Intelligence / Knowledge systems / Organizational learning

Organizational Intelligence
Daily Briefing for Steve

An executive lens on emerging signals, established practices and what they could mean for better decisions.

Editorial pilot22 November 2026 · Pacific · Manual publication; automatic refresh not yet verified

Executive highlights

01Knowledge maintenance is becoming an active AI workflow. Microsoft introduced a SharePoint team knowledge base that Copilot and agents can curate and reference. It is in Frontier Preview, with usage-based billing restrictions. Implication: Explore whether an intelligence register can become a governed, actively maintained knowledge product rather than a passive document. Microsoft, 14 Oct 2026 ↗
02Scanning is most useful when linked to decisions. OECD's 2025 review describes continuous horizon scanning paired with interpretation and policy experimentation, including UK weekly technology scanning and German monitoring dashboards. Implication: Build a visible pathway from signal → assessment → decision or experiment. OECD, 2025 ↗
03Stale knowledge is a governance risk. A systematic review of 65 studies identifies mechanisms by which knowledge systems reinforce obsolete routines. Implication: Make revalidation, contradiction handling and retirement explicit register functions. Review, 2025 ↗

ChatGPT & Microsoft 365 Copilot

01Microsoft 365 Copilot — team knowledge bases (Frontier Preview). SharePoint knowledge bases can be curated by people and agents; access currently requires usage-based billing, and some integrations remain on the roadmap. This is not general availability. Pilot idea: compare provenance and refresh controls with the ChatGPT register. Microsoft announcement ↗
02Microsoft 365 Copilot — reusable skills (GA). November 2026 developer documentation lists Skills in Copilot Chat as generally available, supporting reusable instructions with optional resources and scripts. Application: standardize scanning, evidence grading and jurisdictional synthesis prompts. Verify tenant access. Developer updates ↗
03Microsoft 365 Copilot — notebooks and measurement. October release notes describe web links as Notebook references and expanded Copilot adoption metrics. Application: combine external source collections with evidence on team usage. Availability may vary by tenant and rollout. Release notes ↗
04ChatGPT — enterprise knowledge and governance. Company knowledge, introduced in October 2025, brings connected business sources together with citations and existing permissions. This is a baseline capability, not a new November 2026 release. Recheck configuration and migration plans before operational use. OpenAI ↗ · Enterprise/Edu notes ↗

Changes to previous intelligence

No confirmed revision to an earlier register entry yet. The register was initially empty, so this first issue establishes a baseline. Continue checking Microsoft knowledge-base preview status, pricing and curation claims, and enterprise ChatGPT connector and migration timelines. Record any later reversal with the old wording, new evidence and date.

Practices worth exploring

01UK Government Office for Science: OECD describes a weekly process collecting scientific publications, third-sector reports and other signals, then synthesizing patterns and emerging themes. Transferable feature: A repeatable cadence with human interpretation, not merely a news feed. OECD ↗
02Germany's federal technology monitoring: OECD describes indicators covering 16 technology areas and over 90 technologies, plus an experimental radar. Transferable feature: Combine structured monitoring with exploratory signal detection. OECD ↗
03Adaptive knowledge governance: A 2025 systematic review recommends time-based reviews, diverse discovery methods, flexible governance and periodic challenges to accepted knowledge. Transferable feature: Treat deliberate correction and retirement as core processes. Systematic review ↗

Watchlist and potential experiments

AEvidence-to-action: For two weeks, classify each finding as signal, established practice, actionable change or watchlist. Record whether it led to a decision, discussion or experiment.
BRegister integrity: Seed 3–5 sourced entries, then test a finding that contradicts one entry. Check whether the original conclusion and source remain visible.
CTool comparison: Test the same source pack with ChatGPT and Microsoft 365 Copilot using an identical synthesis rubric, assessing citations, completeness, correction handling and effort.
DWatchlist: SharePoint knowledge-base preview → GA; Copilot skill availability in the tenant; ChatGPT connector governance; automated Page publication.
10-minute challenge

One Idea to Try

The “So What?” Test

Pick one intelligence product your team produces. Ask: What decision could someone make differently after reading this?

Today's challenge: Rewrite its opening paragraph to make the decision relevance unmistakable.

Original exercise inspired by decision-oriented intelligence practice; not an attributed quotation.

The Lighter Side

A little workplace humour

xkcd #303 — Compiling

When the computer is busy, apparently everyone gets a break.

xkcd Compiling: two programmers have a sword fight on office chairs while their code compiles; their manager accepts the excuse.

Cartoon by Randall Munroe · Original xkcd comic · CC BY-NC 2.5 licence. Shared for this non-commercial editorial pilot.