Microsoft

Confidence 0.85 · 3 sources · last confirmed 2026-07-01

Microsoft is the cloud-and-developer-tools giant that owns GitHub (since 2018), partners deeply with OpenAI, and ships the developer surfaces central to the wiki’s agentic-coding corpus: Azure, Azure DevOps, Visual Studio Code, and the responsible-AI pipeline that proxies Copilot requests. Promoted to an entity page on 9 June 2026 via cross-page-presence promotion (precedent: LangChain, Andrej Karpathy): Microsoft is a plain-text mention across ~21 source pages, and the Agentic DevOps keynote — published on the Microsoft Visual Studio channel — is the first source centred on Microsoft’s own agentic developer tooling.

Backlinking note. source_count is set to 3 (formal inbound source-page wikilinks), not ~21 (plain-text mentions). Backfilling the historical mentions into wikilinks is a deferred lint cleanup.

Role in the wiki

Microsoft is the platform-and-cloud vantage on agentic development — the vendor whose IDE (VS Code), cloud (Azure), ALM tooling (Azure DevOps), and safety layer (responsible-AI pipeline) host the agentic workflows that GitHub’s Copilot drives. The channel-author convention applies: Microsoft Visual Studio (the YouTube channel) maps to Microsoft as the publishing entity.

Visual Studio Code

Microsoft’s cross-platform editor — the keynote’s hub for agentic development. The four Copilot interaction modes: Ask / Edit / Agent / Plan (Plan is the newest — the agent states its implementation plan before acting, bounding the over-autonomy failure mode). Instructions files encode house rules + exemplars; agents run locally (CPU/GPU) or push to GitHub’s cloud; profiles swap toolsets; Ollama + the Continue extension enable offline/local-model agents.

Azure DevOps + the responsible-AI pipeline

Copilot agents are coming to Azure DevOps (private preview) — but routed through GitHub (“you still have to link it up to GitHub”). Every Copilot request passes through Microsoft’s responsible-AI pipeline, which validates input and the returned code (security vulnerabilities, NSFW) even when the chosen model is third-party (Anthropic/Google) — sending only enough across the wire and checking the response before it reaches the developer, in milliseconds. See responsible-ai.

Azure SRE Agent

Microsoft’s autonomous cloud-monitoring + incident-response agent (preview). Scope it to an Azure resource group; it performs incident management, can be permissioned to autonomously remediate certain conditions, and files health/security reports — framed as “a site reliability engineer on my team.” A worked instance of an autonomous agent in the IT-operations (not just coding) domain.

OpenAI partnership

Microsoft hosts and routes OpenAI models (and now the Codex agent inside Copilot), alongside Anthropic and Google models. The keynote frames model choice as a developer decision, with Microsoft’s proxy doing the safety validation regardless of model.

Satya Nadella / Build 2026 — the CEO-worldview source

The Possible interview (5 Jun 2026, post-Build 2026) is the wiki’s first platform-CEO worldview source and Microsoft’s second formal inbound source. Nadella frames AI as the future of the firm (not a technology), introduces token capital as the firm-level asset that compounds alongside human capital, and lays out Microsoft’s enterprise-AI stack as the manageability/security answer: the hill-climbing machine (evals/objectives as the new IP), Agent 365 (agent identity/sandbox/policy/observability via Entra/Defender/Purview — see responsible-ai), Foundry asserts (runtime boundary enforcement — see agent-harness), and the agentic development environment (the GitHub app as an inbox of agents with Canvas, for micro-steering of macro-delegation). The dictum “don’t use frontier models for non-frontier problems” anchors his token-efficiency argument. Co-designed silicon (Maya, Cobalt) and the platform-company-defined-by-value-created-on-top trust doctrine round out the strategy.

WorkLab podcast — the owned-media future-of-work vantage

Microsoft’s WorkLab podcast (host Molly Wood) is a separate publishing channel from Microsoft Visual Studio — owned media focused on the future of work broadly, not Microsoft’s own developer tooling. Wood (29 Jun 2026) is the wiki’s first WorkLab source: an interview with LinkedIn’s Chief Economic Opportunity Officer Aneesh Raman on AI-era career and organizational identity (“onlyness,” the 5 C’s, the electricity-bolt-on-failure analogy). The channel-author convention applies here as it does for Microsoft Visual Studio: WorkLab maps to Microsoft as publishing entity, with LinkedIn (a Microsoft subsidiary since 2016) as the guest’s employer — a third, futures-of-work-focused channel alongside the developer-tooling (Agentic DevOps) and CEO-worldview (Nadella/Possible) vantages already held.

Open questions

• Full backlinking. ~21 plain-text “Microsoft” mentions across the corpus await promotion to wikilinks.
• Product sub-pages. Azure, VS Code, Azure DevOps, and the Azure SRE Agent are Dangling product mentions; promote on a second source each.
• The Microsoft–OpenAI commercial relationship. Referenced but not characterised; a dedicated source would let the wiki track it properly.

Mentioned in

• 2025-12-22-randell-gousset-microsoft-agentic-devops-in-real-life — the promotion trigger: VS Code modes, Azure DevOps, responsible-AI pipeline, Azure SRE Agent.
• GitHub — Microsoft subsidiary; the developer-platform half of the keynote.
• 2026-06-29-raman-wood-worklab-job-titles-dont-matter-2026 — WorkLab podcast; Aneesh Raman (LinkedIn) on AI-era career and organizational identity.