Vertesia
by Vertesia (formerly Composable Prompts)
Model-agnostic enterprise platform for document context, agentic RAG, AI agents and processes
Vertesia is an enterprise generative AI platform for IT and content teams that need to turn large document repositories into AI agents, apps and automated business processes. It stores and prepares documents for AI, runs agents and Temporal-based workflows on top of them, and connects to more than 100 models from Anthropic, OpenAI, Google, Bedrock and Microsoft Foundry with your own keys.
Vertesia, formerly Composable Prompts, is a Brooklyn-based enterprise AI platform co-founded by CEO Eric Barroca, the former CEO of Nuxeo, the content management company Hyland Software acquired in 2021. The company raised a $4 million seed round on February 27, 2024, led by Elaia Partners and Illuminate Financial with Motier Ventures, Kima Ventures, m-ai club and Super Capital. The platform has five parts. A context layer stores files with every revision, converts them to text, extracts cited fields, creates embeddings and supports keyword, semantic or hybrid search on object storage and Elasticsearch; its document preparation step, Semantic DocPrep, is patent-pending. A process engine built on Temporal runs long-lived workflows, imports BPMN process maps and routes items to human reviewers. An agent layer offers built-in tools, MCP tools and skills, with code executed in isolated sandboxes. A model layer built on Vertesia's open-source Llumiverse connector supports Anthropic, OpenAI, Google, Amazon Bedrock and Microsoft Foundry with bring-your-own keys, plus fine-tuned or self-hosted models. Studio and App Builder are the build surfaces, and in October 2025 the company added an AI Assistant chat app for employees with a Reasoning Mode that hands multi-step tasks to agents. Vertesia runs as SaaS on AWS, Azure or Google Cloud in the US, Europe and Japan, in a customer's own cloud account, or on any Kubernetes environment, and states SOC 2 Type II certification. Named customers include Accor, which the vendor says migrated off a legacy document platform in 2.5 months, and apparel maker Mad Engine. IDC named it an Innovator twice in 2026.
An enterprise architect or head of content and document operations replacing a legacy document platform and wanting agents and RAG on the same repository.
One governed place to store documents, prepare them for AI and run agents and approval workflows on them, with the freedom to switch model providers.
At a Glance
- Category
- Enterprise Platform
- Pricing
- Subscription, Contact for pricing
- Target Market
- CIOs, CTOs, Enterprise Architects, Enterprise Developers
- Deployment
- Cloud-first, Self-hosted, Multi-cloud
- Headquarters
- Brooklyn, NY, USA
Key Features
- ✓Enterprise context layer
Stores files with full revision history, converts them to text, extracts cited fields and builds embeddings for hybrid search.
- ✓Agentic RAG
Agents retrieve from the prepared repository with keyword, semantic or hybrid search, so answers cite the underlying documents.
- ✓Temporal-based process engine
Runs long-lived business processes, imports BPMN maps and notifies human reviewers by email for approval steps.
- ✓Multi-model support with bring-your-own keys
Connects to more than 100 models across Anthropic, OpenAI, Google, Amazon Bedrock and Microsoft Foundry, plus self-hosted models.
- ✓Sandboxed agents with MCP tools
Agents use built-in tools, MCP tools and skills, and any code they run executes in isolated sandboxes.
- ✓AI Assistant for employees
A governed chat app with role-based access, activity logs, shared agents and a Reasoning Mode for multi-step tasks.
Capabilities
Use Cases
- •Legacy document platform replacement
Accor, per Vertesia, replaced a legacy document platform and migrated its whole repository onto Vertesia in 2.5 months.
- •Document intake and extraction
Incoming files are converted, fields are extracted with citations and records are routed into a reviewed downstream workflow.
- •Governed employee assistant
Employees ask questions over company content in the AI Assistant while admins control access by role and review activity logs.
- •Self-hosted AI in a government cloud
An agency self-manages Vertesia in AWS GovCloud or Azure Government to keep data inside its own authorised environment.
Ideal For
Best For
- ✓Migrating off a legacy document or content management platform onto an AI-ready repository
- ✓Agentic RAG over large internal document collections with cited field extraction
- ✓Long-running document workflows with human approval steps, imported from BPMN process maps
- ✓Teams that want to bring their own model keys across Anthropic, OpenAI, Google, Bedrock and Microsoft Foundry
- ✓Regulated or government buyers who need to self-host in their own cloud account or Kubernetes cluster
Not Ideal For
- ✗Small teams wanting a monthly self-serve plan; every tier is an annual subscription with pricing on request
- ✗Buyers who need ISO 27001 or a FedRAMP authorisation held by the vendor; Vertesia states SOC 2 Type II and supports compliance within the customer's own authorised environment
- ✗Teams that only need a chat assistant over Microsoft 365 or Google Workspace and have no document processing or workflow requirement
Integrations
Deployment
Market Analysis
Pros
- ✓Model choice across major providers with bring-your-own keys reduces lock-in
- ✓Broad deployment options, including self-managed in a customer's own cloud or Kubernetes
- ✓Leadership with long content-management experience from Nuxeo, and a named enterprise migration at Accor
Cons
- ✗No published pricing and annual-only subscriptions make early cost comparison hard
- ✗Small public footprint: its main open-source repos have about 30 stars or fewer, and no Hacker News discussion of the product was found
- ✗ISO 27001 is not listed; HIPAA, GDPR and CCPA are described as supported compliance rather than certifications
- ✗The AI Assistant was covered mainly through vendor-supplied press (KMWorld, October 2025), with no independent hands-on review found
Pricing
Pro
Contact for pricing
- ✓Best for small teams
- ✓Email support
- ✓Standard SLA
- ✓Standard SSO
Expert
Contact for pricing
- ✓Medium to large teams
- ✓Email, chat and phone support
- ✓Shared customer success
- ✓Fine-tuning
Enterprise
Contact for pricing
- ✓Multiple teams or enterprise-wide programs
- ✓Premium SLA
- ✓Custom SSO
- ✓Dedicated customer success
- ✓Optional self-managed or dedicated SaaS
Vertesia sells annual subscriptions in three tiers (Pro, Expert, Enterprise) and publishes no prices; every tier says request pricing. All tiers include the studio, multi-model support, agent builder, agentic RAG and SOC 2 Type II security, while fine-tuning starts at Expert and self-managed or dedicated SaaS is an Enterprise option. Model usage runs on your own provider keys.
Security & Compliance
Connect
Sources
This page was written from 8 sources, 3 on domains other than vertesiahq.com.
- 1.vertesiahq.com — vertesiahq.comvendor
- 2.vertesiahq.com — platformvendor
- 3.vertesiahq.com — pricingvendor
- 4.vertesiahq.com — securityvendor
- 5.vertesiahq.com — aboutvendor
- 6.kmworld.com — Vertesia introduces AI%C2%A0Assistant for employees to use c
- 7.vktr.com — composable prompts secures 4 million in seed funding led by
- 8.github.com — vertesia
Stay Ahead of the Curve
Weekly enterprise AI insights for technology leaders. No spam, no vendor pitches—unsubscribe anytime.
SubscribeRelated Products
Hatz AI
Multi-tenant secure AI platform that lets managed service providers resell governed AI to small and midsize businesses
Articul8
Domain-specific generative AI platform for regulated and industrial enterprises, deployable on premises or air-gapped
RapidCanvas
Agentic AI platform plus field AI engineers that build and run custom AI apps for mid-market enterprises
Kanu
Enterprise AI platform that learns how your teams work and builds agents and apps inside your own AWS or Google Cloud