Agentic Index

AI agent platforms that can be self hosted or run in a private VPC (2026)

Of the 566 agentic AI platforms in the Agentic Index, 188 document deployment inside the customer's own boundary in full, covering self hosted, on premises, air gapped and private VPC options. Only 82 also document bringing your own model or routing between providers. That is 14.5% of the field. This is a bar, not a leaderboard: a platform either documents both in its own public materials or it does not clear, and Partial evidence on either does not count.

The reason the bar has two parts rather than one is the finding underneath the list. Self hosting the software is not the same as containing the data. 106 of the 188 platforms you can run inside your own boundary do not document bring your own model. On those the runtime sits in your VPC and the model call still leaves it, so the prompt, the context and whatever the agent retrieved go to a third party anyway. A buyer who asks only the deployment question gets the wrong half of the answer more than half the time.

Which makes the scarcity ordering the second thing worth carrying out of this page. Model flexibility is the hardest enterprise capability to find in agentic AI, not deployment. Across the same 566 platforms, full coverage runs observability 319, security and identity 277, deployment and residency 188, and model flexibility and routing 135. It is also the binding constraint on this bar: 106 of the 159 platforms one capability short fail on model flexibility alone. Separately, 53 platforms will take your model but will not run in your environment, which is the right answer for a team whose constraint is which model sees the data rather than where the software runs.

The bar, and how the 566 platforms score against it

Capability	What has to be documented	Full	Only blocker
Deployment and data residency	the platform documents running inside the customer's own boundary, self hosted, on premises, air gapped or in a private VPC, or documents where the data physically sits when it cannot	188 (33%)	53
Model flexibility and routing	the platform documents bringing your own model or key, or routing between providers, including local or self hosted models, rather than only calling one vendor hosted model	135 (24%)	106

Full means the vendor publishes evidence meeting the capability in its own public materials, under the Agentic Index verification standard. Only blocker counts platforms that document the other capability in full and fail on this one alone. One honest limit on the first row: deployment and data residency grades a spectrum, from customer controlled self hosting down to documented regional residency inside the vendor's own cloud. Two platforms below clear it on the residency end while their own evidence states that customer self hosting is not the model, and both are named where they appear.

Clears the bar and scores 12.5 or higher of 14 overall

These 12 platforms document deployment inside your boundary and model flexibility in full, and also sit at the top of the Agentic Index coverage score across all 14 capabilities. Ordered by total coverage, ties broken alphabetically.

1. 1.UiPath

   13.5 / 14 capabilities

   enterprise operations agents, multi-agent platforms, agent builders

   Agentic automation orchestrating agents, robots and people end to end. Deployable cloud, on premises or hybrid, with Automation Suite to run the full platform yourself, and Autopilot runs on the customer's cloud of choice or their own infrastructure while supporting GPT, Gemini, Claude and any compatible model.

2. 2.Automation Anywhere

   13.0 / 14 capabilities

   enterprise operations agents, multi-agent platforms, agent builders

   Agentic process automation pairing goal driven agents with RPA bots. Deploys as cloud SaaS and inside a customer virtual private cloud for full data control, with on premise heritage. Deliberately model agnostic, connecting to Amazon Bedrock, Google Vertex AI, Azure OpenAI and OpenAI rather than hosting its own.

3. 3.CrewAI

   13.0 / 14 capabilities

   multi-agent platforms

   Open source multi agent orchestration for collaborative crews. Worth reading the evidence rather than the position: both rows behind this entry rest on the capability research report rather than a named vendor documentation page, so the grade is thinner than the enterprise entries beside it even though the score is the same.

4. 4.Salesforce

   13.0 / 14 capabilities

   enterprise operations agents

   Residency, not self hosting. Enterprise cloud with a trust layer, data residency options and zero copy data access. Salesforce's own filed evidence states customer self hosting is not the model.

   CRM platform whose Agentforce layer runs agents across sales, service and marketing, with the Atlas Reasoning Engine supporting OpenAI, Anthropic and Google Gemini. It clears this bar on data residency and the trust layer, not on self hosting, and its own filed evidence says so. Read the disclosure beside it before treating it as an answer to this question.

5. 5.Agno

   12.5 / 14 capabilities

   agent infrastructure platforms

   High performance agent runtime, formerly Phidata, Apache 2.0 licensed, which is the strongest possible containment position because the runtime is yours. Same evidence caveat as CrewAI: both rows cite the expansion evidence report rather than named documentation pages, a thinner basis than the enterprise platforms carry.

6. 6.Boomi

   12.5 / 14 capabilities

   enterprise operations agents, multi-agent platforms, agent builders

   Integration platform with an agent control tower over enterprise systems. Runs on a flexible multi cloud autonomous runtime with on premise and multi region deployment options for residency, and is model agnostic with MCP support across Amazon Bedrock, Snowflake Cortex and OpenAI, with roadmap support for custom large and small models.

7. 7.Coworker

   12.5 / 14 capabilities

   enterprise operations agents, agent builders

   Enterprise agent platform built on an organizational memory knowledge graph. The most complete containment story in the headline tier: cloud, private cloud, on premises or air gapped deployment with VPC peering, and intelligent routing that scores each task across closed, open source and self hosted models including Llama and Mistral.

8. 8.CrowdStrike

   12.5 / 14 capabilities

   multi-agent platforms, agent builders (secondary lane membership, primary category Security / SOC agent)

   Residency, not self hosting. Cloud native Falcon platform with a single agent architecture and regional cloud options. CrowdStrike's own filed evidence states customer self hosting is not the model.

   Security platform whose Charlotte AI runs agentic detection, triage and response, with AgentWorks offering model optionality across Anthropic, OpenAI and NVIDIA. Like Salesforce it clears on regional cloud options rather than self hosting, and its own filed evidence states that. It also qualifies here on secondary lane membership rather than as a primary platform.

9. 9.Pydantic AI

   12.5 / 14 capabilities

   agent infrastructure platforms

   Open source MIT licensed Python agent framework. The cleanest containment shape on the page: the runtime runs wherever you run Python with no vendor infrastructure in the path, so there is no residency question at the agent layer at all, and it is compatible with virtually every LLM provider with a documented path for custom models.

10. 10.Rasa

    12.5 / 14 capabilities

    multi-agent platforms

    Open source conversational AI with an enterprise framework, self hosted or SaaS, and multi model support. The grade rests on the open source deployment and multi model documentation, so read it as a containment posture backed by code you can run rather than a configurable enterprise residency product.

11. 11.Sim

    12.5 / 14 capabilities

    multi-agent platforms, agent builders

    Open source Apache 2.0 agent workspace. The sharpest single instance on this page: documented Docker Compose and Kubernetes self hosting including a fully offline setup with local models through Ollama, bring your own key at base provider pricing with no markup, and enterprise permission groups that restrict which providers are allowed.

12. 12.SimplAI

    12.5 / 14 capabilities

    multi-agent platforms, agent builders, agent infrastructure platforms

    Enterprise agentic AI operating system spanning SimplAI cloud, customer cloud, private cloud, VPC, on premises and air gapped with no architectural change between them, plus customer defined retention and deletion per agent. One caveat a buyer should carry: bring your own models and fine tuning are documented but gated to the enterprise tier.

The remaining 70 platforms that clear the bar

Every one of these documents deployment inside the customer's boundary and model flexibility in full. They score below 12.5 of 14 on total coverage, which says something about breadth across the whole taxonomy rather than about how well contained they are. Several of the strongest containment stories in the index are narrow open source runtimes that score low precisely because they do one thing.

Common questions

Which AI agent platforms can be self hosted or run in a private VPC?

188 of 566 agentic AI platforms in the Agentic Index document deployment inside the customer's own boundary in full, covering self hosted, on premises, air gapped and private VPC options. Only 82 of those also document bring your own model or routing between providers in full, which is the bar used on this page, because a runtime inside your boundary that can only call one vendor hosted model has not contained your data. That is 14.5% of the field. Graded from public evidence only.

Is self hosting an AI agent platform the same as keeping your data private?

No, and this is the gap the page exists to show. 106 of the 188 platforms documenting deployment in your own boundary do not document bring your own model. On those platforms the software runs inside your perimeter and the model call still leaves it, so the prompt, the context and whatever the agent retrieved go to a third party anyway. Self hosting the runtime and containing the data are two different purchases.

What is the hardest enterprise capability to find in agentic AI?

Model flexibility. Across the same 566 platforms, full coverage runs observability 319, security and identity 277, deployment and residency 188, and model flexibility and routing 135. It is the scarcest of the four, and it is the binding constraint here: 106 of the 159 platforms one capability short of the bar fail on model flexibility alone.

Can you point an AI agent platform at your own model without self hosting it?

On 53 platforms, yes. They document bringing your own model or routing between providers while running in the vendor's cloud rather than yours. That is a real option for teams whose constraint is which model sees the data rather than where the software runs, and it is a different shape of answer from the 82 platforms that document both.

Does every platform on this list actually let you self host?

Two do not, and the page names them. The deployment and data residency capability grades a spectrum, from customer controlled self hosting down to documented regional residency inside the vendor's own cloud. Two cleared platforms have filed evidence stating in terms that customer self hosting is not the model. They stay in the list because the rule is mechanical and applied to every vendor equally, and the disclosure sits beside their entries so nobody reads them as self hostable.

Is this ranking paid or sponsored?

No. No vendor pays for placement, no vendor has reviewed this page, and every grade comes from the vendor's own public materials under the Agentic Index verification standard. 988 vendors are graded against the same 14 capabilities. Data last verified August 22, 2026.

Method: membership is the same 566 platform pool used by the best agentic AI platforms in 2026, drawn from 988 researched vendors. The editorial pages over this pool are one method with different bars, not several opinions. Every grade comes from the vendor's own public materials under the Agentic Index verification standard. No vendor pays for placement and no vendor has reviewed this page. Data last verified August 22, 2026. How this evidence is graded

Related: enterprise security and compliance platforms, platforms that require human approval, agent observability platforms, how every vendor scores on deployment and data residency, how every vendor scores on model flexibility, compare platforms side by side.