Fide AI · Capability and engagement brief
Evidence for your next AI decision.
Fide AI investigates AI model and agent behavior so teams can make a specific training, release, or deployment decision with evidence they can inspect.
We welcome scoped work with frontier-lab assessment teams, independent evaluators, assurance practices, and product teams. Bring a model comparison, an agent workflow, or an evaluation that needs scrutiny.
Technical contributions
Model and system comparison
Compare agreed model versions, prompts, configurations or mitigations under a consistent protocol. Investigate regressions, adversarial behavior, and usefulness tradeoffs, keeping development examples separate from held-out tests.
Best fit: A training, mitigation, checkpoint, or release comparison with a defined behavioral question.
Required inputs: Agreed model versions or endpoints, intended use, and permitted test access. Training data or checkpoints only where available and authorized.
Typical outputs: Comparison protocol, traceable results, failure analysis, uncertainty, and a briefing on the implications for your decision.
Assistant and agent assessment
Test a defined assistant or agent workflow for faithful source use, appropriate responses, permissions, delegation, and truthful action reporting. Distinguish system behavior from what safeguards block, and examine whether people can intervene effectively.
Best fit: One assistant or agent workflow where source fidelity, response boundaries, tools, permissions, or human approval matter.
Required inputs: The intended use, relevant sources and criteria, authority boundaries, a safe test setting, and permitted response or action records.
Typical outputs: Workflow and authority map, evidence-backed findings, remediation priorities, and targeted retesting where commissioned.
Evaluation and grader validity review
Examine whether an evaluation measures what matters and whether its judgments are reliable. Review rubrics, reference judgments, grader calibration, coverage, and false positives or negatives, including the evidence behind benchmark results and performance claims.
Best fit: A benchmark, rubric, automated grader, or evaluation method informing training, selection, procurement, release, or deployment.
Required inputs: The evaluation or grader, the decision it informs, and available rubrics, cases, outputs, scoring rules, and reference judgments. An initial inquiry needs only a non-sensitive description.
Typical outputs: A measurement review memo, grader calibration findings where relevant, reproducible checks where permitted, and recommendations for using or improving the evaluation signal.
What you receive
Depending on scope, you receive a documented protocol, findings with supporting examples, uncertainty and limitations, practical recommendations and a briefing for decision makers. Rerun materials and targeted retesting depend on agreed access and licensing.
These are research contributions with stated limits, rather than client references or completed frontier-lab audits.
- DSEWiki incident-report investigation — Assistant-coded claim assessments await independent human adjudication. This is incident-report research, not a frontier-lab safety audit.
- URLQuery boundary investigation — The original discovery was Transluce’s. Fide’s contribution is reanalysis of the retained sequence; it does not establish every downstream effect.
- Source Delegation and Quotation Fidelity — Results apply to the named Scripture-quotation conditions. They do not establish performance in another domain or a general safety guarantee.
Working arrangements
Agree the decision, users, versions, criteria, access, expert review, deliverables, timing and fee. A project may compare versions, investigate a workflow or review an evaluation; extensions and recurring cycles have their own agreed scope.
Before work begins, Fide and the client agree on the technical and domain expertise needed and who will be responsible for testing and review. Depending on the scope and availability, qualified collaborators from Fide’s network may fill those roles.
We agree the audience, criteria, Fide’s role and reporting rights before testing. Private findings stay with agreed recipients; restricted work has explicit reporting limits. Public independent assessments require advance publication rights without participant approval, subject to defined protections.
Fide retains technical control within the agreed scope. Fees never depend on favorable findings. Data access grants no automatic publication, research, aggregation or training rights. Reports state evidence and limits; they confer no certification or blanket deployment approval. Fide does not currently offer audits.
Fide’s own checks of what it builds are development validation, never an independent opinion; independent assessment of Fide-built solutions comes only from qualified outside assessors. Our assessment of customer work we materially helped design or implement is collaborative validation, never an independent opinion. A different internal reviewer or disclosure alone does not establish independence.
Evaluation independence and reporting →Start a conversation
Share a non-sensitive description of the system, decision, available access and timing. We confirm fit before committing to a scope.
Start with a non-sensitive description. Do not send credentials, confidential logs, or unreleased model details through the inquiry form or ordinary email. We agree an appropriate transfer method before receiving sensitive material.
Start an inquiry in Google Forms ↗ or email [email protected].