Built in your tenant
The workspace is built in your own Microsoft 365 tenant. The audit records Microsoft 365 makes available for it are retained with your firm records under the policy your CCO approves.
AI Workspace, Level 1
A governed AI workspace built in your Microsoft 365 tenant
Your advisors use AI in a workspace built in your own Microsoft 365 tenant. Microsoft Purview audit and retention are configured and tested before go-live. You choose Claude or GPT behind it.
For SEC-registered RIAs.
See how it works
Governed AI for an RIA: From the Policy to the Platform. About 8 minutes.
Open the video in a new tabWhy a policy alone is not enough
Advisors already use ChatGPT, Claude and other AI tools, often on personal logins. A written policy alone may not give advisors an approved place to work, technical controls or a record of what happens there.
Read why a written AI policy does not stop a personal loginWhat Level 1 includes
Your choice of model
Claude or GPT runs behind the same governed workspace. You choose when we build it.
Purview audit and retention, configured and tested
We configure Microsoft Purview audit and retention for the workspace and test that its records can be found before go-live. The results are listed in your Workspace Build Record.
Access follows your permissions
The assistant reads only what each advisor can already open in SharePoint. It is shared only with the advisors group you name.
Daily export to an immutable archive
We configure a daily export of audit records to an immutable archive and test the export before go-live. Your Workspace Build Record lists what was tested.
Data loss prevention, tested
Data loss prevention settings, including sensitivity labels on SharePoint content, are tested before the workspace goes live.
Your AI policy, matched to real use
We review your existing AI policy and suggest corrections so it describes how AI is actually used at your firm. You own the policy and adopt any change.
With Claude as the model, Anthropic processes prompts as a Microsoft subprocessor. See Know before you start.
Your tenant, not a vendor's app
Where the record lives decides who controls it.
| AI chat in a vendor's app | Vendor app plus an archive tool | Dakona AI Workspace | |
|---|---|---|---|
| Where the record is kept | The vendor's service | The vendor's service, copied to an archive | Your Microsoft 365 tenant |
| Who sets retention | Vendor settings | Vendor and archive settings | Your Microsoft Purview policies |
| Who controls access | The vendor's admin console | The vendor's console and the archive | Your Microsoft 365 permissions. Removing access can take up to about two hours. |
General comparison of approaches, not a review of any named product.
With Claude as the model, Anthropic processes prompts as a Microsoft subprocessor. See Know before you start.
How Level 1 is delivered
- 1
Review
We review your Microsoft 365 tenant and your current AI policy.
- 2
Build
We build the workspace and its governance in your tenant.
- 3
Test
Your own advisors test it with real questions before go-live.
- 4
Turn over
You receive a Workspace Build Record: what was configured, what was tested and the known limits. Ongoing support is a monthly managed service.
What Level 1 does not include
- Level 1 covers firm knowledge only: your policies, procedures and standards. Client, custodian and CRM data are not part of Level 1.
- We do not write your AI policy. You own it.
- Configuration backup and restore testing, model change management and retrieval drills are available by change order.
Know before you start
- Claude is supplied by Anthropic, a Microsoft subprocessor. Microsoft's Product Terms and Data Protection Addendum apply to the Claude models we enable. Microsoft does not state where Anthropic processes prompts on this route.
- Whether Claude responses are captured in Microsoft Purview is confirmed for each deployment and recorded in your Workspace Build Record.
- We do not rely on Microsoft's SOC reports for Anthropic's processing. We turn off the Anthropic models that require the provider to retain your data.
- A GPT model is available in the same workspace if you prefer it.
- Retention policies can take up to seven days to show as applied in Microsoft 365.
- Removing an advisor's access can take up to about two hours to take effect.
- You need Microsoft 365 Business Premium or higher and a Copilot licence for each advisor who uses the workspace.
- Dakona provides technology services. Your CCO decides how the workspace fits your compliance program.
Is this a fit?
To use AI Workspace Level 1 you need Microsoft 365 Business Premium or higher and a Copilot licence for each advisor. We confirm this on the discovery call.
Pricing is scoped on a discovery call. Tell us about your firm and we will follow up.
Request a Discovery CallQuestions
Your files and your audit record are stored in your Microsoft 365 tenant. When Claude is the model, Anthropic processes prompts and responses as a Microsoft subprocessor. A GPT model is available in the same workspace.
Claude or GPT. You choose when we build it.
We configure Microsoft Purview audit and retention for the workspace and test that its records can be found before go-live. Whether Claude responses are captured is confirmed for each deployment. Results and known limits are listed in your Workspace Build Record. What is retained and for how long follows the retention policy your CCO approves.
Yes. Each advisor who uses the workspace needs a Microsoft Copilot licence, in addition to Microsoft 365 Business Premium or higher.
No. We review your existing policy and suggest corrections so it matches how AI is used. You own the policy and adopt any change.
The AI Exam File is a fixed-fee review that checks your AI policy against your infrastructure. The AI Workspace is the governed place your advisors work. Firms can start with either.
A written record we give you at turnover. It lists what was configured, what was tested, the results and the known limits.