Melius $25M Raise: Assets Beat AI Optimizers

The cleanest product lesson sometimes arrives in the messiest sentence: “We scrapped the entire codebase; we burned all of it.” That was Melius co-founder Joowon Kim, describing to TechCrunch what happened after the startup’s first AI marketing idea failed to catch. The new product is not another dashboard asking marketers to tune more knobs. It is a platform for generating the workflow artifacts buyers can use immediately: ad campaigns, images, and videos.

The launch that survived the burn

TechCrunch reported that New York-based Melius announced a total of $25 million in funding, made up of a $20 million Series A led by CRV and a $5 million seed round led by General Catalyst. The company also claims it crossed more than $1 million in annualized revenue within two months of coming out of stealth in July, according to TechCrunch. The founders, Joowon Kim, Young Kim, and Arnav Ramu, knew each other from working as engineers at Ramp, the corporate spending and finance software company. That is a useful backdrop: this team did not merely rename a deck and call it a launch, it rebuilt the thing it was selling.

TechCrunch says the first version of Melius launched over a year ago as an AI performance marketing tool. More than six months after building it, the co-founders decided the original idea did not “have legs,” Joowon Kim told TechCrunch. Instead of helping marketers manage and optimize ad spend, the team refocused on generating the campaigns and creative assets themselves. Nearly a year after the initial launch, Melius revealed what it describes as an “agents lab for creative work.”

@title Melius product turn
@source TechCrunch

  Performance tool
          │
          ▼
  Did not have legs
          │
          ▼
  Codebase scrapped
          │
          ▼
  Creative work
          │
          ▼
  Funding announced

@caption Melius moved from ad spend optimization toward creative output.

Why artifacts beat knobs

The buried story in TechCrunch’s report is not the round size, it is the wedge. Melius started with the classic optimization layer: help marketers manage and optimize ad spend. That sounds rational on a pitch slide, but in practice it asks customers to trust a new tool in a budget-sensitive part of the workflow before they have felt any output. This is a pricing page of product risk where every ending is “please connect your ad account.”

The new Melius bet is simpler to evaluate. If a marketer asks for a campaign, image, or video, the output is either usable, close enough to edit, or obviously wrong. That shortens the trust loop. In AI tooling, the winning product is often not the cleverest control panel, it is the artifact that lands in someone’s actual work queue before the meeting ends.

The moat is not the model alone

TechCrunch notes that Melius is far from the only startup helping ad agencies, marketers, and brands generate creative assets. That matters because “we generate ads” is not much of a moat by itself. If the core capability is easy to describe, rivals can crowd the same checkout lane quickly. The defensible ground is likely to be workflow memory, brand fit, approval paths, and repeatable campaign production, not a generic prompt box with a nicer gradient.

This is where the Ramp lineage is interesting without being destiny. Engineers from a finance software company have seen what happens when workflow software earns trust by reducing operational drag. For Melius, the strategic question is whether “agents lab for creative work,” as described to TechCrunch, becomes a system marketers return to daily, or just a nicer vending machine for one-off assets. The former can compound. The latter gets compared feature by feature, and that is a rough sport.

What to watch next

The next logical move is not necessarily more model talk. TechCrunch’s reported traction, more than $1 million in annualized revenue within two months of the July stealth exit, gives Melius permission to prove whether customers come back for whole campaigns rather than isolated assets. Watch retention, expansion inside agencies and brands, and whether the product can move from generating creative pieces to becoming the place where creative work is planned, produced, and refined. That is the difference between a launch that demos well and a product customers build habits around.

For founders, the lesson is refreshingly practical. If your AI product asks users to believe in an optimization layer before they receive anything useful, you may be starting one step too abstract. Melius raised after admitting the first product did not have legs and rebuilding around immediate outputs. The market will decide whether the new wedge holds, but the product lesson is already useful: sell the thing customers can point to, edit, ship, and judge.