DemoBook a demo

From the enterprise blog

Practical answers about private AI, document processing, and agents at work.

33 articles · Page 1 of 3

Enterprise agents
When an AI assistant should say it lacks enough information
Evaluate useful abstention, precise clarification, and partial answers when permitted evidence is missing, ambiguous, inaccessible, or conflicting.
Sarath Chandran
Sarath Chandran
September 8, 20265 min read
Private AI
Which AI agent integrations work without internet access?
Assess offline AI agent integrations by checking endpoint location, authentication, dependencies, and data freshness rather than assuming local plugins are disconnected.
Sarath Chandran
Sarath Chandran
September 2, 20266 min read
Document processing
Normalizing extracted dates, currencies, and identifiers
Keep original values and resolve ambiguity before converting document dates, monetary amounts and identifiers into typed data.
Sarath Chandran
Sarath Chandran
August 28, 20265 min read
Enterprise agents
AI agent error handling: retry, resume, or escalate?
Distinguish confirmed failures from uncertain writes, preserve completed work, and test recovery without duplicating actions or hiding partial results.
Sarath Chandran
Sarath Chandran
August 25, 20266 min read
Private AI
Updating self-hosted AI: what to test before changing versions
Plan version compatibility, representative task checks, migration rehearsals, and rollback decisions before updating a self-hosted AI deployment.
Sarath Chandran
Sarath Chandran
August 21, 20266 min read
Document processing
Multilingual OCR: what to test beyond language support
Build a multilingual document evaluation that distinguishes recognition, reading order, field interpretation and translation.
Sarath Chandran
Sarath Chandran
August 18, 20266 min read
Enterprise agents
Revoking access to documents already indexed by AI
Test permission revocation across new retrieval, answers, citations, previews, and existing conversations using controlled identities and source markers.
Sarath Chandran
Sarath Chandran
August 12, 20268 min read
Private AI
Backing up private AI: what needs to be recoverable?
Define a recoverable private AI backup across application data, files, configuration, indexes, and versioned artifacts, then prove it with an isolated restore.
Sarath Chandran
Sarath Chandran
August 7, 20266 min read
Document processing
One PDF, several documents: testing document splitting
Test page boundaries, repeated document types and missing pages before extracting records from a combined PDF packet.
Sarath Chandran
Sarath Chandran
August 4, 20268 min read
Enterprise agents
How to verify that AI citations actually support an answer
Use a claim-level review to distinguish a valid citation from evidence that actually supports an enterprise AI answer and its qualifications.
Sarath Chandran
Sarath Chandran
July 31, 20266 min read
Private AI
Monitoring self-hosted AI: latency, throughput, and errors
Choose LLM monitoring metrics that separate queueing, first-token delay, generation, failures, and dependency health without collecting unnecessary task content.
Sarath Chandran
Sarath Chandran
July 28, 20269 min read
Document processing
Keeping extracted fields traceable to their source documents
Preserve source identity, page references, original text and transformations when turning documents into structured records.
Sarath Chandran
Sarath Chandran
July 24, 20266 min read

The resource library

There’s more to explore.