New noteWhat changes on January 1, 2027
Solidare AI — Technology built on solidarity

Industries · District managers, school business officials, department heads, clerks and elected boards.

Your ChatGPT prompts are public records. Here is the retention schedule and the policy that keeps you clean.

No one in government wants to be first; they want to be second. Records requests in California and Washington have already asked for AI drafts. The policy has to fit on a staff report and clear the consent agenda.

WHAT YOU NEED TO HAVE READY

  • An AI use policy that fits on a staff report
  • A records-retention line covering prompts and outputs
  • The procurement file: under the threshold, a cooperative contract, or a sole-source justification
  • Who reviewed the AI output, for the auditor
  • An enterprise-account-only rule for student and resident data

WHO WILL ASK FOR IT

  • County counsel or the city attorney
  • The clerk, as custodian of records
  • The auditor
  • The council or board
  • A resident at the podium

HERE IT IS IN 3 WEEKS

AI Use Policy Sprint

3 weeks

A policy on a staff report, a retention line for prompts and outputs, and what to say at the podium when a resident asks whether AI wrote the report.

Start with the AI Use Policy Sprint

What you hold

  • Resident and student records
  • Public records and correspondence
  • Procurement and contract files
  • Employee and payroll data
  • Meeting records and public comments

What keeps you up at night

  • AI outputs that become public records nobody planned for
  • Procurement rules and data-residency requirements vendors don't mention
  • Employee decisions informed by AI under new state rules
  • A meeting-notes bot in closed session
  • A constituent chat assistant that answers confidently and wrongly
  • A vendor that isn't here in three years

Where AI helps first

  • Assistant over ordinances, policies and procedures for staff
  • Public-comment summaries and correspondence drafting with review
  • Records-request triage agents with approval gates
  • Board and council briefings, a resolution, and disclosure language
  • Private AI sized to residency and records rules

How we usually start

Three ways in for public sector.

012–3 weeks

AI Readiness Review

Interviews with the people who do the work, a review across six areas (strategy, data, people, security, governance, workflow), and a written read on what's safe to start now.

  • A readiness scorecard across the six areas
  • Ranked list of use cases by hours saved and risk carried
  • One-page brief for your board, trustees or leadership
What you get →
023 weeks

AI Use Policy Sprint

A policy written for the people who will follow it, with the structure behind it to satisfy a regulator, an auditor or a trustee.

  • AI use policy and one-page do's-and-don'ts card
  • Approval path and intake form for new tools
  • Human-review rule for consequential decisions
What you get →
033–4 weeks

Privacy & Security Review

Find where data is already flowing into AI tools, review who handles it, and fix the biggest exposures first.

  • Data map and shadow-AI inventory
  • Vendor data-handling review with scorecards
  • Private-versus-public decision by class of data
What you get →

Reading for public sector

Law · September 1, 2026

What changes on January 1, 2027

Colorado and California both switch on rules for AI in consequential decisions. If AI touches hiring, discipline, benefits or eligibility, here is what you owe people, and a 90-day way to get ready.

Read the note · 3 min

Private AI · July 28, 2026

When a private AI is the right call, and when it isn't

A model running in your own environment is not a luxury anymore. Here is what “private” and “local” mean in plain language, what it changes, and the one rule for deciding.

Read the note · 3 min

The standard we work to

  • Data that names a person stays inside your environment
  • A person approves anything that touches a person, a record or a dollar
  • Every AI use disclosed in plain language to the people it affects
  • Aligned to NIST AI RMF and designed to ISO/IEC 42001 structure; threat-modeled to OWASP and MITRE ATLAS
  • You own what we build and can leave with it
The seven commitments in full →

Next step

Tell us what you hold and who you serve.

We'll tell you what AI should and shouldn't touch, and what to do first — in one conversation.

NO SALES DECK · NO OBLIGATION · ASK@SOLIDARE.AI