Show the site for
Oh My AIJob helper Beta
Choose a botAsk with

PrivacyNothing you type here is sent to us or saved by us. When you press Go, your question goes to the bot you picked, in a new tab, and that bot’s company handles it under its own privacy policy. Please don’t type personal details.For some bots we also copy your question to your clipboard so you can paste it. How this site handles data

Home · Handbook · Families & differences

Build & rules · AI Policy Handbook

Families & differences

Technology, law, and practice. As of November 3, 2026.

Facts checked November 3, 2026

As of November 3, 2026

Brand names change. Family, data path, and jurisdiction usually decide the governance problem.

Six families public bodies encounter

General assistantsOpen-ended chat over a foundation model. Broad drafting and Q&A; grounding varies by product mode.
Answer enginesRetrieve live sources first, then generate with citations. Built for lookup; citations still need checking.
Suite-embedded assistantsInside email, documents, and calendars, using existing identity permissions. Touches organizational content by design.
Real-time / social toolsWired to live feeds. Useful for situational awareness; weak as an authoritative legal or scientific source.
Task agentsPlan and execute multi-step actions with tools. Highest need for permission boundaries and audit logs.
Self-hosted / private instancesOpen-weight or dedicated instances inside a chosen boundary. More operational work; stronger data-control options.

Features that change risk, independent of brand

Where processing and storage occur

Some tools process prompts only on provider infrastructure in specified regions; others can run on-device or in a customer virtual private cloud. Location determines which governments can compel access under local law. On Jan 31, 2025, the Office of the Texas Governor announced a ban on several foreign-developed apps, including an AI assistant, on state-issued devices and networks.[3]

Whether sources are shown

Citation and RAG features make verification possible. Their presence does not guarantee accuracy; their absence means staff cannot trace a claim to a document without separate research.

Whether inputs are used for training

Consumer terms often allow broader use of prompts for model improvement than enterprise or government terms. Contract language for the specific SKU, not marketing pages alone, is what procurement reviews.

Open-weight versus closed service

Open weights enable self-hosting and offline evaluation. Closed services centralize updates and liability allocation in the vendor relationship. Neither pattern is inherently compliant; controls and contracts decide.

Procurement checkpoint

Data location and compelling authority

A recurring written question from public buyers: in which country are prompts, files, and embeddings stored and processed, and which legal processes can compel the provider to disclose them? Quality and price do not answer that question.

FeatureQuestions commonly recorded in public procurements
Processing locationRegion options? Customer-managed keys? On-device mode?
Citations / RAGCan staff see retrieved passages? Can the corpus be limited to approved sources?
Training on customer dataDefault off for this SKU? Subprocessors? Written residual rights?
Retention & deletionPrompt/log retention period? Deletion on request and at contract end?
Agent permissionsWhich tools can it call? Who approves expansions?