Sign in to continue

Two-Factor Auth

Enter the 6-digit code from your authenticator app.

Set Up 2FA

Scan this QR code with your authenticator app (Google Authenticator, Authy, etc.)

QR Code
OnBase Assistant v2
Role
Switching roles clears the chat
General
Support
Technical
Developer
Solutions Architect
Sources
Select which doc hubs to search
Development & SDK
Release Notes & Community
Module Reference Guides
OnBase Version
Pick a version for precise MRG results
BETA — AI-generated responses may be inaccurate. Always verify results before acting on them.

What are you investigating?

Code Workbench

Convert, upgrade, and validate code for OnBase Unity API — grounded in SDK reference and DataBank standards
VBScript / Legacy C#
Compliant C#
' VBScript — Common workflow patterns
Dim doc = app.Core.GetDocumentByID(docID)
doc.AddToHistory("Processed by import")

Dim remaining = objApplication.BatchDocumentsRemaining

objApplication.PropertyBag.SetProperty("Status", "Complete")
objApplication.PropertyBag.Clear
// C# Unity API — OnBase 25.1 compliant
Document doc = app.Core.GetDocumentByID(docID);
app.Core.LogManagement.CreateDocumentHistoryItem(doc, "Processed by import");

int remaining = wfArgs.BatchDocumentsRemaining;

wfArgs.SessionPropertyBag.Set("Status", "Complete");
// SessionPropertyBag persists across batch — no .Clear needed
' VBScript — Keyword creation & error handling
Set RegString = New regexp
If RegString.Test(emailValue) Then IsValid = "Y"

keyType.CreateKeyword(keywordValue)

If err.Number <> 0 Then
  Err.Raise vbObjectError, , "Keyword failed"
End If
// C# Unity API — Safe patterns
bool isValid = Regex.IsMatch(emailValue, pattern,
    RegexOptions.Compiled, TimeSpan.FromSeconds(1));

if (!keyType.TryCreateKeyword(keywordValue, out var keyword))
    throw new DatabankException(
        $"Unable to create keyword [{keywordName}]");
' VBScript — Document locking & autofill
Dim docLock = doc.LockDocument()
mod.ApplyChanges()
docLock.Release()

Call results.BeginRow()
Call results.AddData("Email", primaryValue)
Call results.EndRow()

pdfType = app.Core.FileTypes.Find(16)
// C# Unity API — Resource discipline
using (DocumentLock docLock = doc.LockDocument())
{   mod.ApplyChanges();
}   // lock auto-released

var row = afArgs.Results.CreateExternalAutofillKeysetData();
row.SetKeyword(CreateKeyword("Email", primaryValue));
afArgs.Results.AddExternalKeywordAutofillKeysetData(row);

const int PDF_FILE_TYPE_ID = 16;
pdfType = app.Core.FileTypes.Find(PDF_FILE_TYPE_ID);
AddToHistory() → LogManagement API
PropertyBag → SessionPropertyBag
objApplication.Batch* → wfArgs.*
CreateKeyword() → TryCreateKeyword()
Regex → ReDoS-safe with timeout
ApplicationException → DatabankException
Lock/Release → using{} auto-dispose
BeginRow/AddData → Autofill API
Find(16) → named constant
↔ Drag the slider to compare
✓ Validated against OnBase SDK and DataBank standards ↑ see examples
Action
VBScript or VB.NET → compliant C#
Migrate C# from one OnBase version to another
Check C# against DBIMX standards and SDK best practices
Break down what this code does, section by section
Describe what you need — get a complete, ready-to-use script
VBScript → C# Unity API — complete, compilable output
Analysis Mode
/
Source
→
Target
From
→
To
OnBase Version
0 lines · 0 chars
Drop file here
Daily Budget
—
Code Workbench convert
BETA — AI-generated responses may be inaccurate. Always verify results before acting on them.

SA Survey Generation

Upload discovery notes, meeting transcripts, or any source material. Box AI fills the SA survey template from it.
Accepted: .doc, .docx, .txt, .md, .png, .jpg, .jpeg  ·  Max 20 MB per file, 10 files per run
Uploaded Files:
     

    Your SA Survey

    BETA — AI-generated responses may be inaccurate. Always verify results before acting on them.

    Ask me anything about OnBase

    Tone:
    Drop, paste, or click to attach a screenshot
    screenshot

    Under the Hood

    Five AI systems running in parallel so you can just ask a question.

    You
    Box AI
    OpenAI
    Claude
    Supabase
    n8n

    What happens when you ask a question

    1
    Intent Classification - AI reads your question and determines what you need: new question, case investigation, follow-up, or documentation lookup.
    2
    Parallel Search - Your question fans out simultaneously to documentation hubs AND case intelligence search. No waiting in sequence.
    3
    Synthesis - AI merges all sources into one cohesive answer with strict anti-hallucination rules. Every fact must trace to a source.
    Box AI - Documentation

    RAG-grounded search across 8 curated hubs: version-specific Module Reference Guides (22.1 through 25.1 + legacy), Release Notes, Development SDK, and DataBank Standards. Auto-selects the right hub based on your role and version.

    Supabase - Case Intelligence

    AI-powered semantic search across multiple data sources: cases, emails, comments, SCRs (known defects), KB articles, and case notes. Data synced regularly from Salesforce.

    Claude (Anthropic) - Brain

    Multiple AI models working together. Fast models handle intent classification, search term extraction, and draft customer responses. Advanced models handle synthesis: merging documentation + case data + live Salesforce into a single grounded answer.

    Salesforce - Live Data

    Paste a case number or Salesforce URL and the pipeline pulls live case details, related SCRs, KB articles, case notes, and account history directly from Salesforce. Version is auto-detected from the case record.

    OpenAI - Embeddings

    Every case, email, SCR, and KB article is converted to a high-dimensional vector for semantic search. Hybrid search combines semantic similarity with keyword matching for fast, relevant results.

    n8n - Data Pipeline

    Automated workflows keep data current by syncing case data, SCRs, KB articles, and notes from Salesforce, then generating search vectors. Runs on a regular schedule.

    Three modes, one interface

    G
    General - Release Notes + MRG (latest version). Standard, Sales, or RFP tone. Great for feature questions, what's-new summaries, and proposals.
    S
    Support - Full case research. Paste a case number or SF URL and get documentation + similar cases + SCRs + KB articles + live Salesforce data in one answer. Draft customer response with one click.
    T
    Technical - Dev SDK + MRG + DataBank Standards + SCRs + KB articles. Built for admins, developers, and solution architects.

    All queries processed under enterprise data agreements (Box, Anthropic, OpenAI). No data is used for model training. MFA-secured with row-level database security.

    What's New in v2

    A complete rebuild of the OnBase Assistant.

    New: Support Research Mode

    • Paste a case number or Salesforce URL and get a full investigation in seconds
    • Searches 6 sources in parallel: cases, emails, SCRs, KB articles, case notes, and documentation
    • Live Salesforce lookup pulls case details, related SCRs, and account history
    • Version auto-detected from case data - selects the right MRG automatically
    • Draft Customer Response button - rewrites your research into a ready-to-send email

    New: Self-Contained Pipeline

    • Direct pipeline is faster and more reliable than the previous architecture
    • AI classifies intent instantly, then synthesizes the final answer
    • Parallel execution: documentation + case intelligence search run simultaneously
    • Anti-hallucination rules: every fact must trace to a source document

    New: Semantic Search

    • Thousands of records vectorized for semantic search
    • Hybrid search combines semantic similarity with keyword matching
    • SCRs (known defects) and KB articles now surface for Technical and Support roles

    Improved: Roles & Interface

    • Simplified to 3 roles: General (with Sales/RFP tone options), Support, and Technical
    • Version-specific MRG hubs: 22.1, 23.1, 24.1, 25.1, and legacy (pre-22)
    • Role preference persists across sessions
    • Unified welcome cards with role-specific examples and search tips
    • Version detection from natural language: "v24", "OB22", "OnBase 23" all work

    Improved: Answer Quality

    • Filters generic IT advice - only OnBase-specific details make it through
    • Code blocks for INI settings, config paths, and commands (copyable)
    • Follow-up detection: "tell me more", corrections, and refinements handled naturally
    • DataBank Standards integrated into Technical role responses