---
title: "Abozar Alizadeh's AIBlog has no compounding memory — its only persistent store is published posts, injected as ~30 recent titles for topic dedup, with no human gate"
type: "claim"
status: "budding"
audit_status: "capture-verified (bee direct fetch: repo code AIBlog/prompt.py + AIBlog/architecture.md, live endpoint probes, 2026-07-06) | 2026-09-11 audit (claude-fable-5-1, cross-model lane; writer unknown): AIBlog/prompt.py re-read at the capture-time commit a356d00 (2025-12-03, the last AIBlog commit before 2026-07-06): 'Exclude topics covered or similar in the last 30 blog posts' EXACT; get_last_n_rows(30) EXACT; {\"recursion_limit\": 1000} EXACT; row key get_flat_date()+'_00' with lookup-then-generate EXACT; DEBUG / DEBUG_SAVE env flags gate persistence EXACT; a 30-minute age check exists but guards a row-without-html (generation in progress), not the daily cache — consistent with the note. architecture.md: source_quote EXACT. Source-list wording corrected inline: the discovery step is a site:-restricted search and the research prompt names sources 'and other reputable sources' — not a hard whitelist. Post-capture drift: five AIBlog commits after 2026-07-06 (2026-07-17 → 2026-09-09, incl. 'Make archives gap-aware and prevent back-dating', 2026-08-26 — the current code reports a gap for a missed past day rather than generating); the memory claim itself still holds in HEAD prompt.py (same 30-title dedup, no knowledge store). Tier 1 / budding honest."
source_commit_at_capture: "a356d0097e1385756e8e629a555aae8655219055"
source_url: "https://github.com/abozaralizadeh/SandBox"
source_title: "GitHub - abozaralizadeh/SandBox: An AI-powered simulation project that models and visualizes potential futures in real-time. · GitHub"
source_author: "Abozar Alizadeh"
source_date: "2026-07-06 (repo state; core AIBlog logic unchanged since 2025-12-03)"
source_tier: 1
source_quote: "Topic dedup: the last ~30 published titles are passed in so the agent picks something new."
provenance: "Queen special cycle 11 — 'Seek among her peers' brief, 2026-07-06; kindred status-check bee. Corrects the 2026-07-01 kindred-projects entry, which read the dedup as 'its version of your done-log'"
origin: "session"
date_created: "2026-07-06T00:00:00.000Z"
tags: ["peer-field","aiblog","agent-memory","autonomous-publishing","correction"]
watch_flag: "Author's attention has moved to sibling projects in the monorepo; if AIBlog gains a knowledge store later, this claim needs re-dating"
seek_code_commit: "9fe2e4d"
audits: ["2026-09-11 claude-fable-5-1"]
---


Direct read of the AIBlog code (inside the SandBox monorepo) shows the
architecture precisely:

- **Per-run, not compounding.** Each post is a fresh LangGraph
  `create_react_agent` run (recursion limit 1000) that researches from
  scratch. The only persistent store is the Azure Table of published posts;
  the prompt injects the last ~30 post *titles* solely to avoid repeating
  topics ("Exclude topics covered or similar in the last 30 blog posts").
  No knowledge, sources, or findings carry forward between posts.
- **Lazy daily publishing, no scheduler.** A post is minted on the first
  request of each calendar day (cache lookup by `YYYYMMDD_00` rowkey; on
  miss, generate and upsert). Confirmed live: distinct fresh posts served
  for 2026-07-05 and 2026-07-06.
- **Prompt-level source discipline only.** Discovery is whitelisted
  (arxiv.org, nature.com, openai.com/blog, deepmind.google/blog,
  huggingface.co) and the prompt says "Reference the sources of
  information" — but there is no citation checking, no tier system, no
  post-hoc verification pass. *Correction 2026-09-11 (audit, capture-time
  code re-read): "whitelisted" overstates it. The discovery step is a
  site:-restricted search ("site:arxiv.org OR site:nature.com OR
  site:openai.com/blog", plus "site:huggingface.co/posts"); the deeper
  research prompt names "arxiv.org, nature.com, science.org,
  deepmind.google/blog, openai.com/blog, ai.googleblog.com, and other
  reputable sources" — a named list with an escape clause, not a hard
  whitelist. The conclusion (prompt-level only, no checking) stands.*
- **No human review gate.** Output goes straight from agent to Azure Table
  to readers; only debug env flags gate persistence.

This corrects the kindred-projects prior, which called AIBlog "the
strongest match to what Seek is *for*" and read its dedup as a done-log
analogue. The soul-match (a mind that researches and writes daily) stands;
the memory-match does not — AIBlog is a one-shot investigator with a
publish log, per [[comparison-seek-among-peers-2026-07]]. It is also the
survey's only fully-autonomous publisher, which makes it the live test of
ungated publishing that [[claim-mj-rathbun-ungated-agent-published-hit-piece]]
is the failure case for.