Ramp’s $22.5B fintech Ramp has just acqui‑hired the entire team of Jolt AI, a three-person startup. Crunchbase News reports that Ramp made the acquisition with "the intent of making its engineers ‘as productive as possible’”. In Ramp’s telling, Jolt’s focus on developer productivity perfectly fits its strategy: Ramp CTO Karim Atiyeh says their job is “making engineers radically more productive, helping them ship faster”.
Jolt founder Yev Spektor echoes this mission, saying “We’re obsessed with making engineers faster and more effective. Ramp shares that obsession,” he wrote. In other words, Jolt’s entire independent product has been folded into a big company’s engineering team. This small deal is actually the tip of a trend: big companies are snatching up tiny AI code teams or forcing them out of business.
Small AI coding startups face harsh realities
Many other independent AI coding startups have already stumbled or surrendered. For example:
- Alex: A Y Combinator‑backed Xcode AI assistant. In Sept 2025 its three-person team joined OpenAI’s Codex division, effectively ending the startup. Founder Daniel Edrisian noted their team had “built the best coding agent for iOS & MacOS apps” before winding down. (Coincidentally, Apple recently added ChatGPT integration to Xcode, erasing Alex’s niche.)
- CodeParrot: A startup turning Figma designs into React/Flutter code. Backed by Y Combinator, it shut down in mid-2025. Cofounder Vedant Agarwala admitted on LinkedIn that they “burned through the $500k we’d raised, and when we hit $1,500 MRR with our final pivot … we couldn’t break through”. In short, even after multiple pivots and demo days, they couldn’t find a sustainable business.
- Subtl.ai : An AI “developer sidekick” platform. Its founder announced they had “started shutting down Subtl.ai” in 2025, noting that despite strong technology, funding ran out and market fit proved elusive. Subtl’s closure followed CodeParrot’s by days, highlighting a wave of early AI startups collapsing.
- Kite: One of the earliest AI coding assistants (launched 2016). It shut down in 2022 after running out of money. Founder Adam Smith frankly blamed timing: Kite was “10+ years too early to market,” and “our product did not monetize”. This cautionary tale showed that even with tens of millions in VC funding, finding a paying market for code AI was extremely hard.
Each of these examples shows the same problem: building a practical, profitable AI coding tool is brutally difficult. As Kite’s founder noted, even Microsoft’s Copilot (with its huge budget) “still has a long way to go,” it might take “over $100 million to build a production‑quality tool”. And Reuters reports that none of these code-gen startups are yet profitable: many face rising compute costs for LLM queries and stiff competition from tech giants. In short, independent teams find it nearly impossible to deliver the quality and ROI that big corporations demand.
Banking
