Ingest raises $21M Series A led by Altimeter to build durable execution infrastructure for AI agents
Sep 16, 2025 with Tony Holdstock
Key Points
- Ingest raises $21M Series A led by Altimeter Capital to build durable execution infrastructure for AI agents, abstracting away distributed-systems complexity that slows product iteration.
- SoundCloud deployed Ingest workflows in six weeks after nine months of failed hand-built infrastructure, shipping a feature that lets artists create vinyl records from uploaded music.
- Ingest positions itself below the application layer, arguing that reliable workflow execution underneath AI agents matters more than the agents themselves as commodity models commoditize.
Summary
Ingest closed a $21M Series A led by Altimeter Capital, with follow-on participation from Andreessen Horowitz and Notable Capital. Jamin Ball at Altimeter anchored the round.
What Ingest builds
Ingest is an async execution platform that lets engineers define workflows as step functions. Developers write the logic and deploy it; the platform handles queues, workers, and infrastructure configuration. When every team has access to the same models and compute, iteration speed becomes the differentiator. Ingest abstracts away the infrastructure layer that slows teams down.
Tony Holdstock, co-founder, traces the product to his time running engineering at a healthcare company. Patient-triggered workflows that should take days routinely took months. He built Ingest to solve that friction for full-stack product engineers who don't want to manage distributed-systems infrastructure.
Agent infrastructure, not agent products
Ingest sits deliberately below the application layer. The execution layer—the runtime that runs AI workflows reliably—is harder to get right than it appears and is frequently underbuilt. Teams building agents on top of commodity models still need durable, observable workflow execution underneath. That is the gap Ingest fills.
Holdstock is most bullish on tools people use daily, such as coding assistants and internal support, and on companies embedding custom agent experiences inside their own products. Decagon's customer support work shows what becomes standard.
SoundCloud case study
SoundCloud is a named customer. The company's CTO spent roughly 9 months trying to hand-build workflow infrastructure for music operations including content moderation, piracy detection, and distribution to Spotify and iTunes. Within one week of adopting Ingest, significant workflows were running. SoundCloud reached production within approximately six weeks. The most recent feature shipped on the platform lets artists create physical vinyl records from music they upload.