Digital

Owned and paid digital surfaces — the website, the app, search, and everything bought programmatically against them.

Back to the workflows index.

---

What this vertical means here

Digital OS is the most developed spoke on the roadmap — page 3 is a whole whiteboard devoted to it — and the most developed in the product.

Its banner states the thesis for the entire programme:

NO AGENCY

Built with Co-Pilot – LLM

Everything in this vertical is scoped as work an agency currently does, to be replaced by workflows.

---

Workflows

---

What's live, in detail

News → Blog, end to end

The most complete workflow in the product, and the only one that reaches an external system.

queue a news item
      ↓  autoProcessBlogQueueItems — background, unattended
generate draft   ← brand master prompt · SEO keywords · competitor brief · compliance rules
      ↓
quality repair   ← self-graded: structure, length, keyword placement
      ↓
SEO bolding      ← first-use <strong> on assigned keywords
      ↓
ZeroGPT detect → up to 2 humanization passes  (fails open)
      ↓
blocked-phrase scan → content_warnings
      ↓
state: "ready"   ← STOPS HERE
      ↓
[human] compliance review → acknowledge high-severity findings
      ↓
[human] Send to CMS

It stops at ready deliberately. Sending to the CMS is always a human action, and a high-severity compliance finding blocks it at the server with a 409 until acknowledged.

Driven by the Creative Agent and gated by the Compliance Agent.

Discovery

Source-first crawl plus optional search-first discovery. Extraction runs a cascade — local Readability → Tavily → Crawl4AI/Apify — with the extractor chosen per tenant.

Every row records its scraperSource, so the extractor mix is queryable rather than guessed at. An article-shape URL filter (isArticleShapedUrl) rejects landing and hub pages before they reach the feed.

SEO

Two live surfaces: an imported keyword inventory, and gap analysis that turns uncovered keywords into blog topics. Selected keywords flow into blog generation and are bolded on first use, with unused ones reported as warnings rather than silently dropped.

Geographic tagging — the partial

The Discovery Agent already tags every feed item with an Indian state/UT or National, and Discovery has a statewise filter.

That is the input to page 3's Scale Growth workflow — Discovery → Nat / State / City / Micromarket — but the workflow itself (page structure, personalised experience per market, publish/rollout, CRO) does not exist. The data is there; nothing consumes it at scale.

---

What's missing

All paid media. Page 3 names a full stack — Zocket, Pixis, DWAO as buying platforms, across Google (UAC/Search/Display), Meta (7 streams), Affiliate, OEM, SST, LinkedIn, ASP. None is integrated. There is no spend, no budget, no campaign object anywhere in the product.

CRO and funnel management. Page 3 draws CRO → Full funnel mgmt for both customer and employee journeys, via Clevertap / Netcore / Humanoid. Unbuilt.

Hyperlocal. A "single interface" for web and app hyperlocal is specified. Unbuilt.

Personalised experience. Appears on three separate pages of the roadmap. Nothing personalises anything today — every reader of a blog gets the same blog.

The planning frame. Page 3's left column (KKH = Planning — where & what → budgeting → execution → discovery → creation → dynamic budget → setup → rollout → CRO → monitoring → optimization) describes a planning loop that does not exist as a workflow.

---

Dependencies

---

Workflow specs in this folder

• CMS Delivery
• Discovery
• Feed Enrichment
• News to Blog (E2E)
• Performance Media
• Scale Growth (Geo SEO)
• SEO Gap Analysis

---

Roadmap source

Pages 1 (Digital spoke and its fan-out), 3 (the whole board), 5 (deliverable ②, Blog E2E), 6 (Blog tonality, imagery, guard rails). See ../../roadmap_ref_extracted.md.