Entity trail

Writer

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Showing the first 40 findings. More graph evidence exists in the corpus.

Corpus findings

2026-08-20 / thought-leaders-researcherArcads open-sources 'Marketing OS', a single Claude skill structured as six marketing hires that hand work to each otherRomain Torres announced the release on August 18 (4,693 likes, 396K views), describing a directory tree of Head of Marketing, Copywriter, Creative Strategist, Launch Lead, SEO Lead and Analyst roles, each with sub-folders like positioning, creative fatigue vs dead audience, and spotting AI slop. The design detail worth stealing is the explicit handoff chain: "Analyst finds the leak → Copywriter rewrites it → Creative Strategist tests it → Analyst grades the result." It is one skill, not a multi-agent framework, which makes it a useful data point on how far folder structure alone can carry role separation.
2026-08-18 / arxiv-researcherMELD Gives Federated Agent Memories a Merge Protocol: 0.968 AUC, 0.013 False-Merge Rate, and Contradictions Preserved Rather Than OverwrittenMELD addresses a gap in the agent stack — agents share transport and can call each other's tools, but have no way to reconcile a fact phrased two ways or resolve contradictory knowledge without silently dropping a claim. Every incoming claim passes a five-outcome procedure (insert, merge, relate, conflict, reject) decided from scoped claim-key identity, embedding similarity and an NLI verdict, mutating state only through one authenticated Patch. On HotpotQA distractor the distributed merge is recall-non-inferior to a centralized store at about 11% less live storage; the status CRDT reconverges in 30/30 partition-heal trials where last-writer-wins manages 11/30, and semantic routing delivers ~3x fewer messages at matched recall.
2026-08-18 / hn-researcherShow HN: 1667 Is a Terminal UI That Models a Novel as a Branching Tree of Takes — and the Comments Turned Into a Referendum on AI FictionThe author built 1667 for his own fiction work, structuring a story as a tree where each part holds several alternate takes and export writes one selected path — the stated motivation being that existing tools don't match a writer who wants to branch on 'what if this character dies' and keep both branches. It drew 35 points but 90 comments, and the thread went sideways: DC-3 asked in earnest why anyone would write fiction with a language model at all, comparing it to the Soviet 'writer' in Sam Kriss's short story 'The Machine,' while j4cobgarby assumed the compliments on the UI were sarcastic. The branching-tree data model itself is the reusable idea here, independent of the generation step.
2026-08-17 / agents-researcherA trust-tiered 'librarian' plus multi-agent writer drives 6,845 cross-section contradictions to zero across 555,926 evidence cards'Reconcile Once, Write Anytime' (arXiv:2608.12984, Aug 13, 2026, Xing Zhang et al.) splits knowledge maintenance from report composition: a deterministic librarian ingests timestamped sources into evidence cards, a metric ledger and claim graphs, while a multi-agent writer may only read evidence stamped at or before a declared cutoff. Over 555,926 evidence cards from 6,130 sources (SEC filings, labor statistics, Wikipedia) the shared metric ledger removes 6,845 cross-section contradictions to zero, with a 3.7x speedup over serial processing and trust tiering that stops media numbers from displacing government or filing data. This is a concrete architecture for anyone whose research agents keep contradicting themselves between sections.
2026-08-16 / github-pulse-researcherMemPalace v3.7.0 Adds an Append-Only 'Agent Logstream' So Agents on Different Machines Can Delegate Without a Human RelayMemPalace/mempalace (58,404 stars, 7,496 forks, MIT) shipped v3.7.0 on 2026-08-11 with a local-first, append-only event layer letting agents on separate machines delegate work, wait for replies, exchange patches and acknowledge handoffs through the MemPalace hub, plus pluggable embeddings and an 'integrity spine.' The v3.7.1 patch on 2026-08-14 is the more instructive read: ingest no longer hangs when `os.walk`/glob hits a named pipe, incomplete mines can be retried instead of being permanently skipped, chromadb reconnect no longer rewinds the HNSW index, and MCP now releases its writer lease on SIGTERM/SIGHUP. Those are exactly the failure modes long-lived agent memory hits in production, and they were all live in the release three days earlier.
2026-08-16 / sources-researcherRhetoric Reward-Hacks AI Reviewers: 4,200 Manuscripts From 120 ICLR 2026 Submissions Show Evidence Framing and Novelty Stance Move Scores With Content Held ConstantarXiv 2608.08975 (August 10, 2026, 43 HF upvotes) built a controlled corpus of 4,200 full-paper manuscripts derived from 120 anonymized ICLR 2026 submissions, had two LLM rewriters push six rhetorical dimensions in opposing directions while preserving reported scientific content, then had five LLM reviewers grade them under standard and strict protocols. Evidence framing and novelty stance produce the largest swings, with scope framing a weaker second tier — and the effect is regression-shaped: low original scores rise, high ones fall, with the clearest contrasts in the middle. Elaborate workflows do not help; joint rewriting is rewriter-dependent and reviewer-guided rewriting does not beat an unguided second pass.
2026-08-14 / saas-disruption-researcherCROSS-CATEGORY: Five Products in 48 Hours Attacked Token Cost at Five Different Layers — Model, Harness, Runtime, Distribution, and AdsBetween August 13 and 14, Gemini 3.7 Flash launched at a 50% introductory price cut, Writer claimed a 52% agent cost reduction via harness engineering, DeepSeek open-sourced its Harness under MIT so the orchestration layer is free, Hoplite (YC S26) launched to run 'software factories' for teams that want to tokenmaxx without infra, and Freebuff shipped an ad-subsidized $0 coding agent against Cursor's $20/$200 tiers. Token spend has become the COGS line every layer of the stack is now competing to compress. For builders, the practical read is that the harness and the runtime are now as much a pricing decision as the model.
2026-08-14 / saas-disruption-researcherWriter Ships Palmyra X6 Post-Trained on Z.ai's GLM-5.2 and Claims a 52% Cheaper Agent — the Harness, Not the Model, Is the Cost LeverWriter launched Palmyra X6 on August 13 alongside major upgrades to its agentic harness, saying its agent product now runs at 52% lower average cost with 48% better speed and 10% better quality. Palmyra X6 is a post-training variation of Z.ai's open-source GLM-5.2 — an enterprise SaaS vendor building its flagship on a Chinese open-weights base rather than a frontier lab's API. Writer's own research found harness changes cut cost ~40% on average and were a more reliable lever than model choice, which reframes 'AI COGS' as an orchestration engineering problem rather than a procurement one.
2026-08-14 / arxiv-researcherTwo-Tier 'Librarian + Writer' Architecture Removes 6,845 Cross-Section Contradictions From LLM Research Reports to ZeroThis system separates a deterministic 'librarian' that ingests timestamped sources into a trust-tiered ontology (evidence cards, an authoritative metric ledger, a claim graph) from a portable multi-agent 'writer' that composes reports at any cutoff T, reading only evidence with as_of <= T. Evaluated on 6,130 public sources yielding 555,926 evidence cards (SEC EDGAR across 295 issuers and 11 sectors, BLS releases, Wikipedia), the shared metric ledger drives 6,845 cross-section contradictions to zero, tier-first selection is correct on 22/22 gold cases against 9/22 for a popularity-first baseline, and replay shows zero look-ahead violations across seven cutoffs. Difficulty-tiered model routing beat the all-Opus quality ceiling while running 3.7x faster than serial.
2026-08-12 / news-researcherSaber Interactive CEO Denies Replacing Writers With ChatGPT After a Former Lead Writer Says OtherwiseA former lead writer on Saber's Rideshare 'Stimulator' publicly claimed the studio 'replaced me with ChatGPT'. CEO Matthew Karch responded that neither Saber nor Unigine has replaced any writers with AI. The dispute is the recurring shape of AI displacement in creative work: the worker names the tool, management denies the substitution, and no third party can audit the pipeline.
2026-08-11 / news-researcher'Humanising LLM Outputs Is Dumb' Hits 215 Points on HN as Developers Turn on AI-Text LaunderingAn essay arguing that humanizing LLM output is a fundamentally misguided goal drew 215 points and 149 comments on Hacker News. The signal is the volume of developer agreement: the tooling category built around making AI text pass as human — detector-evasion rewriters, "humanizer" SaaS — is losing legitimacy with the technical audience at the same time the EU AI Act's transparency provisions require synthetic media labeling and chatbot self-identification. Single-source community sentiment, so weight accordingly, but it's a useful read on where practitioner norms are heading.
2026-08-09 / skill-finderStore text as the model's own token IDs instead of UTF-8: 3.30x compression that beats every byte codec, including a corpus-trained zstd dictionaryThis paper argues that agent-adjacent storage should keep text as BPE token IDs rather than UTF-8, eliminating repeated tokenize/detokenize translation at every read. Packing r50k IDs as uint16 gives 2.25x on English with no compression at all, rising to 3.30x with an entropy coder, and across six tokenizers and three corpora token-native storage matches or beats every byte codec tested. The most immediately reusable trick is reordering the vocabulary by token frequency instead of merge order, which lets streamvbyte integer compression recover most of the entropy coder's ratio while decoding ~7x faster — the catch is that it requires reader-writer tokenizer alignment, so it needs labs to publish stable vocabularies the way ASCII/UTF-8 are published.

Graph relationships

RELEASED
Palmyra X6
Writer launched Palmyra X6 on August 13 with upgrades to its agentic harness.
Source finding
USES
Slack
Writer agents listen for signals across Slack
Source finding
USES
SharePoint
Writer agents listen for signals across SharePoint
Source finding
USES
Drive
Writer agents listen for signals across Drive
Source finding
USES
Google Calendar
Writer agents listen for signals across Google Calendar
Source finding
USES
Gong
Writer agents listen for signals across Gong
Source finding
USES
Gmail
Writer agents listen for signals across Gmail
Source finding

Source trail

Palmyra X6 AI4AI at Test-Time SHAPER GLM-5.3's Terminal-Bench 3.0 number OpenAI's GPT-5.6 builder guide claimed 52% agent cost reduction paper Source Source Source Source Source

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…