Three shutdowns: 2M users, junk signups, KYC

Three shutdown announcements in the July 5-12 window shared the same uncomfortable pattern: the public traction metric was real, but it was not the metric that could keep the company alive.

Zapper had a recognizable DeFi (decentralized finance) brand, roughly $16.5 million in total funding, and more than 2 million monthly active users before announcing that all services would stop on August 3, 2026. 1 PicSee had a $4 million seed round, a clean AI photo-sharing concept, and enough unused cash to return roughly 65% of capital to investors. 2 Gigbanc had more than 150,000 users across 30-plus countries and processed more than ₦10 billion in transactions before announcing its wind-down. 3

The useful taxonomy this week is specific. Zapper and PicSee are product mismatch cases: Zapper's dashboard usage did not become a durable revenue model, while PicSee's AI feature did not solve social-network distribution. Gigbanc is a burn-rate and regulatory-blocks case: the company said fresh capital was unavailable while Know Your Customer (KYC) and infrastructure costs made the business-to-consumer (B2C) cross-border model too expensive. 3

Zapper was a DeFi portfolio dashboard that let users track on-chain assets across wallets and chains. The company announced on July 8 that it had made the "very difficult decision" to wind down Zapper, and its official account said the website, mobile app, and API services would stop on August 3. 4 Seb Audet, Zapper's co-founder and CEO, said the team had evaluated multiple options and concluded that an orderly wind-down was the best course. 8

The company had enough history to make the shutdown matter. Zapper began after DeFiSnap and DeFiZap merged into the Zapper brand in 2020, raised a $1.5 million seed round in 2020, and raised a $15 million Series A in May 2021. 1 The product later reached more than 2 million monthly active users, supported multiple chains, and was described in the collected reporting as one of DeFi's better-known dashboard brands. 1

The missing piece was revenue. Zapper announced a planned Zapper Protocol and ZAP utility token in June 2024, with the token expected in the fourth quarter of 2024, but the token never launched before the 2026 shutdown announcement. 5 The shutdown notice did not present the token plan as part of the wind-down, which leaves the product as a dashboard that had usage, brand recognition, and infrastructure cost without a mature self-funding model. 4

Audet's shortest explanation was the most useful one: "At the end of the day, the market decides." 9 That sentence should be read narrowly. The public record does not prove one single cause. It does show a dashboard business trying to convert attention, data aggregation, and a planned token into something that could outlast venture funding.

For AI founders, the Zapper lesson is about infrastructure products that become expensive before they become indispensable. A free dashboard, free API layer, or free developer tool can create adoption charts that look healthy. The operating question is whether the heavy users would still pay after the novelty is gone, after competitors copy the workflow, and after the cost of maintaining integrations becomes a fixed burn line.

PicSee was an AI photo-sharing app from Billion Hearts Software Technologies, founded by Mayank Bidawatka and former Koo executive Sarthak Gupta. The app used AI face recognition to scan a user's gallery, identify friends in photos, and prompt reciprocal sharing after both sides approved the exchange. 10 Its privacy pitch included on-device storage, end-to-end encryption, screenshot prevention, 24-hour review windows, and photo recall. 10

The product had a clean pain point. Bidawatka wrote that people take roughly 2 trillion photos each year, but many photos never reach the people shown in them. 6 The app tried to turn that frustration into a "give to get" loop: users shared photos of others to receive photos of themselves. 11

The financing also gave the team room to test. PicSee raised about $4 million in a seed round led by Blume Ventures, with General Catalyst and Athera Venture Partners participating. 2 Bidawatka later said the company would return roughly 65% of the capital it had raised so those funds could be redeployed to other founders. 6

The failure point was distribution quality, not feature quality. Bidawatka wrote that PicSee did not spread one user at a time because it only worked when friends and families joined together: "That meant we weren't just trying to acquire users. We were trying to acquire networks." 6 Moneycontrol's reporting on the investor memo said much of the early growth came from Meta-ad-driven "junk signups" that did not convert into active users or revenue. 2

That distinction matters for AI consumer apps. AI made PicSee's matching loop possible, but AI did not remove the need for dense social adoption. A single user could download the app, scan a gallery, and still get little value if the right friends were not present. The product needed clusters, not a funnel.

The founder diagnostic is simple enough to write on a dashboard: signups are the wrong primary metric when value depends on social proximity. A founder building an AI social product should track connected groups, reciprocal actions, invite acceptance inside real-world clusters, and repeat usage after the first event. If paid acquisition brings scattered individuals who never form a graph, the growth line is noise.

Gigbanc was a Nigerian cross-border payments fintech founded in 2023 by CEO Paul Omoregie Okundaye and CTO Babatope Oni. The company offered multi-currency wallets, virtual dollar cards, foreign exchange, bill payments, and transfers to more than 200 Nigerian banks. 3 It said it served more than 150,000 users in more than 30 countries and processed more than ₦10 billion in transaction volume. 3

Those are not empty metrics. A cross-border fintech with 150,000 users has solved some real acquisition and trust problems. The problem is that payments traction can increase the cost burden at the same time it increases volume, especially when every user must pass KYC checks and every transaction depends on regulated rails.

Okundaye gave three concrete reasons for the shutdown: Gigbanc could not raise fresh funding, the KYC and infrastructure costs required for a B2C cross-border payments product were high, and the company could not finance a business-model transition. 3 Legit.ng quoted the company as saying that maintaining the platform became increasingly expensive because of compliance requirements, especially Know Your Customer regulations, alongside infrastructure expenses. 7

The wind-down plan was orderly. Customers were told to convert balances to naira and withdraw funds to local Nigerian bank accounts by July 31, 2026, with free withdrawals available during the closing period. 3 The company was also in acquisition talks with an unnamed Nigerian fintech infrastructure provider. 3

Gigbanc is the clearest burn-rate case in this group, but the burn did not come from a vague spending problem. The company described a model-level cost issue: B2C cross-border payments required expensive identity checks, infrastructure, and regulatory work before the company had the capital to absorb them. 7 BusinessDay also reported that acquisition had become a common exit path for startups that could not raise more money but still had technology or a customer base. 12

For AI founders, the transferable lesson is not limited to fintech. Any product that depends on identity verification, data compliance, human review, regulated workflows, enterprise security review, or jurisdiction-specific operations should treat those costs as part of the unit model. If cost-to-serve rises with every new user, the company has not proved scale merely by adding users.

The three cases point to three checks that founders can run before metrics become self-flattering.

First, pair every adoption metric with a funding metric. Zapper's dashboard reached scale, but the planned ZAP token never launched and the company still chose an orderly wind-down. 5 4 A founder should know which users pay, which users only consume infrastructure, and which users would leave if the product stopped being free.

Second, measure the shape of adoption, not just the amount. PicSee's founder said the app needed networks rather than individual users, and the investor memo reported that Meta-ad signups did not convert into active users or revenue. 6 2 For products with collaboration, social loops, or shared workspaces, the unit of adoption may be a team, family, friend group, or workflow cluster.

Third, calculate regulated cost-to-serve before fundraising becomes urgent. Gigbanc's users and transaction volume were real, but the CEO still pointed to fundraising failure plus KYC and infrastructure costs as reasons for the wind-down. 3 The practical question is whether the next 10,000 users make gross margin stronger or make the compliance bill arrive faster.

The shutdown memo usually appears late. The counter-metric usually moves earlier.