AI Studio
All features
- 01/ 08
Prompt Engineering Studio
Build your system prompts in a visual editor, fine-tune LLM parameters like Temperature · Top-P · Frequency Penalty, and instantly validate with live chat on the same screen.
Prompt Engineering StudioSample screenPrompt studioGPT-4.1 · 128kWhat's the approval limit?Invoices above $5k go to a manager.Thanks.Type a messageTemperature0.8
Top-P0.6
Max tokens2020
Frequency penalty0
- 02/ 08
RAG-Ready Knowledge Base
Upload your corporate documents via drag-and-drop and build your AI assistant's memory in minutes.Azure Cloud or Nisus Local: your data stays where you want it.
RAG-Ready Knowledge BaseSample screenKnowledge baseNisus LocalUploadDocumentChunks- Procurement policy.pdf1842,4 MB
- Invoice rules.docx96820 KB
- Supplier list.xlsx3121,1 MB
- Approval matrix.pdf64540 KB
Indexing · Approval matrix.pdf83%
- 03/ 08
Smart Chunking & Vector Search
Optimize your documents with four chunking strategies: Heading, Sentence, Paragraph, Word Limit.Validate in a chunk tree view, and instantly test results with vector and semantic search.
Smart Chunking & Vector SearchSample screenChunking & searchHeading-basedHeadingSentenceParagraphWord limit- 1. Approval flow
- 1.1 Limits
- 1.2 Approvers
- 2. Exceptions
- 2.1 Urgent buys
- 04/ 08
Function Calling & API Actions
Give your assistant the power to interact with the outside world.Define REST endpoints, model parameters in a hierarchical structure, accelerate with JSON import, and test with a single click.
Function Calling & API ActionsSample screenDefine actionSendGEThttps://erp.local/api/v1/invoicesParameterTypeReq.supplierstringdatedatestatusenum200 OK{"supplier": "Acme","amount": 184500,"status": "open"} - 05/ 08
Assistant Assembly
Prompt, Knowledge Base and Actions: combine all building blocks under a single AI assistant.Configure end-to-end with a 4-step wizard: Describe → Select → Configure → Deploy.
Assistant AssemblySample screenAssistant assembly2/4- Describe
- Select
- Configure
- Deploy
PromptInvoice assistant promptKnowledge base4 documentsActions3 actions selectedModelGPT-4.1 · 128k - 06/ 08
Live Testing & Fine-Tuning
Chat with your assistant in real time: send PDFs, images and evaluate Markdown responses.Optimize streaming, context depth and search precision in a live environment.
Live Testing & Fine-TuningSample screenLive testStreamingfatura_2026_0412.pdfCheck this invoice.Amount and supplier match, ready to approve.Who approves it?The purchasing manager.Session tokens
3,291
Avg. reply
1,4 sn
- 07/ 08
One-Click Deployment
cURL · Python · C# · JavaScript · Postman: get production-ready integration code in your preferred language.Platform agnostic, technology agnostic; just copy-paste and go live.
One-Click DeploymentSample screenDeploymentLive · v1.3https://studio.nisus.local/v1/agents/faturacURLPythonC#JavaScriptPostmanimport requestsr = requests.post("…/agents/fatura/chat",json={"message": "Bu faturayı kontrol et"})print(r.json()) - 08/ 08
Command Center
Monitor your document, prompt, assistant and action metrics in real time.Token consumption, session counts, average response times: all your KPIs on a single dashboard.
Command CenterSample screenCommand centerTodayAssistants
12
Active
7
Documents
284
API calls
14,820
- Invoice assistantActive
- Procurement assistantActive
- HR assistantIdle
Would you like to see the platform with your own scenario?
In a brief 15-minute call, evaluate Central against your own business processes together with our specialists.
