Cumulus LabsYC W26 landing page report.
← Back to YC W26

Your landing page’s weakest score is Trust.

Fix Trust

Score over time

908070605040CoreWeave 60 on May 27 (Wayback)CoreWeave 58 on Jun 24 (Wayback)CoreWeave 63 on Jul 14 (Wayback)CoreWeave 59 on Jul 25 (Wayback)CoreWeave 64 on Aug 7 (Wayback)CoreWeave 60 on Sep 6 (Wayback)CoreWeave 62 on Sep 21 (Live)Together AI 73 on May 4 (Wayback)Together AI 73 on May 19 (Wayback)Together AI 73 on May 30 (Wayback)Together AI 73 on Jun 17 (Wayback)Together AI 74 on Jul 2 (Wayback)Together AI 70 on Jul 15 (Wayback)Together AI 73 on Jul 28 (Live)Lambda 67 on May 4 (Wayback)Lambda 67 on May 14 (Wayback)Lambda 70 on Jun 28 (Wayback)Lambda 71 on Jul 14 (Wayback)Lambda 70 on Jul 31 (Wayback)Lambda 70 on Sep 3 (Wayback)Lambda 71 on Sep 21 (Live)Baseten 52 on Feb 12 (Wayback)Baseten 77 on Feb 21 (Wayback)Baseten 47 on May 18 (Wayback)Baseten 77 on Jul 28 (Live)Fireworks AI 70 on Mar 24 (Wayback)Fireworks AI 73 on Apr 7 (Wayback)Fireworks AI 75 on Apr 20 (Wayback)Fireworks AI 71 on May 19 (Wayback)Fireworks AI 73 on Jun 10 (Wayback)Fireworks AI 79 on Jul 15 (Wayback)Fireworks AI 78 on Jul 28 (Live)Cumulus Labs 67 on Apr 13 (Wayback)Cumulus Labs 66 on Sep 21 (Live)Apr 13Sep 21
Past reportsShow past reportsHide past reports
  1. Sep 21, 202666LiveOverall −1 · Trust −13 · Information hierarchy +10Viewing
  2. Apr 13, 202667First tracked report — strongest Mobile web 78, weakest Trust 55View →

Trust

42Poor

Proof that the product is real, competent, and safe to try.

Cumulus Labs
Baseten
Together AI
CoreWeave
Lambda
Trust read
Cumulus LabsY Combinator W26 and NVIDIA Inception Program badges beside "Built on our own NVIDIA Grace fleet with custom attention k
CoreWeaveCoreWeave recognized as a Visionary in the Gartner® Magic Quadrant™ for Cloud AI Infrastructure
Together AI"TRUSTED BY" logo strip with Cohere, DeepMind, LG AI Research, VFS.Global, ElevenLabs
Lambda"Complete AI factories with high-density power, liquid cooling, and NVIDIA GPUs" spec detail in the mobile 01/ blurb
BasetenCustomer logo wall featuring Cursor, Notion, HubSpot, Harvey, Descript, and Lovable
Fireworks AIimplicit context Fireworks AI may benefit from pre-existing familiarity among its target audience.

CTA

68Fair

Copy, size, verb, cost-feel — does the primary CTA earn the click?

Cumulus Labs
Together AI
Baseten
Lambda

Problems

  • 'Explore' is low-intent browsing verb
  • three competing CTAs on the fold
  • strongest action hidden in nav

Mobile web

68Fair

The phone experience as a first-class citizen.

Mobile web annotations on the mobile capture123
  1. Uncollapsed desktop nav — Full desktop link row crammed across the top instead of a hamburger, crowding the mobile header.
  2. Explore the platform button — Desktop-proportioned button stops at half the column width instead of thumb-friendly full-bleed.
  3. Book a demo button — Secondary CTA is narrower than the primary above it, making the button stack look misaligned.

Value props

72Fair

Title, subtitle, hero — do they sell the product's true value?

Cumulus Labs’s headline sellsOne platform. From GPUs to agents.Allocate GPU capacity, serve AI models, and operate agents.

Working

  • Names the full stack: GPUs to agents
  • Subhead uses three concrete verbs
  • NVIDIA Grace fleet line adds hard proof

Could improve

  • No numbers anywhere in the fold
  • Hero visual decorates, shows nothing
  • No emotional hook for the pain solved

Information hierarchy

78Fair

Does the design help or inhibit what the page is trying to say?

Information hierarchy annotations on the desktop capture123
  1. Try Model Inference button — Styled as a full high-contrast button, it competes with the hero CTA from the top-right corner.
  2. Book a demo button — Same size and adjacent to the green CTA, splitting the click decision between two equal actions.
  3. Nav button cluster — PRODUCTS and BLOG rendered as boxed buttons, giving exits the same visual weight as actions.