ASSEMBLE TEAMS INC. · RESPONSIBLE AI
We use AI to support discovery, research, analysis and engineering. It should make human work more effective without obscuring whose decisions matter.
This page distinguishes our current internal uses, public-site workflows and ideas still in development. It describes principles and requirements, not an independent AI safety certification.
Public practice overview · Reviewed 28 October 2026
OUR POSITION
AI can assist a process. It does not own the outcome.
People remain responsible for product direction, public claims, source verification, mission participation, commercial decisions and appropriate governance. We assess AI by who it helps, who may be harmed and whether an error can be corrected.
CURRENT USE & PRODUCT STATES
Internal AI-assisted work must not be confused with an autonomous product feature. A prototype or research document is not evidence of a functioning public service.
We use AI to help draft, analyze, explore implementation options, identify potential errors and conduct internal research. Human review and evidence checks are required before consequential public claims or release decisions.
Open Directory submissions and listings are not automatically certified or assigned a universal AI trust score. Listing review and publication are human-governed. A public listing is not an endorsement or a guarantee of a match.
Matching, multi-model routing, mission agents, scoring and delegated workflow concepts are under research or development. They are not assumed to be available, validated or authorized for consequential decisions.
For formal capability disclosures we distinguish: Available now, Research Preview, Beta, In development, Planned and Retired. Specific system and architecture changes are dated in AI Updates ↗.
GOVERNANCE PRINCIPLES
Our responsible AI requirements define the conditions for using or releasing AI-supported functionality.
AI must not independently approve payments, sign agreements, decide hiring outcomes, issue legal/compliance determinations or publish authoritative Benchmark rankings.
Research claims require source checks. AI-generated output can be mistaken or biased; material decisions need review paths, correction mechanisms and clear limitations.
Separate public-source research from private submissions. Limit data access, avoid unnecessary sensitive information and assess contractual and security controls before proposed user-facing AI processing.
Consider bias, unequal access, geographic or linguistic disadvantages, misuse and whether people can challenge an AI-supported output. Planned controls are not presented as completed audits.
EVALUATION & OPENNESS
A future AI feature should move through documented testing, human review, privacy/security assessment and clearly labeled release states—not through unverified capability claims.
Record the system's purpose, affected users, data categories, permitted outputs, human approval, risks and known limitations.
Evaluate accuracy, misuse, fairness, privacy and security risks appropriate to the use case. Human reviewers determine whether a feature is ready.
Update the relevant disclosure when model providers, processing purposes, decision authority, data handling or user-facing behavior materially changes.
PUBLIC INFORMATION
Read the policy, check the development record and contact us about possible mistakes or concerns. The presence of a policy does not mean every future system has passed an audit.
Read dated public notices about AI research, systems in development, governance and changes to authority.
Explore updates ↗ DATA HANDLINGHow public-site information is collected, used, corrected and removed; separate permissions for public information.
Read the policy ↗ RELEASE TRUTHWebsite releases and product-stage notes, distinguishing public capabilities from development plans.
View releases ↗ FEEDBACK & ACCOUNTABILITYSubmit questions, report an issue or request a correction to public information or disclosures.
Get in touch ↗Our governance approach is informed by established frameworks, including the NIST AI Risk Management Framework. We do not claim independent certification under NIST, ISO/IEC 42001 or another AI management standard.