Script.itPlain-English automation you can read and edit
Visit

Script.it Review

Plain-English automation you can read and edit

Workflow & agent infrastructure
Last updated August 28, 2026
The take

A scripting agent that turns plain English into automations you can actually read and edit, promising if you can live on the early-adopter edge.

Great for

Non-technical GTM and ops people who want automation they can see and edit, sitting between fiddly visual builders and black-box agents.

The catch

It is seed-stage with almost no independent track record, and the credit meter runs on both agent thinking and every workflow run.

Bottom line

Reach for it when the job is a repeatable knowledge task you want inspectable. Pilot one workflow and watch the meter before you depend on it.

At a glance
Starting price$30/mo (Pro, 300 credits)
Free tierYes (50 credits/mo, no card)
Pricing modelCredit-metered: subscription plus on-demand credits at $0.12 each
Best forNon-technical GTM and ops teams
Key integrationsSlack, Gmail, HubSpot, Salesforce, Notion, Airtable, Google Sheets, 600+ total
Watch out forSeed-stage maturity and credit burn on agent-heavy or frequent runs
Feature checklist
Plain-English workflow builddescribe it, the agent writes the script
Readable, editable outputscripts are Python/shell you can inspect and edit
Visual node/DAG editorconversational build, blocks not drag-and-drop canvas
Prebuilt templatesYes
Scheduled / webhook / event triggerspresets, plain-English timing, or cron
Native integrations600+ apps plus custom REST and webhooks
Human-in-the-loop controlshuman- and agent-in-the-loop, snapshots and revert
Agent-agnostic / skills exportuse built-in agent or Codex/Claude Code, export skills
External messagingSlack, WhatsApp, Telegram, or agent email
MCP supportnot supported at time of Tom Granot's April 2026 review
Self-hostingEnterprise only, GCP-managed
Free tier50 credits/mo, unlimited seats, no card
What 6 real users said
4 positive2 mixed

What you're actually getting

Script.it is a scripting agent. You describe a task in plain English, and it writes a small program made of blocks: code steps plus notes, saved as one reusable script.

The pitch is a middle layer. Visual builders like Zapier and n8n are repeatable but slow to wire by hand, and open chat agents run but hide what they did. Script.it generates a script you can open, read, and edit.

Every run gets its own isolated cloud computer and working directory, and every block output is just a file, so a human and the agent can both read what happened.

Script.it is something different, and I think it's the middle ground we've been waiting for.
Tom Granot, GTM practitioner, SyntaxGTMYouTube · April 29, 2026 ↗

Where it earns its keep

The sweet spot is a recurring knowledge task you keep doing by hand: a scheduled GA4 email report, a monitor that watches Hacker News and Twitter and compiles a PDF brief, a lead-enrichment pass, a weekly digest.

You build it by talking to the agent, connect the two or three tools it needs from the 600+ catalog, and put it on a schedule, webhook, or event trigger.

Reviewer Tom Granot built a GA4 report and a monitoring-to-PDF workflow on camera and called it a well-built product, singling out a demo mode that mocks real APIs before you connect anything.

It's a middle layer, and I absolutely love it.
Tom Granot, GTM practitioner, SyntaxGTMYouTube · April 29, 2026 ↗

Where it'll bite you

The first catch is maturity. Script.it is a seed-stage product founded in 2024 with almost no independent review base, so you are an early adopter betting on a young company and roadmap.

The second is generated code you still have to trust. Scripts are readable, but that only helps if someone actually reads the Python or shell before it writes to a production tool. Keep a technically comfortable person in the loop and use the revert snapshots.

At the time of Tom's April 2026 review it also did not support MCP, and a workflow takes a moment to build before it becomes durable.

This is really how automation is meant to be. This is what we've been looking for all of these years.
Tom Granot, GTM practitioner, SyntaxGTMYouTube · April 29, 2026 ↗

What it costs, really

There is a real free tier, 50 credits a month with unlimited seats and no card. Pro starts at $30/mo and scales across six credit tiers up to 10,000, with annual billing 10% off and on-demand credits at $0.12 each.

The number to watch is the credit meter. Credits cover agent tokens, hosted services like web search and email, and every block run, so an agent-heavy or frequently scheduled workflow spends faster than a simple connector.

Price a real workflow end to end, one build plus a month of runs, before you trust anyone else's cost estimate. Enterprise is custom, with SOC 2 Type II, SSO, and GCP-managed self-hosting.

And maybe more than anything else, this is a well-built AI product.
Tom Granot, GTM practitioner, SyntaxGTMYouTube · April 29, 2026 ↗
It takes a moment to build. They make it very easy, but it takes a moment. After you do, it is very durable.
Tom Granot, GTM practitioner, SyntaxGTMYouTube · April 29, 2026 ↗
12 sourcesEvery claim on this page is traceable. Open the receipts.
  1. 1Script.it homepageCurrent positioning and feature surface; vendor claims.official
  2. 2Script.it pricingFree/Pro/Enterprise, credit tiers, credit mechanics.official
  3. 3Script.it docs: Agent chatBuild/run/iterate flow, snapshots, block results.official
  4. 4Script.it docs: Integrations600+ catalog, custom REST, connection sharing.official
  5. 5aitoolnet Script.it listingPlain-language product description; partly vendor-derived.independent
  6. 6theaidb Script.it listingPositioning vs n8n/Zapier/Make; 0 reviews; directory metadata.independent
  7. 7Pivot News Script.it overviewFeature list and fit boundaries; machine-generated from vendor material.editorial
  8. 8SyntaxGTM Script.it reviewPractitioner video review; one hands-on take.independent
  9. 9SyntaxGTM founder conversationFounder interview with Niv Geron.vendor
  10. 10CB Insights Script.it profileFounded 2024, Tel Aviv, Seed VC, AIX Ventures and Composed Ventures.public
  11. 11Niv Geron LinkedInFounder identity and skills/agent-agnostic thesis.vendor
  12. 1210x EditorialNon-clickable decision marker; not external evidence.editorial

Watch it work

Real tutorials and reviews from YouTube, not ours.

Script.it: The Missing Layer Between Zapier and AI AgentsSyntaxGTM (Tom Granot) · The best hands-on look at Script.it. Tom builds two real automations on camera, a daily GA4 email report and a Hacker News plus Twitter monitor that generates a PDF brief, and explains when scripts beat both workflows and agents. Watch the demo-mode section, where the tool mocks real APIs before you connect anything.Watch on YouTube ↗
Knowledge Worker to Architect (Niv Geron, Script.it)SyntaxGTM (Tom Granot) · A founder conversation with Niv Geron on the thesis behind Script.it: knowledge work as an evolving codebase that agents navigate and anyone can steer without coding. Useful context on where the product is headed, and a founder view, not an independent review.Watch on YouTube ↗

FAQ

Script.it is an AI automation platform its makers call a scripting agent. You describe a task in plain English and it builds a working, reusable script made of code steps plus notes, then runs it on demand, on a schedule, or via a trigger. It positions itself as a middle layer between visual workflow builders like Zapier and n8n and open-ended chat agents.
There is a free plan at $0/mo with 50 credits, unlimited seats, and 600+ integrations, no card required. Pro starts at $30/mo and scales across six credit tiers from 300 up to 10,000 credits, with annual billing 10% off and on-demand credits at $0.12 each. Enterprise is a custom quote.
Credits are the single usage meter. They cover agent tokens, hosted services like web search, the LLM gateway, and agent email, plus each individual block run while a script executes. An agent-heavy or frequently scheduled workflow spends faster than a simple connector, so price a real workflow end to end before trusting anyone else's estimate.
No. You describe what you want in plain English and the agent builds the workflow. Every script is stored as readable code, typically Python or shell, so a technically comfortable teammate can inspect and edit it. That readability only helps if someone actually reviews the code before it writes to a production tool.
Those are visual builders where you wire and maintain a graph of nodes by hand, repeatable but slow to change. Script.it generates the automation for you as an editable script and lets you iterate by chatting instead of rewiring a canvas. The trade-off is maturity: Zapier and n8n have far larger install bases, community support, and templates.
A general chat agent runs a task but the process disappears into the conversation, so it is hard to verify or rerun reliably. Script.it saves the work as a visible script with each block output as a file, so you can see exactly what ran, edit steps, and run it again dependably. It suits recurring, rule-shaped tasks more than one-off open-ended ones.
Script.it advertises 600+ built-in integrations, including Slack, Gmail, HubSpot, Salesforce, Notion, Airtable, Google Sheets, Stripe, Linear, Jira, and Shopify. For anything not in the catalog, you can build a custom integration for its REST API or use custom webhooks. The agent detects which tools a workflow needs and prompts you to connect them.
Recurring knowledge-work tasks fit best: scheduled reporting like a daily GA4 email, monitoring and digests like a Hacker News and Twitter watch that compiles a PDF brief, lead enrichment, data cleanup, content drafts, and CRM or outbound steps. It is not built for full custom app engineering or millisecond-latency real-time pipelines.
Script.it says data is encrypted in transit over TLS and at rest. The Enterprise tier is SOC 2 Type II certified and offers self-hosting on GCP-managed infrastructure, SSO, audit logs, and admin controls. On-prem-only deployment on non-GCP infrastructure with no managed option is called out as a poor fit, so confirm data residency and the DPA against your own policy.
It is worth a pilot when the job is a repeatable task you want to keep inspectable and a non-technical person owns it. In an April 2026 review, GTM practitioner Tom Granot called it a well-built product and the middle ground he had been waiting for. The honest caveat is that it is seed-stage with a thin independent track record, so start with one bounded workflow and watch the credit meter.

Same job, other tools

Also filed under Workflow & agent infrastructure:

Base44AI app builder from natural language Lindy AINo-code AI agent platform LovableAI software-building platform