Most enterprise AI programmes end up building the same things: a way to search company knowledge, a retrieval layer over documents and databases, prompt management, tracing for every model call, a data catalog, pipelines to move and prepare data, and a security layer around all of it. InsightLake is a suite of products for exactly those pieces, built by VerticalServe and designed to run inside your own environment.
Three families
Enterprise platform: building blocks for AI teams
Sixteen products that platform and AI teams compose into production systems. The ones most teams start with:
- Insight Ask: AI search across SharePoint, Teams, OneDrive, Jira, Confluence, Slack, Git, ServiceNow, databases and object stores, deployed in your infrastructure.
- Insight RAG: the retrieval layer, with source connectors, content extractors, pluggable vector databases, real-time and batch pipelines, metadata with access control and audit, and redaction and guardrails.
- Insight AI Assist: an enterprise assistant with specialised agents for productivity, automation and simulations.
- Insight Lense: tracing and observability for model calls and agents, with cost, scores and an audit record of what an agent was allowed to do.
- Insight Prompts: a registry for authoring, testing, approving and deploying prompts.
- Insight Catalog: metadata discovery, lineage and governance across databases, lakes, pipelines and BI.
- Insight Dataflow, Extract and Automation Hub: pipelines, document extraction, and migrations with automated validation.
- Insight DLM: a framework for curating data and fine-tuning domain-specific small and large language models.
Domain solutions: for line-of-business owners
Six vertical products built for domain experts rather than engineers: Insight UW for insurance underwriting, Carts for retail, Variant for genomic variant classification, Tax 360, Fundz for fund intelligence and Ad Scale for media analytics.
AI apps: standalone and ready to use
Ten-plus apps for specific jobs, from Kafka health checks (InsightKafka) and scheduling (CalFlare) to call assistance (Vaantra), content-to-video (DraftPlane) and learning (iLearnRight). Each has its own site.
One deployment model
What ties the suite together isn't a single interface; it's how the products are deployed and governed. Each is built to run in your cloud accounts or on-prem, behind your identity provider, next to your data, with audit trails and access control as defaults. That's the subject of running AI inside your environment.
How one programme uses three families
A made-up regional insurer starts with the platform: Ask for staff search over policies and runbooks, RAG behind it, Lense to trace every answer. The underwriting team then adopts Insight UW, which uses document extraction for broker submissions. A year in, the data team adds Catalog to document where every field the models use comes from. Different buyers, one stack and one deployment pattern.
Where to read more. Every product has its own site with details, and the product catalog lists the whole suite. The reference architecture shows how the platform products fit together.
Choosing where to start
Which family you start with depends on who is buying and what problem is most urgent:
| If you are… | Start with | Because |
|---|---|---|
| An AI or platform team | Ask, RAG and Lense | Search, retrieval and tracing are the base every later product uses |
| A data team | Catalog, Dataflow, Automation Hub | Lineage, pipelines and migrations come before models |
| A line-of-business owner | The domain solution for your work | Built for domain experts, not engineers |
| A team with one job to do | The AI app for that job | Standalone, with its own site and sign-up |
The products are designed to compose: a domain solution can use extraction, a platform team's retrieval layer can serve several applications, and tracing and prompt governance apply across all of them. That's the point of buying from one suite rather than stitching together separate tools. You can still buy only what you need and integrate the rest as you grow.