Back to vendors
I

Inngest

Also known as: Inngest AgentKit

Visit site
Entry priceFree $0/mo · Pro from $99/mo · Business from $499/mo · Enterprise customFull pricing detail

Durable workflow engine for AI applications that runs step-based functions on your own hosts, with model calls offloaded through step.ai, the AgentKit agent framework, agent evals and a self-hostable server under the SSPL.

Inngest is a durable workflow engine for AI applications and background jobs. Functions written with its TypeScript, Python or Go SDKs run as steps with retries, parallelism, fan-out, waits and flow control for concurrency, throttling, debounce, rate limits and priority, and start on events, webhooks or cron schedules. Inngest calls the functions on the customer's own hosts, from serverless platforms to containers, and checkpoints each step so a failed run resumes without repeating finished work.

For agents, step.ai.infer runs model calls on Inngest's infrastructure with the model chosen through its provider adapters, step.ai.wrap brings the OpenAI, Anthropic and Vercel AI SDKs into its traces, and AgentKit composes agents and networks with tools, MCP servers and routing. A running function can pause on an event for a person's approval. Agent Evals attach quality scores to runs, including LLM-as-a-judge scores, and experiments compare prompts or models. Traces, an AI Overview dashboard and Insights show every step, token and cost, and MCP servers let coding agents inspect and operate the platform.

The server is source available under the SSPL, with each release becoming Apache 2.0 on its third anniversary, and can be self hosted by following a published guide. Inngest Cloud is SOC 2 Type II compliant, with SAML, RBAC and audit trails on Enterprise. Pricing runs from a Free plan at $0 to Pro from $99 a month, Business from $499 a month and Enterprise custom, metered on executions.

Vendor details

Canonical URL

https://www.inngest.com

Category

Agent infrastructure

Company status

independent

Use cases & customers

Target customers

backend and full stack developersAI engineering teamsgrowth stage SaaS and AI startups

Deployment options

SaaSserverlesscustomer-owned computededicated infrastructure

In practice

Your agent loop runs for twenty minutes across a dozen model and tool calls and a single rate limit kills the whole request. Step level checkpointing resumes from the failure point rather than restarting the run.

A workflow needs to stop and wait for a human to approve something, possibly for days. step.waitForEvent makes that a native execution primitive instead of a state machine you build and maintain yourself.

You cannot move compute off your own infrastructure. Inngest orchestrates by calling your own HTTP endpoints on Vercel, Lambda, Cloudflare Workers or a container, so execution stays where your code already lives.

Agentic Index coverage score

10.0 / 14 capabilities · 71%

Integrations & Tool Calling Partial

Inngest runs the customer's agent tools rather than supplying them. Its agent tool loop pattern makes each model call and each tool execution a retriable step, and AgentKit gives agents an API for building tools, with MCP servers usable as tools and integrations with outside products (E2B, Browserbase, Smithery, Daytona). Inngest has no connector catalog or framework of its own for authenticated actions, so the tools are the customer's code or third party services.

Sourceinngest.com/docs/ai-patterns/agent-tool-loopsread 2026-09-25

Workflow Orchestration Full

As a workflow engine, Inngest runs functions as steps with retries, parallel steps, fan-out, loops, delays, waits and invocation of other functions, under flow control for concurrency, throttling, debounce, rate limits, batching and priority. For agents, its patterns cover agent tool loops and sub-agent delegation, and AgentKit composes networks of agents with routing.

Sourceinngest.com/docs/learn/inngest-stepsread 2026-09-25

Knowledge Grounding & RAG Not documented

There is no retrieval store that Inngest maintains over the customer's corpus. Its AI steps and AgentKit run model calls and tools, and any indexing, embeddings store or retrieval is the customer's own code or a third party service.

Sourceinngest.com/docs/features/inngest-functions/steps-workflows/step-ai-orchestrationread 2026-09-25

Human Oversight & Guardrails Partial

Durable waits let the customer's code hold for a person. In Inngest's human in the loop pattern, a function proposes an action, notifies someone, and pauses on step.waitForEvent (or waitForSignal) until an approval event arrives or a timeout passes. It is a general wait used for approval, and Inngest has no approval step, approver identity or review screen of its own.

Sourceinngest.com/docs/ai-patterns/human-in-the-loopread 2026-09-25

Security, Identity & Governance Full

SOC 2 Type II compliance is in place, with periodic independent security assessments, and HIPAA BAAs are an add on for Pro and Business and included on Enterprise. Accounts can use SAML authentication, and Enterprise adds SAML, RBAC and audit trails, with signing keys, event keys and API keys as the platform's credentials.

Sourceinngest.com/docs/learn/securityread 2026-09-25

Observability & Auditability Full

Every run is recorded for step by step inspection. Run details and traces show each step, AI calls through step.ai carry prompts, tokens and responses, the AI Overview dashboard turns that metadata into usage, cost and performance views that link to the runs behind them, and Insights queries runs and events. Trace history depends on plan, from 24 hours on Free to 90 days on Enterprise, with trace and log exports and Datadog and Prometheus integrations.

Sourceinngest.com/docs/platform/monitor/ai-overviewread 2026-09-25

Memory & State Persistence Partial

Every step is checkpointed, so an agent's progress is a durable record and a failed run resumes without repeating finished steps, which is state within one run that the code carries on from. Across runs, AgentKit's History Adapter keeps conversations per thread so a later run resumes in context. That store is the customer's own database behind an adapter the customer writes, with no published lifetime, purge path or first party screen to review or edit it, and AgentKit's long term memory runs on Mem0, a third party.

Sourceinngest.com/docs/learn/durable-agentsread 2026-09-25

Deployment & Data Residency Full

Inngest runs in the customer's own environment in two ways. Functions always run on the customer's own hosts (serverless platforms, containers or servers) and Inngest calls them, and the Inngest server can be self hosted under a guide published since its 1.0 release. The server is source available under the SSPL, and each release becomes Apache 2.0 on its third anniversary, as the repository's license file states.

Sourceinngest.com/docs/self-hostingread 2026-09-25

Prebuilt Agents, Templates & Packs Partial

Starting points are published, including AgentKit examples, a starter for a conversational app with persistent history, AI patterns in the docs (agent tool loops, human-in-the-loop, sub-agent delegation), and agent skills and plugins for coding agents. These are examples a developer builds from, not a catalog of agents to adopt.

Sourceagentkit.inngest.com/examples/overviewread 2026-09-25

Triggers & Channel Coverage Full

Functions start on their own. They trigger on events, including webhook events Inngest consumes, on cron schedules and on several triggers at once, and a running function can wait for an event or signal before carrying on.

Sourceinngest.com/docs/guides/scheduled-functionsread 2026-09-25

Model Flexibility & Routing Full

Customers choose the model through Inngest's own tools. The step.ai.infer call offloads a model call to Inngest's infrastructure, with the model named through Inngest's provider adapters (for example step.ai.models.openai with a chosen model), step.ai.wrap wraps the OpenAI, Anthropic and Vercel AI SDKs as steps, and AgentKit takes a model per agent. All of them track requests and responses in traces.

Sourceinngest.com/docs/features/inngest-functions/steps-workflows/step-ai-orchestrationread 2026-09-25

APIs, SDKs & MCP Extensibility Full

Inngest is callable from outside through SDKs in TypeScript, Python and Go, a REST API, and the CLI, and it provides MCP servers for Inngest Cloud and the local Dev Server whose tools match the REST API and CLI commands, so a coding agent can inspect and operate apps, functions, events and runs.

Sourceinngest.com/docs/ai-dev-tools/mcpread 2026-09-25

Testing, Debugging & Optimization Full

Agent Evals attach numeric or boolean quality scores to runs and steps, including LLM as a judge scores and deferred scoring, group related work into sessions, and compare prompts, models or strategies with group.experiment() A/B variants. Scoring is labeled beta, a public beta customers can use today. Scores can be queried in Insights and are metered by plan, from 10,000 on Free to 250,000 on Business.

Sourceinngest.com/docs/learn/agent-evalsread 2026-09-25

Browser & Computer Use Not documented

There is no browser or desktop for an agent to operate, and Inngest has no browser environment of its own. AgentKit integrates with Browserbase, but the browser and its control belong to Browserbase, a third party, and Inngest runs the customer's function around it.

Sourceagentkit.inngest.com/integrations/browserbaseread 2026-09-25

The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded

Product updates

·Human approval / guardrailsVerified

Version 4.26.2 of the Inngest TypeScript SDK adds experimental command shims for Sandboxes. A shim makes a named program inside the sandbox call a handler in the customer's own app first, and that handler can do the privileged work itself, run the real program, or refuse it. Credentials never have to enter the sandbox, and commands that use shims can now run for longer than five minutes.

Bears on: Security / enterprise

View source
·Agent capabilityVerified

Sandboxes are now in public beta on Inngest for Pro and higher accounts using the TypeScript SDK. step.sandbox starts an isolated microVM from inside a function in under a second and runs commands in it, with create, command and destroy recorded as durable steps on the run trace. Sandboxes come in 1, 2 or 4 vCPU sizes, are billed per second of CPU and RAM while running, and get one hour of active runtime each, with paused time not counted.

Bears on: Workflow orchestration

View source
·Workflow orchestrationVerified

Inngest released a feature allowing developers to rerun workflows from any specific step, even if the run has successfully completed. The platform reconstructs all preceding steps as memoized state and only re-executes the selected step and subsequent operations.

Bears on: Memory / state

View source
View all 3 Inngest updates →Tracked since Aug 2026 · Verified from public vendor sources

Pricing

Free $0/mo · Pro from $99/mo · Business from $499/mo · Enterprise custom

Plan fee with included executions (runs times steps plus one), events, span data and scores, with add-ons and overages above them; functions run on the customer's own hosting, billed separately

Free tier

What is public

All four plans are published at inngest.com/pricing with their limits, starting prices and several overage rates, alongside a calculator. Enterprise is sold through sales.

Billing mechanics

Four plans: Free, Pro from $99 a month, Business from $499 a month and Enterprise custom, each with included executions, events, span data and evaluation scores. Executions are runs times (steps plus one). Plan changes take effect immediately and prorate. Because Inngest calls functions on the customer's own hosts, compute sits on the customer's hosting bill, not the subscription.

Cost watchouts

Executions count every step: a run with ten steps is eleven executions, so agent loops use the allowance far faster than background jobs. The functions themselves run on the customer's own hosting, which is a second bill. SAML, RBAC and audit trails are Enterprise only.

Variable cost rationale

Inngest meters executions, events, span data and evaluation scores, and each plan includes an allowance of each. An execution is a run times its steps plus one, so agent loops and fan out workflows use the allowance far faster than simple background jobs and push usage into execution add-ons. Events above Pro's allowance cost $0.50 per million and scores $1.50 per thousand. The plan fee sets the floor, and the included allowances absorb light use. No spending cap on add-ons or overage is published. Compute is a separate bill on top, since the functions run on the customer's own hosting.

Additional watchouts

The pricing FAQ calls Inngest open source, but the server's license is the SSPL, with each release becoming Apache 2.0 on its third anniversary. Governance features (SAML, RBAC, audit trails) need Enterprise.

Overage / add-ons

Executions and workers above the plan are bought as add-ons. Events above the allowance are $0.50 per million and evaluation scores $1.50 per thousand on Pro. Span data above the allowance is billed as overage.

Sales call required

Mixed (some tiers require a call)

Free / trial

Free plan at $0 a month with 50,000 executions, no credit card required. The server can also be self hosted.

Lowest paid plan

Pro from $99 a month

Commercial notes

Serverless deployments can offload model calls to Inngest through step.ai.infer, so the customer's function is not billed for time spent waiting on the provider. The pricing page carries a calculator for runs, steps and scores.

Key ambiguities

Inngest does not publish execution add-on rates above the included amounts. Events above Pro's allowance are $0.50 per million and scores $1.50 per thousand.

Cancellation / refund

Plan changes take effect immediately and are prorated.

Support SLA / resale

Dedicated support comes with Enterprise.

Missing data

Execution and worker add-on rates, and Enterprise pricing.

Agentic Index verified 2026-09-25

Alternatives to Inngest

The closest documented capability profiles to Inngest among agent infrastructure platforms tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.

  • Bernstein10.5 / 14Fuller documented coverage on Integrations & Tool Calling and Human Oversight & Guardrails
  • Confident AI8.5 / 14A lighter documented profile than Inngest
  • DBOS8.5 / 14A lighter documented profile than Inngest
  • Hatchet8.5 / 14A lighter documented profile than Inngest
  • HoneyHive8.5 / 14A lighter documented profile than Inngest
  • Langfuse8.5 / 14A lighter documented profile than Inngest

Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded

Contact us

Found a vendor we missed? Have feedback on the index? We'd love to hear from you.