# Small Business AI Technologies

> Independent, source-aware AI technology evaluation guidance for small-business owners, operations leads, and technical advisers.

Canonical: https://smallbusinessaitechnologies.com/
Owner and publisher: Bridgepath AI Solutions
Editorial review: July 24, 2026
Boundary: This publication does not connect to systems, receive business data, test or rank vendors, certify controls, or replace qualified review.
Note: This is a transparent content index, not a vendor directory, product benchmark, crawler-access claim, or endorsement.

## Start and topic
- [Start here](https://smallbusinessaitechnologies.com/start-here)
- [Evaluate an AI technology path](https://smallbusinessaitechnologies.com/topics/ai-technology-evaluation)
- [Free evaluation resources](https://smallbusinessaitechnologies.com/resources)

## Technology knowledge
- [AI technology layer](https://smallbusinessaitechnologies.com/knowledge/ai-technology-layer): An AI technology layer is one responsibility inside the complete system around an AI feature: the business job, user experience, workflow connection, model service, information path, identity and controls, human review, or operating exit.
- [AI delivery pattern](https://smallbusinessaitechnologies.com/knowledge/ai-delivery-pattern): An AI delivery pattern describes how a capability reaches the workflow, such as a person-led assistant, an embedded feature, a controlled automation, a retrieval layer, or a custom application, and therefore what the business must operate and govern.
- [Connected data boundary](https://smallbusinessaitechnologies.com/knowledge/connected-data-boundary): A connected data boundary states which information sources an AI-enabled workflow may reach, which permissions apply, what can leave each system, who can approve changes, what is retained, and how access and records are removed.
- [Vendor evidence record](https://smallbusinessaitechnologies.com/knowledge/vendor-evidence-record): An AI vendor evidence record ties each material claim to the exact product, plan, configuration, source, date, and evidence type, then marks contractual protections, documented controls, demonstrations, internal tests, sales statements, assumptions, and unknowns separately.
- [Representative AI test set](https://smallbusinessaitechnologies.com/knowledge/representative-ai-test-set): A representative AI test set is a reusable group of approved normal, difficult, incomplete, ambiguous, prohibited, and stop-or-escalate examples that candidates face under the same instructions, configuration, reviewer rubric, and decision conditions.
- [Reversible AI pilot](https://smallbusinessaitechnologies.com/knowledge/reversible-ai-pilot): An AI pilot is reversible when it has one bounded workflow slice, approved information, limited access, a working fallback, predefined evidence and stop conditions, a decision date, and a tested closeout path for accounts, connectors, records, data, billing, and support.

## Evaluation guides
- [Map the AI technology layers before shopping](https://smallbusinessaitechnologies.com/guides/map-the-ai-technology-layers-before-shopping): Distinguish the user experience, workflow connection, model service, business data, control plane, and human decision before comparing products.
- [Define the workflow before the AI category](https://smallbusinessaitechnologies.com/guides/define-the-workflow-before-the-ai-category): Turn a vague request for an AI tool into a bounded workflow with a trigger, inputs, review point, accepted output, and fallback.
- [Set a data boundary before the trial](https://smallbusinessaitechnologies.com/guides/set-a-data-boundary-before-the-trial): Decide which information may enter a trial, who can authorize it, what the provider may retain or use, and what must stay out.
- [Compare AI delivery patterns, not just products](https://smallbusinessaitechnologies.com/guides/compare-ai-delivery-patterns-not-just-products): Understand when an assistant, embedded feature, workflow automation, retrieval layer, or custom application changes the operating burden.
- [Request vendor evidence before confidence](https://smallbusinessaitechnologies.com/guides/request-vendor-evidence-before-confidence): Convert feature claims into dated questions about administration, data use, security, change, support, and exit.
- [Build a small representative AI test set](https://smallbusinessaitechnologies.com/guides/build-a-small-representative-ai-test-set): Test a real range of ordinary, difficult, incomplete, ambiguous, and prohibited examples without turning a demo into proof.
- [Run a reversible AI pilot with an exit](https://smallbusinessaitechnologies.com/guides/run-a-reversible-ai-pilot-with-an-exit): Pilot one workflow slice with named owners, approved data, acceptance evidence, support, stop conditions, and a clean removal path.

## Decision kits
- [AI technology layer map](https://smallbusinessaitechnologies.com/decision-kits/ai-technology-layer-map): A one-page view of the experience, workflow, model, data, controls, human review, and provider dependencies.
- [Workflow evidence brief](https://smallbusinessaitechnologies.com/decision-kits/workflow-evidence-brief): A shared baseline for comparing different AI approaches against the same job.
- [Approved-data boundary card](https://smallbusinessaitechnologies.com/decision-kits/approved-data-boundary-card): A plain-language rule for what may and may not enter an AI trial.
- [Vendor evidence request](https://smallbusinessaitechnologies.com/decision-kits/vendor-evidence-request): A consistent request for the facts behind security, privacy, administration, change, and exit claims.
- [Representative test-set sheet](https://smallbusinessaitechnologies.com/decision-kits/representative-test-set-sheet): A reusable set of normal, difficult, ambiguous, unsafe, and prohibited examples.
- [Pilot decision record](https://smallbusinessaitechnologies.com/decision-kits/pilot-decision-record): A visible adopt, revise, extend, stop, or escalate decision with evidence and closeout.

## Primary sources
- [National Institute of Standards and Technology: Artificial Intelligence Risk Management Framework](https://www.nist.gov/itl/ai-risk-management-framework): AI RMF 1.0 is a voluntary framework for governing, mapping, measuring, and managing AI risk throughout the lifecycle. NIST states that the framework is being revised.
- [National Institute of Standards and Technology: Generative Artificial Intelligence Profile](https://nvlpubs.nist.gov/nistpubs/ai/NIST.AI.600-1.pdf): Current cross-sector guidance for risks that are unique to or amplified by generative AI systems.
- [U.S. Small Business Administration: AI for Small Business](https://www.sba.gov/business-guide/manage-your-business/ai-small-business): Plain-language benefits, risks, terminology, and a start-small adoption boundary for small businesses.
- [U.S. Federal Trade Commission: Artificial Intelligence](https://www.ftc.gov/industry/technology/artificial-intelligence): The regulator's current AI cases, policy material, and business guidance for substantiation, privacy, and consumer protection context.
- [Cybersecurity and Infrastructure Security Agency: Secure by Demand Guide](https://www.cisa.gov/sites/default/files/2024-08/SecureByDemandGuide_080624_508c.pdf): Questions buyers can use to assess whether technology suppliers take responsibility for secure product outcomes.
- [Cybersecurity and Infrastructure Security Agency: Small and Medium Businesses](https://www.cisa.gov/audiences/small-and-medium-businesses): Current small-business resources for secure technology selection, authentication, backups, logging, supply-chain risk, and resilience.