Launch WaveMind developer marketing campaign

[CaspianG]

Objective

Run a public developer launch for WaveMind and turn early feedback into concrete repository work.

Launch assets

• docs/LAUNCH_KIT.md
• docs/RU_LAUNCH_POSTS.md
• docs/assets/wavemind-social-card.svg
• README demo, benchmark, and limitation sections
• Benchmark report and leaderboard in benchmarks/

Scope

• Publish a clear launch post for developers who build long-running agents.
• Show the core product story: persistent memory, dynamic priority, TTL, namespaces, corrections, benchmarks, and local-first operation.
• Share reproducible benchmark links instead of broad claims.
• Track useful feedback from GitHub, X, Reddit, HN, and developer communities.
• Convert recurring questions into docs, examples, or issues.

Acceptance criteria

• GitHub Actions are green before public promotion.
• README, release notes, and benchmark report agree on current limitations.
• Public posts include install command, one runnable demo, and benchmark evidence.
• Feedback is triaged into follow-up issues with labels and clear scope.
• No benchmark or production-readiness claim is made without a checked-in result or reproducible command.

Positioning guardrails

• WaveMind is a dynamic memory layer, not a general-purpose vector database replacement.
• Static vector search can still be faster for plain RAG over large document collections.
• Strong claims should focus on memory behavior: stale suppression, corrections, TTL, namespaces, hotness, auditability, and long-memory evidence retrieval.

[CaspianG]

Launch work started.

Completed public repository actions:

• Enabled GitHub Discussions.
• Set repository homepage to PyPI.
• Updated repository description and topics.
• Updated v2.1.0 release notes with install, demos, benchmarks, and limitations.
• Answered installation help issue can someone help me to install pls #1.
• Opened launch/community discussion: WaveMind launch: dynamic memory for agents and apps #8
• Opened benchmark reproduction discussion: Benchmark evidence and reproduction reports #12
• Added contributor funnel issues:
  • Generate benchmark leaderboard from checked-in results #9 benchmark leaderboard
  • Create README demo GIF from offline dynamic memory demo #10 README demo GIF
  • Write migration guide from Chroma memory to WaveMind dynamic memory #11 Chroma migration guide

External launch channels are prepared but not published yet. The next step is to publish from the approved materials in docs/LAUNCH_KIT.md after confirming the final launch timing and community rules for each channel.

[CaspianG]

Added a runnable customer support / CRM memory demo in commit 000e209. It is an offline launch asset that shows corrected CRM data outranking stale data, temporary discount TTL, customer namespace isolation, and audit-friendly SQLite state. CI is green: tests https://github.com/CaspianG/wavemind/actions/runs/28674175032 and full-check https://github.com/CaspianG/wavemind/actions/runs/28674175060.

[CaspianG]

Added another runnable launch asset in commit 1f10050: examples/research_notebook_memory.py. It demonstrates source-aware analyst memory: confirmed findings with provenance, temporary hypotheses expiring, decisions/action items in one project namespace, and isolation between unrelated research projects. CI is green: tests https://github.com/CaspianG/wavemind/actions/runs/28677312515 and full-check https://github.com/CaspianG/wavemind/actions/runs/28677312497.

[CaspianG]

v2.2.2 launch prep is now live.

Completed:

• refreshed English and Russian launch materials;
• fixed the Russian launch post encoding;
• added the latest production-load benchmark story to docs/LAUNCH_KIT.md;
• published GitHub Release v2.2.2 with clearer notes;
• published wavemind==2.2.2 to PyPI;
• verified local tests, build, and package validation before tagging.

Current public benchmark angle:

• WaveMind is strongest as a dynamic memory layer: TTL, corrections, stale suppression, namespaces, hotness, and auditability.
• Qdrant service is strong at 100k vectors; the first 1M run is included honestly as a tuning target (recall@10 0.506, 45.81 ms).

Next launch work:

• tune 1M recall;
• add real Mem0/Zep/LangGraph comparison adapters;
• prepare a visual Studio/demo asset for public launch posts.