🔬 Concept walkthrough — screens shown here represent directional UI thinking, not final product. Built to clarify the build picture for leadership review. Wording, layout, and interactions will be refined during implementation.
Phase 1 · Editorial Intelligence Platform · Executive Walkthrough

The agent does the scroll.
The editor runs the show.

How Phase 1 turns a daily flood of social, RSS, and indexed-web content into a curated pulse, an AI-assisted composition surface, and a finalized brief — without anyone manually scrolling LinkedIn.

1

Title

Subtitle

intel.[client].com/editor
Design rationale
Time recovered, compounded

What three hours a day adds up to

The editorial team's current 3-hour daily floor for gathering, compounded across a production calendar, is meaningful capacity. The math, scaled out:

Per day
~3
hours recovered
A meaningful chunk of the working morning — the gathering already done when the team starts.
Per week
~15
hours recovered
Nearly two full working days returned to higher-value editorial work every week.
Per month
~65
hours recovered
More than a week and a half of recovered editorial focus, every month.
Per year
~780
hours recovered
Roughly 40% of an FTE editor's annual capacity — nearly 20 working weeks of focus returned to the work that shapes the show.
The picture in one line: nearly 20 working weeks per year of editorial time the team currently spends finding what to talk about — returned to the work of shaping how to talk about it.

Math: 3 hours per day × 5-day production week × 52 weeks. Adjusts to your actual production cadence. FTE equivalence based on a 2,080-hour working year.

Secure by architecture

Designed around your security controls.

The platform runs inside your Google Cloud tenant, governed by your existing IAM, audit, and network policies. Every architectural choice is made to keep data in your controlled environment and to integrate with the security operations you already have.

01 · Data Residency
Data stays in your tenant
Production runs in your Google Cloud tenant. Gathered content, editorial outputs, and configuration all remain inside your controlled environment. No vendor-side hosting.
02 · Data Scope
No customer or client PII
The platform processes public social content and editorial outputs only. No customer records, no advisor-client interactions, no books-and-records material under SEC 204-2.
03 · Identity
Workspace SSO + role-based access
Three named seats authenticated via your existing Google Workspace. No separate user database. Your access reviews and offboarding workflows apply automatically.
04 · LLM Boundary
Gemini via Vertex AI
Every LLM call runs through Vertex AI inside your tenant. Data never leaves Google Cloud. No third-party LLM providers in the data path; no model training on your data.
05 · AI Risk Handling
Prompt injection mitigated
Public content from Reddit and X is treated as untrusted input. Strict prompt structure separates instructions from data. The editor reviews every AI output before it reaches the brief.
06 · Audit Trail
Auditable end to end
Cloud Audit Logs on every GCP service. Application logs capture every brief edit, configuration change, and AI co-pilot interaction with user identity and timestamp.

Note: Because the platform is deployed inside your GCP tenant, your existing security controls — IAM policies, audit log retention, network rules, key management — apply by default. The controls described here supplement those, they do not replace them.