34%
of professional-services workers report using unsanctioned AI tools
Thomson Reuters Institute · 2026Private AI for insurance agencies
Private AI for insurance agencies: carrier and policyholder data stays yours.
Submissions, summaries, and client communications drafted by your own AI, never shared with a public AI provider.
Choose Cloud if your agency does not want to house equipment, or In-Building inside your building. The guarantee applies only to professional firms of 10 to 75 professionals on the In-Building track, with a written time-recovery target and participation requirements. Cloud has no guarantee.
30 minutes · No preparation · No client data
Private AI inside your walls2:04
Captions are included in the video.
The problem
Policyholder information needs an approved boundary.
33%
name data privacy and security the top barrier, up from 23%
Bluevine 2026 Business Owner Success Survey23%
have had a security or privacy issue since adopting generative AI
Xero Future Focus AI researchYour work
Start with the documents your agency already uses.
Submission and proposal drafting
Prepare submissions and proposals from your approved templates and carrier requirements. Your producer reviews the draft before it goes to a client or carrier.
Designed outcomeLess time assembling a first draft
Policy and claims summaries
Turn dense policy and claims documents into client-ready explanations. Your team checks each summary against the source documents before it is used.
Designed outcomeLess re-reading before review
Client communications
Draft renewals, requests, and explanations in your agency’s voice. Your team reviews every word before sending.
Designed outcomeLess inbox drag
Why private AI
Private AI, not public AI. Six reasons.
Your data stays yours
Nothing your team types, uploads, or asks ever reaches a public AI provider. It lives on your machine, in the United States.
Trained on your business
USA-sourced open-weight models, trained on your documents, your process, and your voice, so the answers are yours, not generic.
Nobody else's model, nobody else's prompts
Your AI runs for your company alone. No shared model, no one else's data in it, no one else's questions on it.
No per-seat meter
One private AI for the whole team. Add people, not subscriptions.
Terms that cannot change on you
No policy update from a provider can change where your data lives or what it can be used for.
Built and managed for you
We set it up, train it on your business, host it, secure it, and keep it current. You use it.
How it works
From first conversation to everyday use.
01
Fit call
30 minutes. No preparation. No slides. No client data.
02
Assessment
Two weeks, remote, with a written report.
03
Implementation or installation
The assessment scopes Cloud implementation or In-Building installation.
04
Managed for you
Updates, security, support, and new use cases. Quoted in your assessment.
How we handle wrong answers
What happens when your AI is wrong.
It will be, sometimes. Here is how it is built so that being wrong is cheap to catch and never reaches a client.
Every answer shows its source
The passage comes from your document. Your retrieval system pulls it from the actual file. Your model does not write the source passage.
It says 'I don't know'
When your documents do not contain the answer, your AI says so instead of guessing.
A person approves anything that leaves
Nothing goes to a client without approval from your named reviewer.
You get your own number
The first ninety days record how often drafts are accepted without correction, workflow by workflow. You decide where to rely on your AI.
Your people can talk to it. Only administrators can change it. Every question, answer, and change is logged.
Your AI can only see what you can already open. Permissions come from your systems, not from the AI.
It learns how your firm writes. It looks up what your firm knows, one question at a time, only from documents the person asking is allowed to open.
The 90-day Absorption Method
A written target. Measured in hours returned.
For CPA, law, insurance, and wealth management firms of 10 to 75 professionals on the In-Building track. We agree to a written 90-day time-recovery target. If the target is not met and participation requirements were satisfied, we keep working at no additional charge. Cloud has no guarantee.
Baseline
Measure three workflows and agree to the time-recovery target in writing.
Adopt
Train champions, redesign the approved workflows, and address friction each week.
Review
Use the dashboard to compare hours returned with the baseline. Your professionals review the work.
Participation requirements and the target are written into the engagement. The professional-firm program explains the framework.
Map the rules that apply to your agency.
State insurance data security laws may be modeled on the NAIC Insurance Data Security Model Law (#668). Applicable requirements depend on state enactments, your agency’s domicile, licenses, activities, and any exemptions. The model itself is not a single nationwide law.
The NAIC Model Bulletin on the Use of Artificial Intelligence Systems by Insurers, adopted in December 2023, addresses insurers’ AI governance. It is directed at insurers. Your agency and its advisers should review applicable state action and carrier requirements rather than assume the bulletin applies uniformly to every agency.
Gramm-Leach-Bliley safeguarding obligations also need to be addressed where they apply to your agency. The assessment maps the relevant jurisdictions, data flows, and technical controls for review with your advisers.
NVIDIA and Apple hardware hosted in the United States. USA-sourced open-weight models, trained on your documents, your process, and your voice.
PrivateStride provides technical and implementation support, not legal advice or a certification of compliance.
The first step
Start with an assessment. In writing.
AI Assessment
Cloud-Hosted Private AI Assessment: $4,500
Two weeks · Done remotely · Written report
Your own dedicated private AI, hosted in a US data center we contract. For entrepreneurs, founders, and executives who want private AI without owning or housing equipment.
The report quotes implementation and monthly service.
Credited in full toward implementation within 90 days.
Book a fit callAI Assessment
In-Building Private AI Assessment: $9,500
Two weeks · Done remotely · Written report
For companies and firms that want the equipment inside their walls. NVIDIA and Apple hardware hosted in the United States.
The report quotes installation and Managed AI by quarter. It also covers site and network review, equipment placement and power, the installation plan, and training after installation.
Credited in full toward installation within 90 days.
Book a fit callBoth assessments are done remotely, by video and phone. Nobody travels.
Cloud has no guarantee. The written 90-day time-recovery target applies only to eligible professional firms of 10 to 75 professionals on the In-Building track. If the target is not met and participation requirements were satisfied, we keep working at no additional charge.
Most private AI implementations cost less than a third of the fully loaded cost of your next hire.
This is for
- Independent insurance agencies and brokerages ready to choose Cloud-hosted or in-building.
- Partners ready to back a better workflow with real leadership.
- Leaders willing to assign champions and review the work.
This is not for
- Not an unsupervised coverage or claims decision system
- Not consumer AI with a policy memo around it
- Not for leaders unwilling to assign champions
Your questions, answered.
Can our producers use ChatGPT with policyholder data?
Only for a defined workflow and data class your agency has approved. Review the exact service, account, data flows, access, retention, and terms with your advisers before entering policyholder information. A personal account or product label is not approval to use that data.
Which state laws apply?
It depends on your agency’s domicile and the states you are licensed in, as well as your activities and any applicable exemptions. The assessment maps the relevant jurisdictions and requirements for your agency and advisers to review. It does not replace a legal determination of which laws apply.
Cloud-hosted or in-building for an agency?
Choose Cloud if you do not want to house equipment. Your private AI runs for your agency alone in a US data center. Choose In-Building if you want it inside your building. The fit call recommends which fits. The written 90-day time-recovery guarantee applies only to eligible professional firms of 10 to 75 professionals on the In-Building track. Cloud has no guarantee.
Does it work with our agency management system?
Integration with your agency management system is scoped in the assessment. We review approved data exports, access, permissions, and integration data flows before implementation. The written report defines the work and quote.
Who reviews what goes to a client or carrier?
Your producers and designated reviewers remain responsible for the work. Outputs are drafts to check against source documents and carrier requirements before sending. Private AI is not an unsupervised coverage or claims decision system.
How long until we are running?
Start with a 30-minute fit call. Either assessment takes two weeks, remotely, and produces a written report. The report scopes Cloud implementation or In-Building installation. The full 90-day Absorption Method is the in-building firm program. No sandbox, trial, or demo is provided. No AI system is delivered until implementation is paid.
Who stands behind the work
DJLRC, Inc.
An Arizona corporation, d/b/a PrivateStride.
6751 N Sunset Blvd, Suite 320Glendale, AZ 85305
+1 (212) 960-8293
partners@privatestride.com
David Le Rouse (DJ)
Leads fit and assessment. BS Accountancy and MBA, Arizona State. Thirty years across technology, accounting, and strategy consulting.
Alex
Leads the technical review.
Book a fit call
Thirty minutes. No preparation. We never need your data to have the conversation.
Tell us about your company. We will discuss Cloud-hosted or in-building and the assessment that fits.
Book a fit call 30 minutes · No preparation · No client data