Independent deployments will determine whether ChatGPT Company Knowledge delivers reliable, permission-aware retrieval across organizational data for Business, Enterprise, and Edu users.

state: expiredheat: lowuncertainty: highconvergesscott: mediumenterprise-rag knowledge-systems agent-memoryOpenAI

What is this?

Company Knowledge is an OpenAI feature for ChatGPT Business, Enterprise, and Edu that retrieves organizational information from connected workplace apps such as Slack, Google Drive, SharePoint, and GitHub. OpenAI says it preserves each user’s existing source permissions, supports administrator-controlled apps, and is powered by a version of GPT-5. The supplied results establish availability and vendor claims, but provide no independent deployment evidence testing retrieval reliability or permission enforcement; the web answer’s attribution to Amazon conflicts with the cited sources.

Why it matters to Scott

OpenAI is a consequential new entrant validating Scott’s cross-silo organisational knowledge territory, although its connector-based retrieval is not yet evidence of his stronger compiled, governed semantic-layer architecture. The unsupported reliability and permission claims create a concrete publishing and evaluation opportunity using his Capability Audit and Route-Invariant Grounding frameworks, rather than merely another generic RAG launch.

ip:framework.bi-for-soft-dataip:concept.capability-auditip:framework.route-invariant-groundingip:concept.containment-by-representationdev:project.mcp-ip-wikiradar:concept.ragradar:concept.knowledge-systemsradar:concept.enterprise-airadar:cloudflare-ai-search-private-data

queries asked of Scott's wikis

permission-aware enterprise RAG evaluations
connector-based organizational knowledge retrieval
RAG authorization and access-control leakage
enterprise knowledge system deployment lessons
agent memory versus governed company knowledge
retrieval reliability across heterogeneous work apps

Measured heat

no measured readings yet — the hourly heat pass fills this in

How the heat travelled

no chain yet — the hourly chain pass fills this in

Evidence (2) — ⭐ canonical anchor

source	object	author	score	comments
🟧 hn	Company Knowledge in ChatGPT (Business, Enterprise, and Edu)	ssuds	2	1
🟧 echo.blog ⭐	OpenAI’s original announcement says: “Today we’re introducing company knowledge for ChatGPT Business, Enterprise, and Edu.” It describes sea	OpenAI	—	—

Interpretation history

2026-08-28T05:24:03Z
Repeated checks through the monitoring ladder produced no independent deployment, retrieval-quality, or permission-isolation evidence, so the launch-era validation episode has faded without advancing. Future substantive implementation evidence should open a fresh episode rather than keep this vendor-claim watch alive.
2026-08-26T04:31:54Z
No new evidence accompanied the staleness trigger, leaving Company Knowledge a vendor-established capability without independent deployment, retrieval-quality, or permission-isolation validation. The hot adjacent agent-memory topic does not advance this specific case; retain a weekly monitoring cadence.
2026-08-24T03:29:17Z
The latest check is only repetitive amplification; no independent deployment, retrieval-quality, or permission-isolation evidence changes the vendor-established evaluation target. Keep the case open on a weekly cadence because enterprise validation may emerge slowly.
2026-08-22T02:32:43Z
No independent deployment, retrieval-quality, or permission-isolation evidence has emerged, so the case remains an unvalidated vendor capability rather than a demonstrated enterprise knowledge system. Further checks should move to a weekly cadence unless substantive implementation evidence appears.
2026-08-20T02:28:53Z
Another stale reobservation supplies no independent deployment, retrieval-quality, or permission-isolation evidence. The surrounding topic remains active, but this case has not advanced beyond a vendor-established evaluation target.
2026-08-18T01:31:54Z
The case remains a vendor-established evaluation target with no independent deployment or permission-isolation evidence. Repeated stale reobservation adds no meaning, so monitoring can move to a slower cadence.
2026-08-16T01:22:53Z
No independent deployment or permission-isolation evidence has appeared; the slight engagement change adds no meaning. The case remains an evaluation target rather than validation of OpenAI’s reliability claims.
2026-08-14T00:36:32Z
No independent deployment, implementation, or permission-isolation evidence has arrived; the case remains a vendor-established capability awaiting real-world validation. This reobservation adds no substantive meaning beyond the already-routed launch.
2026-08-14T00:35:05Z
grounded: converges/medium — OpenAI is a consequential new entrant validating Scott’s cross-silo organisational knowledge territory, although its connector-based retrieval is not yet eviden
2026-08-14T00:33:16Z
origin walked (codex/luna, conf 0.99): anchor hn.story.49292963 -> echo.blog.1ec8e8f39d by OpenAI
2026-08-14T00:32:19Z
case created — OpenAI’s first-party documentation establishes a concrete enterprise retrieval capability that can be evaluated in deployed workflows.

Decision trace

08-28 15:24expireRepeated checks through the monitoring ladder produced no independent deployment, retrieval-quality, or permission-isolation evidence, so the launch-era validation episode has faded without advancing.
08-28 15:24alert_silentThe only delta is negligible engagement on already-known launch material; there is no new evidence or action for Scott, and the case has exhausted its monitoring window.
08-28 15:24alert_routeThe only delta is negligible engagement on already-known launch material; there is no new evidence or action for Scott, and the case has exhausted its monitoring window.
08-28 15:23drop_targetsquiet through full ladder or over cap 8
08-26 14:31repriceNo new evidence accompanied the staleness trigger, leaving Company Knowledge a vendor-established capability without independent deployment, retrieval-quality, or permission-isolation validation. The
08-26 14:31alert_silentThis is a scheduled stale reobservation rather than a consequential delta; it can wait until an independent deployment report, retrieval benchmark, or permission-control assessment appears.
08-26 14:31alert_routeThis is a scheduled stale reobservation rather than a consequential delta; it can wait until an independent deployment report, retrieval benchmark, or permission-control assessment appears.
08-24 13:29repriceThe latest check is only repetitive amplification; no independent deployment, retrieval-quality, or permission-isolation evidence changes the vendor-established evaluation target. Keep the case open o
08-24 13:29alert_silentA one-point engagement increase is not consequential evidence; wait for an independent deployment report, retrieval benchmark, or permission-control assessment.
08-24 13:29alert_routeA one-point engagement increase is not consequential evidence; wait for an independent deployment report, retrieval benchmark, or permission-control assessment.
08-22 12:32repriceNo independent deployment, retrieval-quality, or permission-isolation evidence has emerged, so the case remains an unvalidated vendor capability rather than a demonstrated enterprise knowledge system.
08-22 12:32alert_silentThis is another stale reobservation with no consequential new evidence; it can wait for an independent deployment report, retrieval benchmark, or permission-control assessment.
08-22 12:32alert_routeThis is another stale reobservation with no consequential new evidence; it can wait for an independent deployment report, retrieval benchmark, or permission-control assessment.
08-20 12:28repriceAnother stale reobservation supplies no independent deployment, retrieval-quality, or permission-isolation evidence. The surrounding topic remains active, but this case has not advanced beyond a vendo
08-20 12:28alert_silentThe score change is inconsequential and no substantive evidence has arrived; wait for an independent deployment report, retrieval benchmark, or permission-control assessment.
08-20 12:28alert_routeThe score change is inconsequential and no substantive evidence has arrived; wait for an independent deployment report, retrieval benchmark, or permission-control assessment.
08-18 11:31repriceThe case remains a vendor-established evaluation target with no independent deployment or permission-isolation evidence. Repeated stale reobservation adds no meaning, so monitoring can move to a slowe
08-18 11:31alert_silentNo new evidence or consequential event occurred; the next briefing can wait for an independent deployment report, retrieval benchmark, or permission-control assessment.
08-18 11:31alert_routeNo new evidence or consequential event occurred; the next briefing can wait for an independent deployment report, retrieval benchmark, or permission-control assessment.
08-16 11:22repriceNo independent deployment or permission-isolation evidence has appeared; the slight engagement change adds no meaning. The case remains an evaluation target rather than validation of OpenAI’s reliabil
08-16 11:22alert_silentThis is only a stale reobservation with negligible engagement movement and no substantive evidence; it can wait for an independent deployment report or security evaluation.
08-16 11:22alert_routeThis is only a stale reobservation with negligible engagement movement and no substantive evidence; it can wait for an independent deployment report or security evaluation.
08-14 10:36repriceNo independent deployment, implementation, or permission-isolation evidence has arrived; the case remains a vendor-established capability awaiting real-world validation. This reobservation adds no sub
08-14 10:36alert_silentOnly a legacy-state re-evaluation occurred, with unchanged engagement and no new evidence; the established launch was already routed and can wait for independent deployment results.
08-14 10:36alert_routeOnly a legacy-state re-evaluation occurred, with unchanged engagement and no new evidence; the established launch was already routed and can wait for independent deployment results.
08-14 10:35alert_shadowOpenAI has introduced cross-workplace search with citations and claimed enforcement of existing source permissions, directly entering Scott’s organisational-knowledge territory and creating a timely e
08-14 10:35alert_routeOpenAI has introduced cross-workplace search with citations and claimed enforcement of existing source permissions, directly entering Scott’s organisational-knowledge territory and creating a timely e
08-14 10:35groundOpenAI is a consequential new entrant validating Scott’s cross-silo organisational knowledge territory, although its connector-based retrieval is not yet evidence of his stronger compiled, governed se
08-14 10:33promote_anchororigin walk conf 0.99
08-14 10:32createOpenAI’s first-party documentation establishes a concrete enterprise retrieval capability that can be evaluated in deployed workflows.