Dataset
18 records
One workspace where people and AI coworkers plan, execute and ship real client work. Projects, data, documents and agents in one place - not another chat transcript.
Built inside our own digital studio at Justmade. Now opening to selected design partners.
Built for the work studios actually do
Not another chat wrapper. StudioLayer is a real workspace: structured data, versioned documents, event signals and agents that can actually do the work.
Text, number, boolean, select, relation, file, JSON. Records edit in-place, ship through the content API to any frontend.
Each agent has a role, tools, skills and memory: reusable playbooks that teach them how your studio works. They pick up tasks, ship deliverables, log everything.
Revenue, workload, agent spend: always current, no reporting day.
Every change tracked. Roll back anything, by human or agent.
Hand a task to a person or an agent: same board, same accountability.
Your content flows straight into your website or app. No plugins, no re-entry, no migrations.
Compose briefings and client pages from live data. No BI tool needed.
Not autocomplete. Not a chatbot. A full agent with a role, memory, tools and taste. They read the project before they start: your data, docs and decisions. No context pasting.
Full-stack developer
Ships code. Terminal-mindset. Reads the repo, writes the PR, runs the tests, files the diff.
Editorial writer
Drafts memos, launch posts, landing copy. Knows your tone-of-voice from your docs.
Ops orchestrator
Runs missions across a team of agents. Blueprints, phases, retries, rollups.
Growth analyst
Reads signals, spots outliers, writes the weekly. Owns the dashboard on Monday.
Example agents. Stats reflect StudioLayer's own use at Justmade.
Build your own perfect coworker: any role, any skill set, any tools. Tuned to how your studio works, not the other way around.
A mission takes a goal, drafts a plan, spreads the work across agents and people, and keeps iterating until the result holds up. You review at the checkpoints, not in the weeds.
Every mission begins as a plan: phases, owners, deliverables. You approve it before anything runs.
The orchestrator holds the full picture and briefs each agent with exactly what their step needs: nothing more. Focused agents, sharper output, no context drift.
Missions run in cycles. Each pass reviews what came back, adjusts and runs again: the mission learns what the project needs.
Work moves between steps as parcels: validated handoffs that say exactly what is expected. Nothing ships half-empty. Lock one and it is the deliverable.
AI tools generate output. StudioLayer gives agents the context, structure and tools to carry the work forward - so they finish jobs, not just answer prompts.
Datasets, documents and views on structured nodes. Everything an agent produces lands somewhere permanent, versioned and reviewable - not in a chat scroll.
A role, a memory, an address, an audit trail. You assign work the way you assign it to people - and you review it the same way too.
Bring your own LLM keys, run agents on your own machines, pipe content to any frontend. Your work, your models - no lock-in by design.
Agents pick up the queue when you put down the laptop - but client work needs more than a clever model. It needs guardrails.
Every output traces back to your own projects, docs and data.
Role, tone and playbooks per agent. It ships sounding like you.
Unsure means asking a teammate, never guessing. You review at checkpoints.
Every action logged. Self-host and bring your own LLM, built for GDPR.
Tests green, PR filed, diff waiting for your morning review.
Scored, enriched, logged in your pipeline dataset. Follow-ups drafted.
Built from the client's brief, your rates and past projects. You edit, you send.
Signals rolled up, outliers flagged, one page. Coffee still hot.
Every action logged. Every deliverable reviewable. You stay in control, they do the hours.
Stop managing tools.
Start managing outcomes.
Plug in your own keys and route each agent to the model that fits the job. Swap providers per agent, no lock-in, and you pay the provider directly for tokens.
+ any OpenAI-compatible endpoint · self-hosted models via your own gateway
A skill is a playbook: how you write proposals, how you review code, how you talk to clients. Write it once, grant it to any agent, and they follow it, every time.
Agents work where your work already lives. One toggle connects an account, the whole toolset comes with it. GitHub is live today; the rest roll out through the beta.
Anything with an API fits through MCP. Or skip the wait: hand an agent the API docs and it teaches itself the tool, then saves what works as a skill for the whole team.
Your clients open a live preview of their site, click any element and type what should change. No CMS training, no back-and-forth emails with screenshots.
Copy tweaks and content edits land straight in the headless CMS. Live on the next publish.
Ask for a new section or layout change and it turns into a support ticket for your studio, with the exact element and context attached.
Every request is reviewable. Nothing ships without your team behind it.
"Can we make this headline shorter and mention the spring offer?"
TypeScript SDK. Signed webhooks. Per-node keys. Zero-config MCP. Everything is an API - or a bring-your-own-runtime.
import { createClient } from '@studiolayer/client' // per-node scoped key, cached reads const studio = createClient({ apiKey: import.meta.env.STUDIOLAYER_API_KEY }) // typed dataset handle const posts = studio.dataset<Post>('blog', 'posts') const all = await posts.list() // or run a saved, server-defined query const featured = await studio.query('featured-posts')
mission: launch-v1.4 orchestrator: atlas phases: - discover · kiro -> market-scan - design · vera needs discover - build · forge needs design - ship · atlas · human gate
End-to-end types from your schema. Autocomplete for every field.
Every tool exposed as MCP. Plug any client. No lock-in.
Every write emits a signal. Trigger anything downstream.
Agents execute on runners you control. Local machine or your own server.
From how we run our own studio at Justmade. Private beta - your experience will vary.
One plan per member for the workspace. Agents run on top and bill for the tokens they actually burn, on your own keys. Start with a 30-day free trial, no card up front.
Expected beta pricing · may change during private beta
For studios that ship weekly. Usage-based agent runs on top.
Start 30-day free trial ->30 days free · no card up front
Dedicated runners, private models, SLA and onboarding.
Talk to us ->Expected beta pricing · subject to change during private beta
The workspace, agents, missions, datasets, documents, views, tasks, the content API and BYO LLM keys are running - it's what we run Justmade on. Runners, connectors beyond GitHub, the visual CMS and enterprise features are in active development, marked live / beta / planned throughout the page. We're in private beta, opening to selected design partners.
It replaces the pile: the PM board, the docs, the data tables and the AI chat tab, in one workspace. The difference is that the same place where you plan the work is where agents and people actually do it and where the output lands - structured, versioned, reviewable.
Only through the project they're attached to: scoped nodes, datasets and documents, with per-project keys. Every read and tool call is logged on the org activity stream. Agents run on your LLM keys, and we never train on your data.
Yes. Any OpenAI-compatible or Anthropic-native provider, keys rotated per org, project or agent. Agents can run on your own machines through a paired runner, so code and data never have to leave your infrastructure. Runners are in beta.
Digital studios and agencies that ship client work: websites, webshops, brand systems, products. Small teams that want AI coworkers doing real work inside a structured workspace, not another chat tab bolted onto the side.
No. Chat is one surface. The core is a structured workspace: typed datasets, versioned documents, event signals, tasks and missions. Agents are one of the ways work gets done in it - alongside humans and your own scripts.
Yes. Any provider that speaks OpenAI-compatible or Anthropic-native works. Rotate keys per org, per project, or per agent. They're encrypted at rest and never exposed back to agents or the UI.
You do. Full export at any time - projects, datasets, documents and history. Agents run on your LLM keys if you want, and we never train on your data.
Justmade, a digital studio in Belgium. We've been building websites, webshops and products for clients for years: we know exactly what studio work looks like, where the hours leak, and what a tool has to do to actually help. Nothing on the market fit, so we built StudioLayer for ourselves first. It runs our studio today - and we're now opening it to selected design partners.
All content encrypted at rest, TLS in transit, per-project keys, full audit log on the org activity stream.
Now in private beta. Join the waitlist and help shape the product.
AI agents are becoming coworkers. StudioLayer is the workspace they'll work in. Get the memo with traction, roadmap and the numbers.