Familiar on day one
If your team can use a consumer chat assistant, they can use BisChat. Same rhythm: type a question, read the answer, keep going. No query language, no rollout program, no training week.
LivePrivate AI for teams that have to answer for it
The AI chat your team wants.
The control your business needs.
BisChat gives your team a familiar AI workspace on open models. You control retention and get an exportable audit trail. Enterprise runs in your own AWS account.
AI is already part of daily work. The question now is whether your business can show how it was used.
◆ Workspace live in minutesFull audit logging · SSO roadmapUS-owned · Atlanta, GA
request / governed-workspacerecorded
Request received
s.chen · vendor-agreement.pdf
01Policy check passed
workspace · source · retention
02Open model executed
llama-3.3-70b · zero retention
03Audit record exportable
request · response · model · time
04
Open-weight models onlyRetention you setEvidence ready on demand
Meet BisChat
It looks like the AI chat your team already knows.
It answers to you.
BisChat is the private AI chat workspace inside PrivateStack: the everyday chat window your team opens to ask questions, draft, and search your own documents, with the models, the retention, and the audit trail under your control.
The product, not a diagramReal product capturesDemonstration workspaceNo customer data
The workspace feels familiar before anyone needs a manual.
A focused place to start a conversation, search, take notes, and work with the model selected for the task.
See how private models appear to an administrator
Administrators can see the private model inventory.
The product makes the available model choices visible instead of hiding them behind a generic chat box.
A clean chat window. Ask a question, get an answer, get back to work.
Which models your team can use: Llama, Mistral, DeepSeek and other open models, enabled per workspace.
Answers grounded in your own documents and knowledge bases, not the open internet.
Which sources are connected, and who is allowed to reach them.
Your documents, not the open internet
Point BisChat at the files and knowledge bases your team already works from, so answers come back grounded in your own material instead of whatever a public model happened to absorb.
A private workspace for your team, with their own conversations and files.
How long those conversations are retained before they age out.
The same familiar experience every day, in any browser. Nothing new to learn.
A complete, exportable log of every request. On Enterprise, you also control where the whole thing physically runs: inside your own AWS account.
retention.window = customer_setrequest.logging = completeevidence.export = available
Every conversation is on the record
100% of requests are logged: who asked what, when, and against which model. Export the trail as an evidence pack when an auditor, examiner, or client asks how AI was used.
Hosted
Isolated workspace with disclosed zero-retention inference.
Enterprise BYOC
The full stack runs inside your own AWS account and VPC.
BisChat is what your team sees. PrivateStack is the platform underneath it: on Solo and Team, each team gets its own provisioned workspace, live in minutes; every request is logged and exportable; and hosted inference runs zero-retention through a disclosed subprocessor, so prompts are never stored by the model provider and never used for training. On Enterprise, the whole stack deploys inside your own AWS account: AWS is the only cloud we support for BYOC today. Single sign-on is on the roadmap, and SOC 2 Type II is in progress: ask under NDA and we'll share the current timeline and control set.
Control, with evidence
See what the controls produce.
Inspect the evidence yourself.
The Console separates operational verification from the records you export for review. Both views below use synthetic demonstration data.
Model policy
Open-weight models enabled per workspace
Retention
Zero-retention hosted inference through a disclosed subprocessor
Auditability
100% of requests logged and exportable
Enterprise BYOC
Deployed inside your own AWS account and VPC
Operational verificationSource-derived product viewSynthetic exampleNo customer data
Zero-Retention Verification
All automated retention controls verified
Synthetic verification run
View individual checks and SHA-256 checksum
Verified by automated controls. Independent audit in progress. This is not a certification.
Check the retention control instead of taking it on faith.
The Console source surfaces the latest automated retention verification result, individual checks, and a checksum. This view recreates that current component with synthetic values.
Review data handlingEvidence exportSource-derived product viewSynthetic exampleNo customer data
Compliance Evidence Pack
Date-ranged audit logs, model inventory, retention configuration, and team roster.
Start date
Demo range
End date
Demo range
Generate evidence pack
Evidence supporting a compliance program. Not a certification.
Package the records an evaluator needs to inspect.
The current Console source generates a date-ranged JSON bundle of available audit logs, model inventory, retention configuration, and team roster. It is not a certification.
Open the Trust CenterWhat is live, and what is still underway
- Full request logging and evidence-pack export are live today.
- SOC 2 Type II certification is in progress.
- Enterprise SSO and SIEM export remain roadmap items.
Who it's for
Built for teams that have to answer for it.
The workspace is the same. The evidence burden is not.
Finance
AI your compliance team will actually approve.
Give your team Llama, Mistral, DeepSeek, and other open-weight models in a governed workspace: every prompt logged, zero-retention inference, no training on your data. Enterprise deployments run inside your own AWS account.
Healthcare
Don't let a paste box create your next breach notification.
PHI-grade AI belongs inside your own perimeter. PrivateStack Enterprise deploys Llama, Mistral, DeepSeek, and other open-weight models into your own cloud environment: HIPAA-ready architecture, full audit trail, and PHI that never leaves your VPC.
Legal
AI that never sees your client's matter.
Run open-weight models in a workspace your firm governs: zero-retention inference that never stores or trains on your prompts, and a complete audit trail of who asked what, when. Enterprise: deployed inside your firm's own cloud tenant, where matter data never crosses your perimeter.
Manufacturing
Keep your engineering data out of someone else's AI.
PrivateStack gives your team open AI models in a governed workspace: your specs, supplier pricing, and process data are never retained by a model provider and never used for training. Enterprise deployments run inside your own AWS account.
Government
AI your inspector general could audit.
Public-sector and federal-adjacent teams need AI that answers for itself: every request logged and exportable, a US-owned vendor on the record, and a deployment path that stays inside your own perimeter. PrivateStack is priced at a flat per-seat rate procurement can actually plan around.
Pricing
Start solo. Scale with us.
Team monthlyTeam yearlyTeam saves 20% with annual billing
Direct signup
Solo / Entrepreneur
One seat, one person
A private workspace for one person. Not a smaller version of Team.
$49/ month
Valid invite code required at checkout
Redeem your invite- One private BisChat workspace
- Unlimited messages
- 5 knowledge bases
- 25GB storage
Scope limits
- 1 seat, no additional members
- No API access
- No custom models
Need seats for other people? See Team.
Team
Up to 9 seats
$99/ user / mo
or $79 / user / mo billed annually, sales-assisted
- BisChat workspace for every seat
- Every model, unlimited messages
- Full API access, 20 knowledge bases
- 100GB storage
- SSO (Okta / Google / Azure), coming soon
- Team collaboration tools
- Custom branding and dedicated support
Need more than 9 seats? See Enterprise.
Enterprise
Tailored to your needs
Custom
Volume and dedicated infrastructure
- Everything in Team, plus:
- Private networking (PrivateLink / VPC peering)
- Custom data-residency commitments
- SIEM export, scoped as an Enterprise roadmap item
- Security questionnaire and BAA support
- Dedicated infrastructure and custom availability targets
- On-premise deployment option
- 24/7 dedicated support
Deploy your private AI.
Keep control. Keep your budget. Open models in a workspace you govern, with the logs to prove it.
Solo / Entrepreneur is $49/month for one seat. Valid invite code required at checkout.
Not ready for a call? Download the AI Governance Checklist.
A two-page, self-scored checklist for regulated teams evaluating private AI.