Describe what you want done.

Actuant builds and runs it.

An automation platform where AI agents are first-class, not bolted onto a flow engine. Author in plain language, refine on the canvas, and run on the model you choose.

prompt“Every morning, triage inbound support tickets.”

Start building free Explore the product

Free tier · 250 runs/mo · no card required

Automate across the tools your work already runs on

The product

A workflow is a graph of steps with a trigger.

Steps fall into four families. The AI steps are where Actuant departs from a conventional flow engine.

Triggers

Start a run: webhook, schedule, polling, or manual.

Actions

Do work in the world: HTTP requests and app connectors.

Logic

Control flow: branch, loop, filter, merge.

AI steps

A simple LLM node for one-shot calls, plus a first-class Agent node that plans, selects tools, loops, and reports its reasoning.

In practice

From a sentence to a running workflow.

A few of the workflows teams run on Actuant. Each one is a trigger, a handful of steps, and an agent making the calls a static rule can’t.

Customer support5 steps

Support ticket triage

Classify, prioritize, and draft replies the moment a ticket lands.

See how it works

Sales & growth5 steps

Lead enrichment & routing

Turn a raw form fill into a scored, enriched, routed opportunity.

See how it works

Research & monitoring5 steps

Research agents over new sources

Watch the sources that matter and file what's relevant, summarized.

See how it works

Data & operations5 steps

Database to spreadsheet sync

Keep the spreadsheet the business lives in current, automatically.

See how it works

Finance & operations6 steps

Invoice processing & approval

Read every invoice, match it to the PO, and route only the exceptions.

See how it works

Engineering & on-call6 steps

Incident triage & on-call summary

Turn a noisy alert into a triaged incident with a written first assessment.

See how it works

The difference

Most tools graft AI onto a flow engine. Actuant is built around it.

Conventional automation tools win technical users with flexibility, but their authoring is entirely manual. Actuant competes on being AI-native, on authoring in plain language, and on running every step on the model you choose.

Agents as a primitive

An Agent node autonomously decides which steps and connectors to use to reach a goal, rather than every branch being wired by hand.

Natural-language authoring

Describe the outcome in plain language and Actuant drafts the workflow on the canvas for you to refine. The speed and accessibility wedge.

Any model, your key

Run each node on the model you choose and bring your own key (BYOK), with a provided model available instantly for anyone starting out.

Agent-grade observability

Every run shows step inputs and outputs alongside the agent's tool calls and reasoning, so debugging an agent is as clear as debugging code.

Pricing

Generous run caps. Software-like margins.

On paid tiers, heavy inference runs on your chosen model, so the big run allowances cost almost nothing to offer. Soft caps with per-1,000-run overage, no margin traps.