title: AI Brief
AI Brief — what's new in enterprise AI
A curated digest of what's happening in enterprise AI — filtered for what's actually practical for Scotts, not the hype cycle.
How this is filtered
We pull from a small set of trusted sources (AI Daily Brief, vendor enterprise blogs, select enterprise-AI commentary), look at the last 14 days, and shortlist 5 items per fortnight: 3 Gemini-stack picks + 2 broader enterprise AI picks. The Gemini bias reflects what our associates use company-wide today; the broader picks keep us aware of the landscape (Claude, governance, agentic patterns, EU AI Act timeline). Every item gets a "why this matters for Scotts" line — that's the part that does the real work.
Two fortnights stacked. The current digest sits at the top; the prior fortnight's stays visible below it so the trajectory is legible (e.g., last fortnight's Shared Gems → this fortnight's Shared conversations). Older fortnights roll off the page each cycle.
What gets boosted:
- Gemini stack — Gemini app/Enterprise/Workspace, NotebookLM, Gems, Vertex AI updates that affect our daily tools
- Anthropic / Claude enterprise news (we move to Claude Enterprise in FY27)
- Governance, policy, audit, compliance for enterprise AI
- COE patterns, change management, adoption frameworks
- Agentic deployment patterns and tooling
- Cost / ROI signal (token economics, deployment patterns)
What gets suppressed:
- Raw research papers and model benchmark wars
- Consumer features that don't map to enterprise use
- Hype-cycle takes without a concrete enterprise hook
Cadence: Refreshed every Wednesday (the day before office hours).
Top 5 — week of 2026-08-12
Covers 2026-07-29 → 2026-08-12. Anchored on the MCP 2026-07-28 spec shipping and EU AI Act GPAI obligations activating August 2. Hand-curated this fortnight; the scheduled filter wires up before the next digest.
Workspace Studio can now automatically add sources to Gemini Notebooks
The news. A new "Add Source to Gemini Notebooks" step is rolling out in Workspace Studio from August 6, allowing users to automatically pull content from Drive, Docs, Slides, and other Workspace sources directly into a Gemini Notebook as part of a Studio workflow. The feature removes the manual step of opening a notebook and adding sources separately — sources are populated in the background as part of a broader automation chain.
Why this matters for Scotts. This is a direct workflow unlock for the NotebookLM Enterprise intake card's core use case: keeping notebooks populated with current operational content without manual curation overhead. Combined with the auto Drive-sync feature from Issue 3, notebooks can now be both seeded and kept current automatically — making the case for a pilot significantly easier to scope and present to the Governance Committee.
Gemini Enterprise custom MCP servers just got easier to configure
The news. Gemini Enterprise has simplified custom MCP server setup by removing the requirement for description fields. The platform now automatically calls the server's tools/list endpoint to discover tools, parameter schemas, and capabilities directly — routing requests using explicit tool definitions rather than admin-written descriptions. This reduces setup friction and improves routing accuracy for private data connectors.
Why this matters for Scotts. The custom MCP server feature (in Preview since July) is the architectural path to connecting Gemini Enterprise agents to internal Scotts systems. This update lowers the barrier to standing up that connection: IT teams no longer need to hand-craft descriptions for each tool, reducing both setup time and the risk of misconfigured routing. Anyone scoping the Blocked Stock Workflow or Production Planning Tool on Gemini Enterprise should note this simplification before beginning a proof-of-concept.
New Gemini Enterprise overage billing controls arrive August 17 — admins should review now
The news. New overage billing controls for Gemini Enterprise are rolling out on August 17, 2026. Admins who have received the billing update email ([Billing Update] New Gemini Enterprise overage billing controls launching Aug 17, 2026) will see a change to how overages are handled. Organisations not on the new billing model are unaffected. Full details are in the Admin console under Gemini Enterprise billing settings.
Why this matters for Scotts. Before August 17, an admin should confirm whether Scotts' Gemini Enterprise account is on the new billing model and review the overage thresholds and controls available. As usage grows across the associate base — particularly if MCP connectors or agentic workflows are added — unexpected overage charges are a real risk without appropriate spend controls in place. This is a low-effort action that protects the budget case for continued Gemini Enterprise investment.
Next two: outside the Gemini stack — MCP's biggest protocol update yet (directly relevant to FY27 Claude Enterprise integration) and the EU AI Act GPAI obligations that activated August 2.
MCP goes stateless — and Claude is the first major client to support it
The news. The MCP 2026-07-28 specification — the largest protocol update since MCP launched in November 2024 — shipped on July 28. The headline change is a stateless core: MCP servers can now deploy on serverless and edge infrastructure without managing session state, making them far easier to scale and operate. Authentication hardened to OAuth 2.0/OIDC, with enterprise managed auth allowing admins to provision MCP connectors centrally via their identity provider (Entra, Okta). Claude became the first major client to add support the same day. MCP now has 950+ connectors and 400 million monthly SDK downloads — a 4x increase this year.
Why this matters for Scotts. Both Gemini Enterprise (custom MCP connectors, Preview) and the FY27 Claude Enterprise platform build on MCP as their agent integration layer. The stateless architecture and enterprise managed auth in 2026-07-28 are the two features that make MCP viable for Scotts-scale deployment: servers can run on managed infrastructure without bespoke session handling, and connector access can be provisioned through existing identity systems rather than user-by-user setup. This is the infrastructure story underpinning the Agentic Workflow Builder intake card.
EU AI Act GPAI systemic risk obligations are now live — general-purpose AI models in scope
The news. The EU AI Act's transparency obligations (Article 50) and general-purpose AI model (GPAI) systemic risk requirements activated on August 2, 2026. Providers of GPAI models with systemic risk — including frontier models from Google, Anthropic, OpenAI, and xAI — must now maintain technical documentation, conduct adversarial testing, report serious incidents, and implement cybersecurity protections. Enterprise deployers using these models inherit disclosure obligations: any AI-generated content in customer-facing or employee-facing contexts must be labelled as such.
Why this matters for Scotts. The August 2 activation is the compliance milestone that was flagged as imminent in our Issue 3 brief (before the HRAIS deadline correction in Issue 4). The immediate practical obligation for Scotts is the Article 50 transparency requirement: any associate-facing or customer-facing AI tool that generates content should carry a disclosure. The AI Governance Committee should confirm which deployed tools are in scope and whether current disclosures are sufficient — this is a lower-effort action than the full HRAIS compliance program, but it is legally required now.
EU AI Act official page → · AI Governance Institute August 2 briefing →
Top 5 — week of 2026-07-28
Covers 2026-07-14 → 2026-07-28. No major anchor event this fortnight — picks drawn from the NotebookLM rebrand, Gemini Enterprise model upgrades, Claude Opus 5's launch, and a governance alert on agentic tooling as shadow IT. Hand-curated this fortnight; the scheduled filter wires up before the next digest.
NotebookLM is now Gemini Notebook — and it's deepening its Workspace ties
The news. Google rebranded NotebookLM as Gemini Notebook on July 16, rolling out to all Workspace customers over the following weeks. The product functionality and APIs are unchanged, and existing notebooks remain accessible. The rename signals Google's intention to evolve the tool beyond a standalone research app into a connected layer across Gemini products. The enterprise edition is similarly renamed Gemini Notebook Enterprise, though the subscription page still shows the old NotebookLM Enterprise label for now.
Why this matters for Scotts. The Hub's NotebookLM Enterprise intake card is the most mature near-term adoption candidate on the ideas page. The rebrand is a signal that Google is deepening investment in the product rather than sunsetting it — a positive indicator for the business case. Admins should note that the Admin console and Help Center labels will update over the coming weeks; no configuration changes are required.
Gemini 3.6 Flash arrives in Gemini Enterprise Agent Designer
The news. Gemini 3.6 Flash is now available in Gemini Enterprise Agent Designer workflow agents, giving builders access to Google's latest Flash model tier when designing and running agents. Updates take up to a day to propagate to existing workflow agents. As a related heads-up: Gemini 3.5 Flash is being retired from the global region in the Gemini Enterprise app on August 4 — any agent prompts specifically tuned to 3.5 Flash behaviour should be reviewed before that date.
Why this matters for Scotts. Any agent being built under the Agentic Workflow Builder intake card, or scoped through the Hub's solution pipeline, should be designed against 3.6 Flash rather than 3.5 Flash from this point forward. The August 4 retirement of 3.5 Flash in the global region is an operational deadline: a quick audit of any agents already in Agent Designer is worth running this week to confirm there are no hardcoded model references.
Gemini Alpha is now Gemini Beta — what admins need to know
The news. Google's Gemini Alpha early-access programme is being renamed Gemini Beta, with the label change rolling out in the Admin console and Help Center over the coming weeks. No admin configuration changes are required and no features are affected. New Gemini Beta features will continue to be announced in the Google Developer Programme forum and the Gemini Beta Help Center; organisations must be active Workspace customers and register to access GDP content.
Why this matters for Scotts. If the Hub or IT team has any documentation, runbooks, or training materials that reference "Gemini Alpha", those should be updated to "Gemini Beta" ahead of the Admin console label change completing. More broadly, this is a maturity signal: moving from Alpha to Beta framing typically precedes GA releases, so features currently in Gemini Beta are likely on a shorter path to general availability than their Alpha predecessors were.
Next two: outside the Gemini stack — Claude Opus 5's arrival reshapes the FY27 platform economics, and a governance alert on agentic developer tools as the new shadow IT.
Claude Opus 5 launches at half the price of Fable 5 — and outperforms it on key enterprise benchmarks
The news. Anthropic released Claude Opus 5 on July 24, available across the Claude API, Claude.ai Pro/Max/Team/Enterprise, Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry. It is priced identically to Opus 4.8 ($5/$25 per million tokens) but benchmarks at near-Fable-5 performance on coding and knowledge work, and supports zero data retention — a significant governance advantage over Fable 5, which requires 30-day retention. An effort toggle (low/medium/high) lets enterprises balance cost and capability per task. The model ID is claude-opus-5.
Why this matters for Scotts. The FY27 Claude Enterprise transition now has a clearer cost model: Opus 5 delivers near-frontier capability at the same price Scotts would already pay for Opus 4.8, with zero data retention unlocking use cases previously blocked by data residency concerns. The effort toggle is directly relevant to agentic workflows where not every step needs maximum reasoning depth. Worth updating the FY27 platform evaluation brief with Opus 5 pricing and the zero-data-retention flag before the next Governance Committee review.
Agentic developer tools are the new shadow IT — and the blast radius is bigger
The news. A widely-cited July 15 analysis from the AI Governance Institute argues that standard shadow IT controls — blocking unauthorised SaaS subscriptions, revoking access — do not apply to agentic coding assistants, which operate at codebase level, transmit code as part of their core function, and expose data in proportion to the developer's own privileges. The same week, China's agent governance rules took effect (July 15) and Illinois mandated third-party AI safety audits, creating simultaneous compliance deadlines for multinationals. Google separately proposed a US federally-overseen frontier AI safety body on July 5.
Why this matters for Scotts. Any associate using an agentic coding tool — Antigravity CLI, Claude Code, GitHub Copilot, Cursor — with access to internal systems represents a governance exposure that existing IT controls were not designed to catch. The AI Governance Committee should add agentic developer tooling to its next intake review: a lightweight inventory of which tools are in active use, what data they can access, and whether any reach internal systems is the minimum baseline before August's EU AI Act transparency obligations activate.
AI Governance Institute analysis → · July 3 agentic shadow IT alert →
Past digests
Older digests are archived chronologically. We don't keep weekly digests around forever — items older than 30 days roll off the front, and only the ones referenced by a Solution Library entry stay linked from the library.