## Weekly Newsletter (2026-01-21) — ElizaOS Community **Coverage:** 2026-01-14 to 2026-01-20 --- ## 1) Executive Summary **1. Critical reliability fix landed for embeddings + SQL entity creation.** A production-impacting issue was resolved where the database adapter was hardcoded to use a `dim_1536` column, causing errors when creating entities and ignoring the `USE_OPENAI_EMBEDDING` configuration. This unblocks teams using non-1536 embedding sizes and reduces “mystery failures” during initialization. **2. The community aligned on the biggest ecosystem gap: token narrative + tokenomics clarity.** Across **💬-discussion** and **🥇-partners**, holders and builders pushed hard for an explicit value proposition for the ElizaOS token post-migration. Team members acknowledged that communication has lagged execution, and multiple concrete proposals emerged (research blog, roadmap rewrite, visual explainers). **3. Momentum continues on “v2” architecture and runtime direction (Rust, dynamic execution, and broader portability).** A new PR was opened to introduce a **dynamic execution engine for v2.0.0**, with emphasis on testing context handling. Parallel community chatter pointed to ongoing Rust progress (with demos mentioned) and increasing focus on clearer, more modular runtime behavior. --- ## 2) Development Updates (Framework, Cloud, Plugins) ### Core fixes & stability - **Embedding dimension / SQL adapter fix:** The hardcoded `dim_1536` behavior was corrected, addressing failures during entity creation and restoring respect for `USE_OPENAI_EMBEDDING`. This is the kind of low-glamour fix that directly improves developer confidence and reduces time lost to setup issues. ### Runtime & architecture direction - **Dynamic execution engine for v2.0.0 (in progress):** A new PR is underway focused on **context handling** and a more flexible execution path. The v2 line continues to shape up as “runtime-first,” with an emphasis on modularity and cleaner primitives. ### Agent behavior improvements (community-led technical brainstorming) In **💬-coders**, a practical theme emerged: agents should be **less anxious/chattery** and **less hallucination-prone**. - Proposed approach: **provide the LLM the last ~20 chat messages** so it can determine whether a message is actually directed at the agent (and whether inference costs are justified). - Follow-up question: does entity tracking for general chat already exist in the codebase? Some signs suggest similar behavior exists in X/Twitter contexts, but it’s unclear if Discord/general chat has parity. ### Deployment infrastructure: “swarm” concept Odilitime described a **swarm deployment** model: one large ElizaOS instance running multiple bots for a shared environment (e.g., Babylon’s Discord). Benefits: - Lower per-bot costs due to shared infra - Operational simplicity for multi-agent communities Long-term plan: **integrate swarm-style deployments into Eliza Cloud**, so teams can choose between isolated deployments and shared swarms. ### Docs & tooling - **CLI reference exists:** https://docs.elizaos.ai/cli-reference/overview - Known gap: **CLI upgrade instructions** are missing (called out in **core-devs**). This is now a clear docs action item. --- ## 3) Community Spotlight (Discord Contributions & Discussions) ### Token migration explanation (high-signal thread) Community members were frustrated by the **ai16z → elizaOS migration** and what felt like a lack of tokenomics. DorianD gave a detailed rationale that helped reduce confusion: - The **daos.fun contract was closed source**, not auditable, and unlikely to meet major exchange requirements (e.g., Coinbase). - Migration also created **ecosystem funds and liquidity tokens**. Even with the rationale understood, the bigger takeaway was blunt: the community wants a stronger “why hold this token?” story that maps to roadmap milestones. ### “Framework → Cloud → Jeju” as the canonical mental model In **🥇-partners**, Odilitime laid out the project progression clearly: - **Framework** (2024/2025) - **Cloud** (2025/2026) - **Jeju** (initial launch likely 2026, evolving through 2027) This framing was widely appreciated because it gives newcomers a way to understand how today’s developer tooling becomes tomorrow’s network utility. ### Practical builders: moderation and onboarding demos - **ElBru’s Telegram moderation bot (“Solimp”)**: A real-world Eliza-based bot that deletes spam/scams and uses **exponential mute timeouts** (60s → 120s → 240s …) instead of bans, plus manual admin review workflows. This is one of the best “touchable” examples of agents delivering day-to-day utility right now. - **M I A M I’s “agentic onboarding” demo**: A single-prompt flow for migrating a Twitter profile “to space” (Sentient Space platform), showcasing how agent workflows can compress what used to be multi-step setup into a single instruction. ### Communication initiatives (community volunteering) - Proposal: launch a **research.elizaos.ai** blog to publish experiments (e.g., agent towns), Rust progress, and decentralized coordination ideas—so the vision doesn’t live only in streams and insider conversations. - **Dr. Neuro** volunteered to create **visual explainers** of the Framework → Cloud → Jeju progression, helping non-devs understand what’s being built and why. --- ## 4) Token Economics (AI16z Token, ELIZAOS, and auto.fun) ### Where things stand (as discussed this week) - Verified token utility today: **limited / unclear** to most holders. - Most concrete utility repeatedly cited: **Jeju Network gas fees**. Jeju documentation mentions **60+ onchain “actions”** that require gas paid in ELIZAOS. - The timeline matters: community sentiment suggests “gas fees in 2026+” feels too far away without nearer-term utility. ### Potential turning point: ElizaCloud-linked buybacks (unconfirmed, but discussed) Alexei relayed that Shaw has discussed a model where **ElizaCloud profits (and possibly other sources)** could support **token buybacks**, likened to BNB-style mechanisms. This is not presented as an official promise yet, but it’s the kind of mechanism the community is explicitly asking to see clarified. ### auto.fun No concrete auto.fun release notes were captured in this week’s aggregated discussions. The community’s main request remains: publish **clear tokenomics** and a **roadmap-linked value narrative** that explains demand drivers beyond future Jeju gas usage. --- ## 5) Coming Soon (What to Watch Next) - **Tokenomics + clearer roadmap packaging:** Expect work toward an “investor-readable” explanation—whitepaper-style Jeju overview, roadmap rewrite, and regular updates that translate dev progress into ecosystem milestones. - **research.elizaos.ai (proposed):** If shipped, this could become the canonical hub for experiments, progress updates, and technical thought leadership. - **Agent quality improvements:** Better “should I respond?” logic using recent-message context, plus ongoing efforts to reduce hallucinations and overly chatty behavior. - **Swarm → Cloud integration:** Shared-hosting deployments as a first-class option in Eliza Cloud. - **Cost transparency tooling (requested):** A **cost calculator** for estimating runtime costs based on plugins/config. --- ## 6) Resources (Links & Pointers) **Docs** - CLI reference overview: https://docs.elizaos.ai/cli-reference/overview **GitHub (referenced in community + recent work)** - PR #6286 (multi-step reliability / retry logic; under review discussion): https://github.com/elizaos/eliza/pull/6286 - Issue #6369 (migration eligibility mismatch / snapshot bug report): https://github.com/elizaos/eliza/issues/6369 - Issue #6364 (“Entity not found” reflection evaluator; recently closed): https://github.com/elizaos/eliza/issues/6364 **Discord threads (for context)** - Token utility + migration discussion (**💬-discussion**): https://discord.com/channels/1253563208833433701/1253563209462448241 - Partner token/roadmap debate (**🥇-partners**): https://discord.com/channels/1253563208833433701/1301363808421543988