Depot

40x faster Docker builds and 10x faster GitHub Actions — remote builders, persistent NVMe cache, and Depot CI that AI agents can trigger via API.

Depot Review: The Build Acceleration Platform That Rebuilt CI for the AI-Driven Development Era

CI pipelines were designed for 2015 workflows — when writing code was the bottleneck. In 2026, AI coding assistants generate code at unprecedented velocity, and the bottleneck has shifted: code integration and build validation are now the constraint. An hour-and-45-minute build cycle that was acceptable when a developer submitted one PR per day is untenable when an AI agent is submitting dozens. Depot was built to remove this constraint — remote Docker builders with persistent NVMe-backed cache that deliver 40x faster container builds, GitHub Actions runners that run 30% faster than GitHub-hosted at roughly half the cost, and Depot CI — a fully programmable, API-first CI engine built specifically for the AI agent integration era, where agents trigger jobs, monitor status, pull logs, and compose workflows without relying on traditional UI-based CI tools.

Quick Summary

Depot is a build acceleration platform founded in 2022 and backed by Y Combinator, Felicis, and Pioneer Fund — providing remote Docker container builds with persistent NVMe-backed caching, GitHub Actions runners with 30% faster CPUs at roughly half the GitHub-hosted cost, and Depot CI — a programmable API-first CI engine for AI-accelerated development — with customers reporting 2x to 55x build time reductions and 50% plus CI cost reductions. Plans from $20/month (Developer) and $200/month (Startup) with per-second metered usage.

Is it worth using? Yes for development teams whose CI build times are creating meaningful developer friction — particularly teams producing Docker images, running multi-architecture builds, or using AI coding assistants that generate code faster than traditional CI can validate it.
Who should use it? DevOps engineers, platform engineers, and software development teams who run Docker-based CI pipelines and want dramatically faster build times with a persistent cache layer, and AI-first development teams building agentic workflows that need programmatic CI triggering.
Who should avoid it? Teams with very low build frequency where the monthly plan cost exceeds the developer time savings from faster builds, and teams running primarily non-Docker build workflows where Depot’s core Docker acceleration advantage is less applicable.

Verdict Summary

Best for

Engineering teams with slow Docker builds in CI — Depot’s persistent NVMe-backed layer cache and emulation-free native ARM/Intel builders eliminate the cold cache rebuilds that make Docker CI slow, delivering the 40x build time improvements that customers like PostHog report
Teams running GitHub Actions who want faster, cheaper runners — Depot’s managed GitHub Actions runners provide 30% faster CPUs and 10x faster cache at approximately half the cost of GitHub-hosted Linux runners, without changing any GitHub Actions YAML
AI-first development teams and platform engineers building agentic workflows — Depot CI’s fully API-driven, programmable CI engine allows AI agents to trigger builds against uncommitted changes, compose workflows dynamically, and pull logs programmatically without UI-based CI tooling

Not for

Teams with infrequent or very low build volumes where the $20/month Developer plan’s 500 included Docker build minutes cover all needs without meaningful savings over standard CI
Organisations that need a full CI platform replacing their existing CI infrastructure — Depot is a build acceleration and programmable CI engine that complements rather than replaces the full CI/CD pipeline
Teams using primarily non-Docker build systems where Depot’s layer cache and remote builder architecture provides less applicable acceleration

Rating
⭐⭐⭐⭐ 4.2 / 5

What Is Depot?

Depot is a build acceleration company founded in 2022 — backed by Y Combinator, Felicis Ventures, and Pioneer Fund — that started with a focused problem: Docker builds in CI are slow because every run starts cold and rebuilds layers that barely changed. Depot’s solution was remote builders with persistent NVMe-backed caching that keeps the layer cache warm across runs, combined with native Intel and ARM builders that eliminate the emulation overhead that slows multi-architecture builds.

In early 2026, Depot launched Depot CI — a programmable CI engine built specifically for the AI-accelerated software development era. Where traditional CI systems wait for GitHub events and expose configuration through YAML files, Depot CI is fully API-driven: agents trigger jobs, compose workflows dynamically, check status, and pull logs without relying on UI-based tooling. Jobs start in seconds rather than minutes, support custom runner images pre-loaded with dependencies, and include SSH access for debugging — making it the first CI engine designed from the ground up for AI agent integration.

How Depot Works

Replace docker build with depot build. Swap the standard docker build command for depot build in any existing CI pipeline — no Dockerfile changes required. Depot routes the build to remote infrastructure with persistent cache automatically.
Persistent NVMe-backed layer cache. Every build layer that has not changed is served from Depot’s persistent NVMe cache rather than being rebuilt — the same Dockerfile produces dramatically faster builds on every run after the first, regardless of which CI runner picks up the job.
Native multi-architecture builds. Depot provides native Intel and ARM builders — no emulation required for multi-architecture image production. Multi-arch builds that take 10x longer with emulation complete at native speed.
Drop-in GitHub Actions runners. Replace GitHub-hosted runners in any GitHub Actions workflow by changing the runs-on label to use Depot’s managed runners — 30% faster CPUs, 10x faster cache, available in Linux, ARM, and macOS configurations.
Trigger Depot CI via API. For AI agent integration, Depot CI accepts job triggers through its fully API-driven interface — agents submit jobs programmatically, monitor status, and pull logs without requiring GitHub events or YAML configuration changes.
Debug with built-in SSH access. Depot CI includes SSH access to running jobs for debugging — engineers and agents can inspect the job environment in real time rather than inferring state from logs alone.

Key Features

Remote Docker builders with persistent NVMe-backed layer cache — up to 40x faster container builds
Native Intel and ARM builders — no emulation for multi-architecture image production
Managed GitHub Actions runners — 30% faster CPUs at approximately half the GitHub-hosted cost
Depot CI — fully programmable, API-first CI engine for AI agent integration, launched March 2026
Job start in seconds — under-second job initiation versus minute-scale starts on traditional CI
Custom runner images pre-loaded with dependencies for faster job startup
SSH access to running jobs for real-time debugging
Depot Registry — container image registry included on Startup and above
GPU-enabled builds on Business tier for ML model building and AI workloads
Docker build autoscaling on Business tier
macOS M2 runners available on Startup and above
Depot Cloud runs on AWS — available in us-east-1 and eu-central-1
Customers: PostHog, Suno, and Y Combinator alumni portfolio
Build time reductions of 2x to 55x reported by customers
CI cost reductions of 50% plus reported by customers

Real-World Use Cases

PostHog Docker build acceleration: PostHog, known for shipping weirdly fast, previously waited 1 hour and 45 minutes for Docker builds. After Depot, builds complete dramatically faster — the team credits the persistent cache and remote builder as essential to maintaining their deployment velocity at scale.
Multi-architecture image production: A platform engineering team building Docker images for AMD64 and ARM64 was running emulated ARM builds on Intel CI runners — each build taking 3x longer than native. Depot’s native ARM builders complete multi-arch builds at native speed without any Dockerfile changes.
AI agent CI integration: An engineering team builds an AI coding agent that generates PRs and validates them through automated tests. Depot CI’s API allows the agent to trigger builds against uncommitted changes, poll for completion, and pull logs — the agent gets a full feedback loop without any CI configuration changes or GitHub webhook requirements.
GitHub Actions cost reduction: A startup running 50,000 GitHub Actions minutes per month on GitHub-hosted Linux runners switches to Depot’s managed runners by changing runs-on in the YAML. CPU performance improves 30%, cache performance improves 10x, and the monthly bill reduces by approximately 50% — all without changing any other workflow configuration.

Pros and Cons

Pros	Cons
40x Docker build acceleration with persistent NVMe cache — the most impactful CI improvement for Docker-heavy pipelines without any Dockerfile changes	Build-focused scope — Depot accelerates builds and adds programmable CI but does not replace a full CI platform managing complex multi-stage deployment pipelines
Depot CI API-first design for AI agent integration — the only CI engine built from the ground up for programmatic agent triggering rather than UI-based workflow management	Usage-based pricing scales with build volume — teams with very high build volumes should model costs carefully before committing to the Startup plan’s included minute allocation
GitHub Actions runners at approximately half the GitHub-hosted cost with 30% faster CPUs — immediate savings without changing any workflow YAML	No perpetual free tier — the Developer plan requires $20/month after the initial trial period for production use
Y Combinator, Felicis, and Pioneer Fund backing — strong financial position for continued infrastructure investment	Business plan GPU-enabled builds and dedicated infrastructure require custom pricing through a sales conversation
50% plus CI cost reductions and 2x to 55x build time improvements reported by customers — validated ROI across portfolio	Available only in us-east-1 and eu-central-1 AWS regions — teams with strict data residency requirements in other regions should verify availability

Pricing & Plans

Developer — $20/month

1 user
500 Docker build minutes included
2,000 GitHub Actions minutes included
2,000 Depot CI minutes included
25 GB cache storage
Community Discord support
Additional usage billed at metered rates

Startup — $200/month

Unlimited users
5,000 Docker build minutes included
20,000 GitHub Actions minutes included
20,000 Depot CI minutes included
250 GB cache storage
Email support
User RBAC
macOS M2 runners
Depot Registry
Optional SSO add-on available

Business — Custom pricing

Dedicated AWS infrastructure
GPU-enabled builds
Docker build autoscaling
Static IPs
AWS PrivateLink
SSO/SAML/SCIM
Audit logging
Slack Connect support
Advanced compliance controls
AWS Marketplace billing

Metered usage rates (above included minutes):

Depot CI: $0.00005/second/vCPU
GitHub Actions runners: $0.004/minute (2-vCPU Linux)
Docker builds: $0.04/minute
Cache storage: $0.20/GB/month

Best Alternatives & Comparisons

Trunk AI — Better for flaky test detection, CI failure analysis, and merge queue management alongside build acceleration
GitHub Actions — Better for the broadest CI/CD ecosystem integration — Depot adds acceleration on top rather than replacing it
Buildkite — Better for self-hosted CI with full pipeline control for teams needing maximum infrastructure flexibility
CircleCI — Better for teams wanting a fully managed CI/CD platform rather than a build acceleration layer

Frequently Asked Questions (FAQ)

What is Depot?

Depot is a build acceleration platform providing remote Docker builders with persistent NVMe-backed cache (40x faster builds), managed GitHub Actions runners (30% faster, half the cost), and Depot CI — a programmable API-first CI engine for AI agent integration. Backed by Y Combinator, Felicis, and Pioneer Fund.

How much does Depot cost?

Depot plans start at $20/month (Developer, 1 user) and $200/month (Startup, unlimited users), with usage above included minutes billed at metered rates. Business is custom pricing. No perpetual free tier.

What is Depot CI?

Depot CI, launched in March 2026, is a fully programmable API-first CI engine built for AI-accelerated development. AI agents can trigger jobs programmatically, compose workflows dynamically, check status, and pull logs without requiring GitHub events or YAML configuration — enabling complete agent feedback loops for code validation.

Does Depot replace GitHub Actions?

No — Depot complements GitHub Actions by providing faster, cheaper managed runners and accelerated Docker builds that plug into existing pipelines. Depot CI provides an additional programmable CI engine for workflows that require API-driven triggering beyond what GitHub Actions YAML supports.

What kind of build speed improvements can teams expect?

Customer-reported improvements range from 2x to 55x faster build times. PostHog reduced a 1 hour 45 minute build to dramatically less through Depot’s persistent cache. GitHub Actions runners deliver 30% faster CPU performance. The actual improvement depends on how cache-friendly the existing Dockerfile is and how much of the build was being rebuilt unnecessarily.

Is Depot available in all regions?

Depot Cloud runs on AWS and is available in us-east-1 (US East) and eu-central-1 (EU Central). Teams with strict data residency requirements in other regions should verify current availability with Depot before committing.

Final Recommendation

Depot is the most practical and immediately impactful build acceleration investment for engineering teams whose Docker CI pipelines are creating developer friction — and the most forward-looking CI platform for teams building AI-agent-driven development workflows in 2026. The persistent NVMe-backed cache, native multi-arch builders, and GitHub Actions runner cost reduction deliver measurable ROI from the first week for any team running Docker-heavy CI. Depot CI’s API-first programmable engine is the infrastructure layer that AI agent development requires — and building it in 2026 positions Depot as the default CI choice for the development era where AI agents generate more code than human developers. For any engineering team asking why their builds are still slow in 2026, Depot is the answer.

Next steps