App Store Sales Dip, AI Agent Workflows Diversify, and SMB M&A Activity Picks Up

Key Takeaways

1. App Store sales fell for the first time in a decade, but non-game apps still grew 14.6% and web revenue share rose, complicating the 'death of apps' narrative. — via 1 2
2. Prompt injection is a real blocker for agentic workflows: levelsio keeps bug boards disconnected from AI, while Alex Lieberman argues local dev is dead and pushes isolated cloud environments. — via 1 2
3. Scaling agents is becoming a multi-account operation: a workflow using 14 Claude Code accounts and 150+ agents on a large enterprise codebase surfaced as an example of untapped potential. — via 1
4. SMB M&A listings show strong recurring-revenue multiples: new acquire.com listings range from a $316K-revenue AI newsletter to a $1.4M Clover app. — via 1 2
5. Global remote hiring continues to be a major cost lever, with one founder citing a 400-person, office-less company with only 20 American employees. — via 1

1. AI Agents and Developer Workflows

1. levelsio said he won't connect his bug feedback board directly to AI because attackers could exploit prompt injection to add a backdoor disguised as a normal fix. He currently copies bug reports and feature requests into Claude Code manually, making security a primary constraint in agentic product design. — via 1
2. Alex Lieberman declared local development basically dead: multi-agent systems can't parallelize on a real stack, overwrite each other, and create too much mental overhead. His proposed fix is to give each agent its own isolated cloud environment (like @AmpCode orbs), which produces smaller, cleaner diffs and reduces conflicts. — via 1
3. Tony Dinh pointed to a workflow that runs 150+ agents across 14 Claude Code accounts to handle a large enterprise codebase. He admitted it made him realize his own AI-agent usage is still far from that level, highlighting the gap between current practice and advanced orchestration. — via 1
4. Ben Tossell shared a benchmark where Factory Droid finished a test in 8 minutes with 20 tool calls at $1.60, while Claude Code took 19 minutes, 40 calls, and $1.87. The comparison is informal, but it underscores how quickly AI agent performance and pricing are being contested. — via 1

2. Market Data and Performance Signals

1. levelsio cited App Store data showing sales declined for the first time in a decade, despite more apps than ever. He added that non-game apps grew 14.6% globally YoY, gaming fell 4.5%, and the web revenue share on RC rose from 20% to 26%, arguing this does not mean the end of apps. — via 1 2
2. Arvid Kahl flagged DragonflyDB claims of 25x Redis throughput with 80% lower resource consumption, saying he was surprised and wondered where the catch is. This highlights the growing competition in high-performance data infrastructure, though the numbers are presented as an observation rather than verified benchmark. — via 1
3. Shaan Puri summarized Poppi's growth from a podcast episode: revenue rose from $3M to $500M in roughly 4.5 years, a rebrand took 9 months, and a last-minute Super Bowl ad tripled brand awareness overnight. He also noted the Coke contract blocked Poppi from selling at its own Lakers official soda arena, making the brand's rise a lesson in constraints and awkwardness as an underrated advantage. — via 1

3. Business, M&A, and Global Hiring

1. acquire.com listed several small businesses for sale, including an AI news and intelligence newsletter with $316.4K TTM revenue and a $396K asking price; a class-action settlement app at $1.1M TTM and $3.8M asking; a Muslim prayer/Quran app at $202.2K TTM and $300K asking; and a Clover merchant cash-discount software at $1.4M TTM and $10.2M asking. These listings show the wide range of cash-flowing businesses moving through the SMB M&A market. — via 1 2 3 4
2. Nick Huber cited Somewhere.com as a global remote case study: 400+ employees, no office, and only 20 Americans on staff. He added that he runs a COO in Cape Town, management teams in Johannesburg and Durban, and an engineering department in Bogotá, demonstrating the savings and leverage of distributed hiring. — via 1
3. Starter Story featured Rashid, who makes $32K/month from three seemingly boring database websites. His biggest idea came by accident when he built a product to test whether other founders shared his own financing pain, a reminder that customer discovery can start with personal problems. — via 1
4. Sam Parr observed a near-universal pattern: promoting the best salespeople into sales leadership is almost always a mistake because management and selling demand different skills. He also noted top sellers often make moves that look unreasonable to non-sellers but work anyway. — via 1