What is the difference between Spinrun Connect and Spinrun Agents?

Connect gives an AI you already run — Claude, ChatGPT, Cursor, any MCP client — governed access to your apps. Agents is Spinrun running the agent for you: templates, schedules, triggers, approvals and channels, on top of the same gateway.

Which models can an agent use?

Any model the platform gateway offers, chosen per agent and changed at any time. Claude, OpenAI, Gemini and open-weight models are all on the list; swapping one out changes nothing about the agent's tools.

Do agents hold my credentials?

No. Connections are made once, by a person, through managed OAuth. The agent calls the gateway with its own scoped key; the gateway holds the tokens, refreshes them and revokes them when you say so.

How do approvals work?

Each connected app carries read, write and destructive rules. A call the rules mark for review pauses the run and lands in the approvals inbox with the proposed action attached. Approve it and the run continues; reject it and the agent is told why.

Can I build agents from Claude Code or the CLI?

Yes. An agent is one JSON document, and the Build endpoint lets any MCP client or the spinrun CLI read, change and publish it — with every reference checked before it is written.

Where does my data live?

Core data is hosted in Frankfurt (AWS eu-central-1 and Vercel fra1). Nothing you connect is used to train models, and the compliance hub lists every subprocessor.

The free plan includes a monthly credit pool, no card required. Paid plans raise the pool and unlock artifacts, the sandbox and team management. Credits meter the work agents do, not the number of people supervising them.

Spinrun Agents

AI agents built by
your team

Pick a template or describe the job. Connect the apps it needs. The agent runs on a schedule, a trigger or a Slack mention — and asks before anything risky.

Get started for free Browse templates

See one work

Roll out specialized agents in minutes

What were our key metrics last week? I need a quick summary for the leadership sync.

Spinrun

Thinking

Leadership needs the standard five KPIs in a hurry, so I'll work from the metrics sheet that holds our agreed definitions. That keeps the formulas consistent with the last few weekly briefs.

Google Sheets

Reading sheets

Thinking

Definitions look unchanged from last week. MRR and active users are best read straight from the warehouse rather than the cached sheet snapshot, so I'll grab those next.

PostgreSQL

Querying database

Thinking

Usage and revenue numbers landed cleanly. Pipeline is the most volatile number so I'll pull it live from Salesforce instead of trusting the staging export.

Salesforce

Checking pipeline

Thinking

Pipeline jumped nearly 40% — that's worth calling out explicitly with the two enterprise deals as the cause so leadership doesn't read it as noise.

Here is last week's KPI snapshot across your connected sources:

Metric	This Week	Prior Week	Change
MRR	$482,300	$471,100	+2.4%
Active Users	12,847	12,390	+3.7%
Support Volume	318 tickets	402 tickets	-20.9%
NPS	61	58	+3 pts
Pipeline Created	$1.24M	$890K	+39.3%

Pipeline created jumped nearly 40% driven by the Meridian and NovaTech enterprise deals entering discovery. Support volume dropped after the v4.2 patch resolved the SSO bug. Want me to format this as a Slack brief for #leadership?

Yes, post it to #leadership and flag the pipeline jump.

Spinrun

Thinking

I'll format the brief in the same template leadership has been reading so the pipeline highlight stands out at the top, then attach the underlying records so anyone can click through.

Slack

Posting to #leadership

Done. Posted the KPI brief to #leadership with the pipeline increase highlighted. I also added a note linking the $1.24M to the two new enterprise opportunities with source records attached.

Weekly Active Users

Last 14 days

MTWTFSSMTWTFSS

Total: 12,847+3.7% vs prior week

How it works

Pick it. Connect it. It runs.

Nothing to install. Everything happens in the browser, on apps your workspace has already connected.

01
Pick a template, or describe the job
Start from one of the templates below, or write one sentence and let the agent draft its own instructions.
02
Connect the apps it needs
Managed OAuth for 1,000+ toolkits through Spinrun Connect. The agent never sees a password or holds a token.
03
It runs — and asks first
On a schedule, a webhook or a Slack mention. Writes that matter wait for a human in the approvals inbox; every call lands in the log.

Template library

Start from a template

Ready-to-run agents for sales, data, support, marketing, ops and research. Each names the apps it expects; connect them once and it works.

36 templates

Capabilities

Everything you need to make AI work

Every model out of the box
no vendor lock-in

Connect to internal
and external data

Recurring tasks to keep your agents
running in the background

Social Monitor

Mondays at 8 AM

New Mention

Lead Qualifier

Every submission

Qualified

Security Audit

Every 8 hours

Resolved

0H2H4H6H8H

A human approves the risky parts

Approval requested2 min

Sales follow-up

Send 9 follow-up emails

Drafts attached · Gmail · 9 recipients

ApproveReject

A canvas to orchestrate
multi-agent workflows

Memory that cites its sources

What did we promise Acme in the last QBR?

Q3 roadmap: SSO by October, per the QBR deckDrive
Acme asked for a dedicated Slack channelNotion
Renewal at $48k, flagged as at riskHubSpot

Build your agents from Claude, Codex and the CLI

Build from your coding agent

Describe the agent. Spinrun publishes it.

Every agent and workflow is one JSON document. Claude, Codex, Cursor or the CLI reads it, changes it and puts it back — every reference checked before anything is written, published as you.

Claude · connected to Spinrun Build

Make an agent that triages new Linear bugs every weekday morning and posts a summary to #eng.

Drafted Bug triage. Publishing it to your workspace:

spinrun_build_put_agentdocument

{
  "kind": "spinrun.agent/v1",
  "name": "Bug triage",
  "model": "claude-sonnet-5",
  "tools": { "apps": ["linear", "slack"], "connections": "all" },
  "schedule": { "cron": "0 9 * * 1-5", "prompt": "Triage overnight bugs." }
}

Published bug-triage v1 · applied: version, publish, schedule

Read the Build docs

Channels

Meet your agents where you work

Interact with agents like co-workers in Slack, WhatsApp, email — or pick up the phone.
Tag @Spinrun and ask.

PhoneSoon

TeamsSoon

GmailSoon

Chat widgetSoon

data-analysisLive agent channel

M

Max3:12 PM

@Spinrun Compare this month's revenue to last quarter

💬 2🔥 1

Spinrun replied · 4 min ago

S

Sarah4:42 PM

@Spinrun Who are our top 3 most engaged customers?

💬 1👀 1

Spinrun replied · 2 min ago

C

Carlos8:50 PM

@Spinrun Where are we losing users in onboarding?

💬 1🎯 1

Spinrun replied · just now

How it compares

Not a script. Not a chatbot.

Cron jobs need a maintainer and a coding agent on a laptop needs a babysitter. A Spinrun agent runs unattended, inside the rules the workspace set.

	Scripts and cron	A coding agent on a laptop	Spinrun Agents
Nothing to install	a server to keep	a laptop to keep open	runs in the browser
Runs unattended			schedules and triggers
Human approval gates		Partlyper prompt	an approvals inbox
Every call logged			at the gateway
Any model, swap freely
1,000+ apps, OAuth handled	you wire each one	Partlykeys in config files	through Spinrun Connect
Team roles and shared spaces
Who maintains it	You	You	We do

Safety & security

Same gateway, same rules, same log

Every tool call an agent makes goes through Spinrun Connect. Read, write and destructive rules are checked on each call, credentials stay in the vault, and the activity log is the one the rest of the workspace already reads. How Spinrun Connect works →

Fine-grained data access controls, per team and per action.

GDPR
A Romanian company under the GDPR and ANSPDCP, with a DPA that binds on acceptance and a countersigned copy on request.
EU AI Act aligned
Human oversight at the gateway, transparency duties named, no conformity claim we cannot back.
Core data hosted in Frankfurt, EU
Database, authentication and application functions run in Frankfurt. What is not region-pinned, we say so.
No training on your data
Neither we nor the model providers we route through may train on your content.
SCCs for non-EEA transfers
Standard Contractual Clauses, with encryption and access control behind them, wherever data leaves the EEA.

See our compliance hub →

Pricing

Start free. Pay for the work, not the seats.

Agents meter the tool calls they make against a monthly credit pool. Teammates watching them cost nothing, and nothing runs past the pool.

See plans and credits

FAQ

Frequently asked questions

Get started for free today.

Sign up and put the first agent to work in minutes, or talk to us about rolling agents out across a team.

Get started for free Talk to sales

AI agents built by
your team

Roll out specialized agents in minutes

Data Analysis Agent

Support Agent

CRM Agent

Meeting Prep Agent

Call Analysis Agent

Pick it. Connect it. It runs.

Pick a template, or describe the job

Connect the apps it needs

It runs — and asks first

Start from a template

Everything you need to make AI work