Agent operations, engineered

Run AI agents with surgical precision.

Neurokit turns prompts, tools and models into dependable pipelines. Compose a crew, wire it to the services you already use, and ship it to production — every run metered to the token.

No subscriptionsPay per token330+ ready crews
crew-builderrunning
run #3128− 412
Anthropic & OpenAI models120+ service connectorsAppend-only token ledgerAPI-first runsEU-hosted billing
The toolkit

Everything between an idea and a running crew

Six instruments, one bench. Each is small enough to learn in a minute and sharp enough to use every day.

spec 01 · calibrated ✓

Visual crew composer

Drag agents and tasks onto a canvas, connect the edges, and the execution graph is your documentation.

spec 02 · calibrated ✓

120+ connectors

Slack, Telegram, RSS, webhooks, HTTP — credentials stay encrypted, agents get scoped access.

spec 03 · calibrated ✓

Bring-your-own models

Anthropic, OpenAI, Gemini, xAI and more. Route each agent to the model that fits its job.

spec 04 · calibrated ✓

Token-exact billing

An append-only ledger records every debit with the balance after it. Audit any run down to the step.

spec 05 · calibrated ✓

Runs as an API

Trigger any crew with a single authenticated POST. Same engine, same ledger, same guarantees.

spec 06 · calibrated ✓

Fails safe

Reserves are taken up front and released on failure — only successful steps are ever charged.

See the instrument work

A two-minute tour of a crew in production

From composing agents to the metered ledger line — this is the whole loop, exactly as it runs.

T
Triage

Scan overnight advisories

claude-sonnet-5
A
Assessor

Measure exposure in our stack

gpt-5.2
V
Verifier

Confirm findings, sign the brief

claude-sonnet-5
Procedure

Four steps from blank canvas to running crew

1

Compose

Define agents with roles, goals and models — or start from one of 330+ templates.

2

Wire

Connect tasks into a graph. Add triggers, branching and the services each step operates.

3

Run

Execute on demand, on schedule or over the API. Watch each step stream its output.

4

Meter

Every token spent lands in the ledger with the balance after it. Top up when you need to.

nk / telemetry live
production-ready crew templates0+
services your agents can operate0+
integer ledger — no float, no drift0
monthly fees. Tokens onlyNo
Template library

330+ crews, calibrated and ready

Content factories, research desks, support triage, ops monitors — clone one and make it yours.

Research

Market brief crew

Researches a topic, extracts drivers & risks, ships a reviewed executive brief.

4 agents · sequential~620 /run
Content

Blog factory

Outlines, drafts and edits an SEO-aware article from a single keyword.

3 agents · sequential~540 /run
Engineering

Bug triage manager

Reads the issue, delegates a repro and a fix proposal, summarizes for the team.

3 agents · hierarchical~710 /run
Sales

Lead enrichment

Gathers firmographics, scores fit, drafts a tailored first-touch email.

3 agents · sequential~480 /run
Finance

Earnings digest

Parses a quarterly report, flags surprises, writes a one-page digest.

4 agents · sequential~660 /run
Support

Ticket responder

Classifies a ticket, retrieves the right docs, drafts a reply to approve.

3 agents · sequential~390 /run
330+

Browse the full library

Every template is one click from its first run.

Pricing

Buy tokens. Spend tokens. That's the model.

No seats, no tiers, no renewals. Top up a balance and every run meters against it.

Starter

€10
tokens
800
bonus
—
≈ runs
2
  • All 6 model providers
  • Sequential process
  • Built-in tools & actions

Pro

€50
tokens
4,400
bonus
+300
≈ runs
10
  • Everything in Maker
  • Priority queue lane
  • Scheduled triggers

Scale

€120
tokens
11,000
bonus
+1,000
≈ runs
27
  • Unlimited API keys
  • Priority queue lane
  • Low-balance alerts

Studio

€250
tokens
24,000
bonus
+3,000
≈ runs
60
  • Everything in Scale
  • Team invites
  • Best per-token rate

Larger packs simply price the token lower.

FAQ

The questions engineers ask first

The single unit every run is metered in. Model usage, connector calls and engine time all convert to tokens, so one integer balance covers everything.

Neither. A top-up is a one-time purchase that sits on your balance until your crews spend it. There is no subscription to cancel because there is no subscription.

Yes — add Anthropic, OpenAI, Gemini, xAI or any OpenAI-compatible endpoint. Your keys are AES-encrypted at rest and only decrypted inside the run engine.

The engine reserves tokens before a run and releases the reserve on failure. Only steps that completed successfully are charged — the ledger shows both movements.

Every crew is callable over an authenticated REST API with the same billing guarantees as the UI. Keys are stored hashed and can be revoked instantly.

Credentials are encrypted, API keys hashed, and the ledger is append-only. Your prompts and outputs stay inside your workspace.

0+crew templates
0+connectors wired
0%ledger visibility
0subscriptions
calibration ready

Put a crew on the bench today.

Sign up, claim your starting tokens and run your first pipeline in the next ten minutes.

Start with welcome tokens