swfte
로그인팀에 문의하기AI로 구축하기

04 Studio

Describe the work.Studio builds it.

Build agents and workflows visually or by describing them. Policies decide what they may do, and a person takes over when it matters.

Build with AI Talk to our team

[ 04 ] StudioBuild. Govern. Run.

01 In this layer

Agents and workflows.

  • Build by describingAvailable
  • Policies attached to runsAvailable
  • Hand-off to a personAvailable

Build

StudioNexus IDEOur MarketplaceTemplatesIntegrations

Agents and workflows

Governed AgentsGoverned WorkflowsAI SolutionsGoverned agentsCompliance workflowsAll use cases

03 SEE IT WORK

See Studio at work.

Describe it

Start with a sentence, not a blank canvas.

Choose what you are building, from an agent or workflow to a journey, chatflow, widget or MCP server, and describe the job. Studio names each phase while it drafts, checks and saves, then hands you a draft to review.

See every screen

02 [ 04.A ]

From a sentence to a governed workflow.

Describe the job. Studio assembles the steps from building blocks you can inspect and change.

Available

03 [ 04.B ]

A person approves what matters.

Approval steps pause the run until someone decides. A gate never approves itself.

Available

04 [ 04.C ]

Every run leaves a record.

Each run is written to a run ledger you can read and verify.

Available

[ 03.A – 03.L ] INSIDE STUDIO

Inside Studio, screen by screen.

Twelve chapters, each a real screen from the console, assembling as you scroll.

S1 [ 03.A ]Describe it

Start with a sentence, not a blank canvas.

Choose what you are building, from an agent or workflow to a journey, chatflow, widget or MCP server, and describe the job. Studio names each phase while it drafts, checks and saves, then hands you a draft to review.

Available
  1. 01Pick a mode
  2. 02Describe the job
  3. 03Watch it draft
  4. 04Review the draft

S2 [ 03.B ]Workflow canvas

The work, drawn as steps you can read.

Each node says what it is, what it takes in and where its output goes. Typed ports carry data along the edges, a condition splits the path in two, and a failure can follow its own error route.

Available
  1. 01Trigger and steps
  2. 02Typed ports
  3. 03Branch
  4. 04Error path

S3 [ 03.C ]Node library

Start from blocks, or save your own.

AI steps, flow logic, data tools, triggers and connected services sit in one searchable library. Drop a block onto an edge and it slots in between. Save a configured node and your whole team can reuse it.

Available
  1. 01Open the library
  2. 02Filter and search
  3. 03Drop on an edge
  4. 04Save your own

S4 [ 03.D ]Configure a step

Set up one step, then try it.

Pick a credential and an operation, pull in another step's output, and decide what happens if it fails. Then run only that step with a test input and read its result on the card before you run the rest.

Available
  1. 01Open the step
  2. 02Credential and operation
  3. 03On error
  4. 04Run this step

S5 [ 03.E ]Test run

Watch each step run, then fix it.

Start a test run and every node reports its state as it goes, with activity grouped by step. When one fails, the drawer shows which step and why, and Retry starts the run again once it is fixed.

Available
  1. 01Start the run
  2. 02Step by step
  3. 03A step fails
  4. 04Retry and read

S6 [ 03.F ]Human approval

A person signs off on the next action.

When a run reaches a gate it pauses and shows the exact action the agent wants to take. The reviewer approves it, edits it and approves the edited version, or denies it, and the run carries on from that decision.

Available
  1. 01The run pauses
  2. 02Proposed action
  3. 03Edit the action
  4. 04Approve edited

S7 [ 03.G ]Agent canvas

An agent, laid out part by part.

Model, instructions, knowledge, tools and channels each have a place on the agent canvas. Attach a source or a tool where it belongs, then run the tree audit to see what is still missing before anyone talks to it.

Available
  1. 01Identity and modules
  2. 02Attach knowledge
  3. 03Attach a tool
  4. 04Audit the tree

S8 [ 03.H ]Knowledge bases

Give agents knowledge you can test.

Upload documents to a knowledge base and follow each one through parsing, cleaning, splitting and indexing. Then ask a test question and see which passages come back, and how closely they matched, before an agent relies on them.

Available
  1. 01Knowledge bases
  2. 02Ingest
  3. 03Ask a test question
  4. 04Retrieved chunks

S9 [ 03.I ]Run ledger

Read back what ran, in order.

The run ledger lists workflow, agent and worker runs with a plain outcome for each, failures included. Open a run's trace to read every event in sequence and see whether its hash chain verifies.

  1. 01Every run, one list
  2. 02Read the record
  3. 03Open the trace
  4. 04Check the chain

S10 [ 03.J ]Deploy

From draft to a running service.

Publish a version, then deploy it with the target, size, estimated cost and runtime mode in front of you. Watch it come up phase by phase, and see each deployment's facts kept apart, including the ones not yet known.

Available
  1. 01Publish a version
  2. 02Review before deploying
  3. 03Watch it come up
  4. 04Every deployment, plainly

S11 [ 03.K ]Across industries

Same blocks, different work.

The blocks that build, pause, record and ship work are the same in every team. The scenarios below are illustrations, not customer deployments.

S12 [ 03.L ]Why Studio

Build it, run it, show what it did.

One workspace takes work from a sentence to a deployed service, with a person in the loop where it matters and a readable record of every run.

  1. 01Eight pieces, one workspace
  2. 02From a sentence to a service
  3. 03Explore each piece

Start from a sentence

Describe the job, review the draft, then refine it on the canvas.

See how

Real nodes, wired by hand

Triggers, AI steps, branches and integrations, connected handle to handle.

See how

Your own building blocks

Browse the node library and save any step as a custom node.

See how

Try one step first

Run a single step with a test input and read its output on the card.

See how

A person signs off

A gate pauses the run: approve, edit and approve, or deny.

See how

Agents with knowledge you can test

Lay out an agent part by part and see which passages its knowledge base returns.

See how

Every run on the record

Plain outcomes in the run ledger, and a trace whose hash chain you can check.

See how

Deploy with the facts in view

Size, estimated cost and runtime before you deploy; unknown facts stay marked unknown.

See how

Where Nexus is connected, Checks lists its findings for review.

Build with AI Talk to our team

Illustrative demo

How it works

From a sentence to a governed agent.

01

Build by describing.

Say what the agent should do. Studio assembles the workflow from building blocks you can inspect and change.

Available
Studio screen: Build by describing.
02

Bound it by policy.

Guardrails keep private information out and the agent on topic. Policies apply at runtime on runs that have one attached.

Available
Studio screen: Bound it by policy.
03

Hand off to a person.

When a conversation needs a human, the agent passes it on with a summary, and your team takes it from there.

Available
Studio screen: Hand off to a person.
04

See every run.

Step inside any agent: every conversation and every action, recorded for the people who answer for it.

Available
Studio screen: See every run.

06 Agent thread

Describe a workflow. Watch it get built.

One sentence becomes a workflow with its systems, an approval gate and a policy, tested before it goes live.

  • AvailableBuilding from a description, connecting systems, approval steps that pause a run, test runs and publishing.
  • Designed forWriting the policy from the builder. Policy is enforced today on runs that have one attached.

Illustrative example with a fictional company. Not a real customer, not real data.

Studio · builder

Priya Shah, Operations, Fenwick Logistics

/new-workflowRoute refund requests over the threshold to finance for approval

I will draft it, connect your systems and test it before anything goes live.

  1. Drafted the workflow5 stepsAvailable
  2. Connected the systemsTicketing, Payments, EmailAvailable
  3. Added an approval gateFinance Operations, over thresholdAvailable
  4. Set the policyno card data in prompts, EU region onlyDesigned for
  5. Ran a test3 cases: 2 automatic, 1 to approvalAvailable

Workflow

Ticket inCheck amountApproval gateRefundNotify

Test run

  • Case 1, under thresholdrefunded automatically
  • Case 2, under thresholdrefunded automatically
  • Case 3, over thresholdwaiting for Finance
Available

Policy

No card data in prompts. Runs stay in the EU region. Refunds over the threshold wait for a person.

Designed for

Ready to publish. Refunds over the threshold will wait for Finance Operations.

Available
Illustrative example
Enterprise IntelligenceAll layersGovernance

Frequently asked questions

Swfte Studio로 무엇을 구축할 수 있나요?

Swfte Studio에서 AI 에이전트, 대화형 챗플로우, 자동화된 워크플로우, 멀티채널 경험을 구축할 수 있습니다. 고객 지원 봇부터 영업 자동화, 문서 처리부터 데이터 분석까지—AI와 관련된 것이라면 Studio에서 구축할 수 있습니다.

Studio로 첫 번째 에이전트를 얼마나 빨리 배포할 수 있나요?

Studio templates give you a starting point, and the time to a first working agent depends on your use case, data and integrations. For custom enterprise solutions, we scope timelines with you.

Swfte Studio를 사용하려면 코딩이 필요한가요?

코딩이 필요 없습니다! Studio는 비즈니스 사용자부터 개발자까지 모든 사람을 위해 설계되었습니다. 시각적 드래그 앤 드롭 인터페이스는 AI 경험을 블록 조립처럼 간단하게 만들어 줍니다.

Studio는 어떤 통합을 지원하나요?

Studio는 100개 이상의 도구와 기본적으로 연결됩니다: Salesforce, HubSpot, Slack, Microsoft 365, Google Workspace, GitHub, Zendesk, Notion 등. 또한 API를 통해 모든 맞춤형 시스템에 연결할 수 있습니다.

Keep exploring

Guides

Other products

[ 07 ] One platform

Capability and control, in one platform.

Start with one layer and prove it. Add the next when you are ready. The controls come with you.

Build with AI Talk to our team

Build this in Studio

Describe what you need in plain language. Studio builds the agents and workflows, and you keep every version.

Start building freeLearn more