From adoption evidence to patient access
Connect the research to the scheduling, intake, insurance, referral, follow-up, and human-escalation workflows healthcare teams operate.
Explore healthcare workflows →Home / Research
Claire Research
Evidence for better AI operations.
Original analysis for leaders deciding where AI should act, where people must remain involved, and how operational automation changes economics, risk, governance, and customer experience.
Featured original research
Current Claire Research
Evidence-led reports with explicit methodology, source labels, limitations and stable citation guidance.
Evidence Review · Published August 31, 2026Research to workflow
The Home Services Call-to-Job Gap
A 58-row evidence review of missed calls, after-hours HVAC demand, booking, capacity and the rush into AI reception.
Read the home services report →Evidence Review · Published August 31, 2026The Property Maintenance Coordination Gap
A 78-row review of resident communication, emergency triage, vendor coordination, AI adoption and the work between request and resolution that remains poorly measured.
Read the property maintenance report →Evidence Review · Published August 31, 2026The Freight Status Gap
An 80-row evidence review of manual shipment-status work, what AI vendors claim, what independent research can verify, and the benchmarks freight still does not have.
Read the freight report →Industry Research · Published May 28, 20262026 State of AI Receptionists in U.S. Medical Practices
Original research with methodology, citations, limitations, market evidence, and operational recommendations. The report was featured by Healthcare IT Today, and this citation-bearing URL is preserved as a protected authority asset.
Read the full report →Research taxonomy
Authority built around operational questions.
Future studies will be published only when there is a defensible method and useful evidence—not to manufacture statistics or fill a content calendar.
Workflow performance
How AI workers behave in production: completion, exceptions, handoffs, reliability, cost, and operational ownership.
Control in real systems
Permissions, data boundaries, evaluation, auditability, policy, and the human decisions that should remain explicit.
How work actually happens
Original studies of intake, scheduling, dispatch, status, follow-up, and administrative burden in specific operations.
From pilot to operation
What deployment requires across systems, workflow design, testing, change management, and exception handling.
Economics without invented proof
Evidence-based analysis of labor leverage, missed interactions, revenue leakage, implementation burden, and measurable outcomes.
Citation-ready operating data
Future anonymized and permissioned benchmarks with explicit samples, methods, definitions, and limitations.
Authority flywheel
Research should improve decisions—and commercial pages.
Original research can earn journalist and industry citations, build topical authority, and help commercial pages rank. The link must work both ways: research should point readers to relevant operational workflows, while industry and solution pages should cite the evidence that genuinely supports them.
The Freight Status Gap is now published as a secondary-evidence review. Planned primary research on check-call workload and property-maintenance coordination remains uncollected until a real method and defensible sample are in place; the two published evidence reviews do not claim original outcomes.
How Claire Research should earn trust
Show the method, evidence, limitations, and decision relevance.
Research is useful when an operator can understand where the evidence came from and what it does not prove. Major studies should identify the research question, population or market observed, collection period, inclusion criteria, sample context, calculation method, source notes, limitations, publication date, and update history.
Third-party findings, estimates, practitioner language, product telemetry, and Claire product interpretation should be labeled separately. We do not turn a competitor claim into a market fact, a synthetic demonstration into a customer outcome, or a small sample into a universal benchmark.
For operators
Translate evidence into workflow decisions: where demand enters, what work repeats, where delays occur, which exceptions require people, and what baseline should be measured.
For buyers
Explain how to evaluate vendors through workflow completion, system actions, human control, failure behavior, implementation burden, and evidence quality.
For product teams
Use repeated market needs to identify whether Claire should configure an existing primitive, integrate a system, build a new capability, or decline the use case.
For future studies
Aggregate only ethically permitted, appropriately anonymized data and never publish customer-sensitive records or invented benchmark results.
When quantitative evidence is unavailable, the page should say so. Qualitative operator language can inform a workflow hypothesis, but search estimates, commercial outcomes, competitor adoption, and customer results must not be invented.
Every study should end with decisions the evidence can reasonably support and questions it cannot answer. Operators can then use the findings to form a bounded pilot, establish a baseline, and collect better first-party evidence without presenting correlation, survey opinion, search language, or product telemetry as causal proof.