Glow Back to the investor site

Diligence

AI-native operating model

Not "AI-powered" as positioning. The directory that would traditionally need an operations team was built by a pipeline, and the run logs show it.

Automated — shipped and verified

  • Business discovery across zones and categories
  • Classification and enrichment
  • Deduplication and brand/branch resolution
  • Area banding and scoring

Automated — designed, not yet live

  • Conversation seeding and community follow-up questions
  • Content tagging
  • Spam triage
  • Advertiser reporting
  • Trend detection

AI with human review

  • Moderation escalation
  • Claims naming individual service providers
  • Listing disputes
  • Quality audit of published pages

Human owned

  • Product and strategy
  • Closing advertisers
  • Brand partnerships
  • Strategic accounts

The claim is operating leverage, not a company without people. Where today's models aren't reliable enough — anything involving reputation, disputes or money — humans own the work outright.

The modelled headcount comparison

Held off the main narrative as supporting evidence rather than a headline. Human sales headcount is identical in both columns.

MAUTraditional FTE AI-native FTETraditional OpExAI-native OpEx
1M10224₹9.42cr₹2.42cr
5M511121₹47.09cr₹12.12cr
10M1,022 242₹94.17cr₹24.24cr

Sanity check: Yelp runs roughly 4,400 employees against a comparable local-review audience — about 4.4 FTE per 100k MAU, sitting between this model's traditional benchmark (10) and its AI-native assumption (2.2). That is the right place for it to sit.

Geographic expansion economics

FootprintAreasLanguages BuildMaintain / yrLocalisation First year
Bangalore122₹0.8L₹0.5L₹36L₹37L
Top-8 metros965₹6.3L₹3.7L₹90L₹100L
50 cities2809₹18.3L₹10.9L₹162L₹191L
100 cities42011 ₹27.4L₹16.4L₹198L ₹242L ($252k)

Two findings. First, supply construction across 100 cities costs about $28,500 — less than two operations hires, and the figure the narrative headlines. Second, localisation is 82% of the first-year geographic cost, not data acquisition. The ₹2.42cr first-year total is kept here rather than on the main page: it is correct, but it bundles a large modelled localisation component and reads as a budget rather than an insight.

The precise claim, and its limit

Supported: national supply coverage scales on software. Mapping and maintaining beauty businesses across 100 cities needs no proportional field organisation.

Not supported: that total operations headcount stays flat. Moderation scales with content volume, native-language QA with languages, and advertiser servicing with advertisers.

The defensible sentence is therefore narrower than the ambitious one: Glow can achieve national geographic coverage without a national field organisation, because supply construction and maintenance run on software. Community and advertiser operations still scale with usage and revenue — sub-linearly against a traditional operator, but not flat.