:** reduce repeated eligibility questions; increase ratio of product questions vs price questions in discussion channel.

---

## 5) Key Risks & Mitigations

1) **Risk: n8n integration expands scope uncontrollably**  
   - Mitigation: MVP definition + “3 workflows first” + workflow validation gate.

2) **Risk: Cost shocks from Knowledge plugin reduce trust**  
   - Mitigation: default-safe config, explicit warnings, cost-mode docs, and “budget RAG” preset.

3) **Risk: Token/airdrop ambiguity dominates narrative**  
   - Mitigation: official cadence + FAQ, even if answers are partial.

4) **Risk: Scam/social engineering**  
   - Mitigation: pinned PSA + moderation workflow + “never share wallet” policy in onboarding docs.

---

## 6) Watchlist (Signals to monitor next)
- Volume of support questions about:
  - `EMBEDDING_PROVIDER` enum
  - contextual embeddings costs
  - airdrop eligibility
- Adoption of `@elizaos/core@next` (downloads, issues filed)
- Early funnel conversion and TTFA for workflow automation
- n8n instance operational stability (workers/scaling) once real users arrive