How a Gamified News App Is Trying to Make Facts Compete With TikTok

Adam Harder has had a unique career in the media industry. He grew up in Ohio, joined the Air Force to pay for college, and spent six years as a military broadcast journalist stationed in Belgium, South Korea, and Saipan (Mariana Islands). He then moved into tech marketing at Cisco and DigitalOcean before founding InPress, a news aggregator app built around gamification and social matching. The premise: Make consuming factual information more engaging than doomscrolling. His shorthand for it - "make the veggies taste like chocolate" - has become the app's organizing philosophy.

InPress is not original reporting. It aggregates content from free-to-access outlets via OPoint, a news aggregation service that licenses content from thousands of free-to-access outlets worldwide - but not from major paywalled publishers like the New York Times or Wall Street Journal. 

Inpress surfaces AI-generated two-line summaries ("gists"), and layers on Duolingo-style streaks, leaderboards, and a matching algorithm that connects users based on how they think and feel about the news they consume. 

Users rate each article on three dimensions - emotion, interest, and importance - and those ratings feed a vectorization model built on Gemini and OpenAI embeddings that maps each user's evolving worldview in high-dimensional space. Harders explains this mechanism with a vivid planet-and-skyscrapers analogy: Every article rating moves your planet through infinite space, you end up closest to those who share your most active interests.

InPress has been live for 16 months - first in Washington DC, then New York, then nationwide since last April. Total spending to date: $220,000. Total downloads: 13,000. Harder is disarmingly candid about what that means. "In the grand scheme of tech companies, that is literally nothing," he says. "We haven't been spending on massive user acquisition."

What he does lean on: engagement metrics that are harder to fake than download counts. Users spend 8.5 minutes per session on InPress. On average, they read and rate 12.6 articles a day. Day-30 retention sits at 43%. Across all users, 316,000 article surveys have been submitted. The average user age is 31, with the densest cohort between 23 and 31. "There's no reason for a 22-year-old in today's environment to be reading and rating 12.6 written pieces of mobile journalism a day," Harder says. "That speaks to the thesis."

He's less forthcoming about daily-to-monthly active user ratios - the metric Duolingo uses as its benchmark of stickiness (37 percent). Harder says his numbers are "in the thousands" but declines to give a percentage, noting that the nationwide launch only happened in April 2025. InPress 2.0, a substantially rebuilt version of the product, launches in April 2026.

Three revenue streams are in development, none yet active at scale. First: a B2B behavioral data platform selling anonymized, non-inferential audience insight to publishers and brands. Letters of intent from interested companies are in place; the platform is still being built. The pitch is that InPress data captures how users actually think and feel about factual information - not algorithmic proxies skewed by engagement optimization. "We can tell you how any demographic thinks, feels, and cares about anything happening in the news," Harder says. "And it's not softened by biased coverage."

Second: a ClassPass-style premium content model. Users pay a monthly fee for credits to unlock paywalled articles from partner outlets, while keeping full access to gamification and social features. (ClassPass is a US-based subscription service that gives members monthly credits to book classes across multiple gyms and studios - without committing to a single venue.) This requires publisher partnerships InPress doesn't publicly have yet (though Harder confirms at least one is under NDA and expected to go public in roughly two months).

Third: ad placement agreements with media partners, already signed for "several hundred thousand dollars in committed value" - but deliberately not yet activated. Harder is holding them back until the product is ready to convert traffic to sustained retention rather than a one-time spike. The current $2 million pre-seed funding raise, backed by Flybridge Capital, is primarily earmarked for user acquisition once InPress 2.0 is live.

InPress has no New York Times or Wall Street Journal - major outlets that don't work with OPoint and wouldn't partner with a 13,000-user app anyway. Harder frames this as a startup sequencing problem rather than a structural flaw: Prove the thesis first, earn the partnerships later. He says paywalled outlets have already approached InPress wanting to be included. "We've basically told them to hold on."

The AI content question requires careful framing. InPress markets itself as free of AI slop, and it is - in the sense that no AI-generated articles appear in the feed. But the gists are AI-generated, built on prompts designed to summarize without giving the story away. Harder says editorial staff review edge cases, and that the format - one line of summary - makes errors rare. He also requires partner outlets to commit to no AI-written content as a condition of inclusion.

Harder's one-year outlook: 

Harder has no exit plans, but he names social platforms, digital media companies, and advertising and data firms as logical eventual acquirers - but only, he says, under conditions that preserve neutrality and non-manipulative distribution.

As Harder puts it, the core question driving the whole thing is: "How do we make factual information competitive in the attention economy?" It's not a new question. What InPress is betting is that the answer isn't better journalism but better game design.