Hosted private AI assistant for teams

DocsGPT Cloud is the hosted, multi-tenant DocsGPT at app.docsgpt.cloud, operated by Arc53 and priced per seat. Teams start in minutes; organisations that need a dedicated instance or their own infrastructure move to Managed or On-premises.

Start free Book a demo

$20/seat/month · app.docsgpt.cloud · operated by Arc53

Last updated: 2026-09-19

Deployment posturesoperator · data boundary · LLM

Cloud
operatorArc53
data boundaryArc53 multi-tenant cloudEU region
network boundary
llm locationCloud provider
Managed
operatorArc53
data boundaryDedicated VPSyour region
network boundary
llm locationCloud provider or local engineor your own key (BYOM)
On-premises
operatorCustomer
data boundaryYour network
egress optional (hybrid)
llm locationLocal engineor your cloud LLM (hybrid)
Air-gapped
operatorCustomer
data boundaryYour networkno egress
no external network
llm locationLocal engine only

Text version

Cloud: Arc53 operates a multi-tenant service and the LLM runs at a cloud provider. Managed: Arc53 operates a dedicated instance in your chosen region; the LLM runs at a cloud provider, on a local engine on the instance, or with your own key (BYOM). On-premises: you operate DocsGPT inside your network; the LLM runs locally, or (hybrid) prompts cross the boundary to a cloud provider you choose. Air-gapped: zero external network; LLM, embeddings, vector store and storage all run inside your perimeter.

6+ h → under 5 min

IP application drafting: 6+ hours to under 5 minutes per application, with lawyer review as the final step.

Source: Balt Alnoor · Cloud deployment

18.3k GitHub stars

The hosted service runs the same open-source platform, MIT-licensed, with 226+ contributors.

Source: GitHub

What you get

Everything in the open-source platform, hosted: connectors, cited answers, agents and workflows, teams and analytics.

Knowledge

Connectors and uploads

Google Drive, SharePoint, Confluence, S3 and file uploads with OCR for scanned documents.

google drive · sharepoint · confluence · s3 · pdf · docx · xlsx

Answers

Cited answers

Every answer cites its source document and page; the source panel opens beside the chat.

citations · page numbers · source panel

Agents

Agents and workflows

Classic, Agentic, Research and Workflow agents with tools, schedules and shareable links.

classic · agentic · research · workflow

Teams

Teams and sharing

Workspaces, roles and Viewer / Editor sharing per agent and source.

teams · roles · viewer · editor

Analytics

Usage analytics

Token and cost attribution by agent, conversation volume and adoption.

tokens · cost by agent · conversations

Models

Cloud model set

Frontier models from OpenAI, Google and Anthropic, plus top open-weight models — GLM, Kimi, DeepSeek — provided through Arc53. The set rotates as new models ship. Bring your own model in the registry with your own key and base URL.

openai · google · anthropic · glm · kimi · deepseek · byom

docsgpt — chat

What notice period applies if we terminate the ACME master services agreement early?Termination for convenience needs 90 days' written notice ¹. The 2026 renewal amendment shortens this to 60 days for the professional-services schedule only ². Notice goes to the addresses in clause 21.

$20/seat/month · billed per seat · start free, add seats when the team joins · full pricing →

Where your data goes

What leaves, what stays.

On DocsGPT Cloud, documents and embeddings are stored in Arc53's cloud environment in the EU region; prompts and retrieved passages go to the LLM provider selected for the workspace. Customer data is not used to train models, and deletion is available on request.

Where data goes on DocsGPT Cloud
	Destination	Status
Documents and embeddings	Arc53 cloud storage	EU region
Prompts and retrieved passages	The selected LLM provider	Named on the sub-processor list
Training on customer data	Not used to train models	No — your documents, prompts and answers are never used for training
Deletion on request	Removed from Arc53 storage	Available on request

Every sub-processor, with its region, is listed in Annex III of the DPA.

Security and data handling →

Identity and administration

Sign in with email and password, or with OIDC single sign-on against your identity provider. Each member shows how they were added: manually or through your IdP.

Teams and roles: admin and user roles, Viewer / Editor sharing per agent and source. Usage analytics attribute tokens and cost by agent.

email + password · oidc sso · teams · roles · viewer / editor · analytics

Admin, Security & Analytics →

The Customer Support team: six members, each with how they were added — three added manually, two via SSO, one via SCIM — an admin or member role and a delete button, then six shared resources — two agents, a wiki source and two prompts — each marked Editor or Viewer. Add member opens a dialog, alex.smith@meridianfreight.com is typed with the role member and added, and the list becomes seven members. Back on Teams, three team cards are listed: Customer Support, People Ops and Revenue Ops, with their member and shared-resource counts. — app · Teams › Customer Support

When Cloud is not the right posture

Three cases where a dedicated or self-operated posture is the right call.

Regulated data

If the data is regulated and the platform must be single-tenant, use a dedicated instance.

Managed →

Data-residency rules

If policy requires your own data centre or private cloud, run it on-premises — with a local LLM or a hybrid cloud LLM.

On-premises →

No external network at all

If policy forbids any external network, run it air-gapped with a local model only.

Air-gapped →

Upgrade path to Managed

The same platform runs on a dedicated VPS operated by Arc53, in your chosen region, with your identity provider connected. Sources and agents are exported from Cloud and imported into the Managed instance on request.

A Managed instance is priced as a flat monthly fee rather than per seat: from $2,000/month. Managed dedicated AI instance →

Frequently asked questions

Is my data used for training?: No. Your documents, prompts and answers on DocsGPT Cloud are not used to train models, and Arc53 engages model providers only on terms that prohibit them from doing the same. Data is deleted from Arc53 storage on request. Data handling is on the Security page; every sub-processor is listed in Annex III of the DPA.
Where is my data stored?: In Arc53's cloud environment in the EU; prompts are sent to the model provider you select. Details are on the Security page, and each sub-processor's region is in Annex III of the DPA.
Is DocsGPT Cloud multi-tenant?: Yes. Cloud runs on shared infrastructure operated by Arc53 with per-organisation isolation of data and access. Organisations that need a dedicated instance use Managed.
Can I bring my own model key on Cloud?: Yes. The model registry supports BYOM on the hosted plan: register a custom model with your own key and base URL, alongside the models Arc53 provides.
How do I move to a dedicated instance later?: Book a call; Arc53 provisions a Managed instance and migrates your sources and agents on request.

Start with your documents today

Start free at app.docsgpt.cloud, $20/seat/month when the team joins. Or book a 30-minute demo on your use case.

Start free Book a demo See pricing

What you get

Where your data goes

Identity and administration

When Cloud is not the right posture

Upgrade path to Managed

Frequently asked questions

Is my data used for training?

Where is my data stored?

Is DocsGPT Cloud multi-tenant?

Can I bring my own model key on Cloud?

How do I move to a dedicated instance later?

Start with your documents today