HugoScore hugoscore.org

Condition-specific coaching and care AI

Chronilogix

Chronilogix offers AI coaching with goals and step plans between visits. Mixed: its notice contradicts itself on whether conversations are used for training. No control position, because the base model developer is not named. Documents only, not tested.

Document-based draft Read full report Open directory

Checklist assessment dated October 30, 2026. Independent confirmation is incomplete. Hugo has not personally verified this service.

Mixed7 of 12 agency checks metDocument-based draft. Independent confirmation incomplete; not product-tested.How categories work

Agency checklist · October 30, 2026

Mixed 7 of 12 yes

Document-based draft. Independent confirmation is incomplete. Hugo has not personally verified this service.

Assessed function: Public descriptions of Roni and Millie, consumer app listing, institutional sales pages and published policies. No authenticated or partner deployment tested.

A no can mean missing documentation or institution-dependent control; it does not by itself establish harmful behavior. These categories are not clinical safety or product-quality ratings.

  • Public documents do not establish actual product behavior.
  • The independent review screened rules and selected evidence; it did not replicate all twelve answers for this tool.
See all twelve answers and their evidence
QuestionAnswerEvidence / reviewSource and reasoning
1. Told it is AIYesdocumented
Not independently replicated
Chronilogix AI is an AI model powered by large language models (LLMs) and is under continuous training. Source

Terms section 2.4. The App Store listing also says 'Chat with an AI-powered coach'.

2. Operator and model maker namedNonot-documented
Not independently replicated
Quotation not recorded; evidence repair pending.

Operator is named: Chronilogix, Inc., Delaware, Fairfield CT (privacy policy). The terms describe 'proprietary fine-tuned AI models' and LLMs but no model maker is named.

3. Patient chooses to use itYesdocumented
Not independently replicated
Sign in and choose your coach Source

Consumer app path: the user downloads the app and signs in. Employer, plan and partner deployments are a separate channel and were not assessed. For those, this answer would be deployer-dependent.

4. Patient sets the taskYesdocumented
Not independently replicated
Chronilogix adapts its guidance based on your goals, priorities, and progress. Source

The user sets goals in the consumer app. The coaching method (motivational interviewing) is fixed by the operator, and institutional buyers pitch outcomes like claims reduction, so this may differ in deployments.

5. Patient can decline or stopYesdocumented
Not independently replicated
Sign in and choose your coach Source

Self-started tool rule: Q3 is yes and no care depends on it for the consumer app. Source is the Q3 source. In a hybrid or plan-delivered program this could be different.

6. Patient sees the AI outputYesdocumented
Not independently replicated
Chat with an AI-powered coach Source

The patient sees the chat directly. Privacy policy says transcripts are stored. FAQ says interactions can be summarized for a clinician.

7. Patient can correct or challengeNonot-documented
Not independently replicated
Quotation not recorded; evidence repair pending.

The privacy policy gives a right to rectify inaccurate personal data, but nothing documents a route to correct or contest an AI coaching output. Terms section 2.4 only warns about errors.

8. Patient can get a copyYesdocumented
Not independently replicated
You have the right to receive the personal data you have provided to us in a structured, commonly used, and machine-readable format Source

Privacy policy section VII, by emailing the privacy address. The right is stated as limited to consent or contract based processing.

9. Patient can deleteYesdocumented
Not independently replicated
If you delete your account, all associated data is permanently removed from our servers Source

Section V adds 'except for aggregated and anonymized data used for AI improvement'. Conversations are kept until the user deletes them.

10. Training or sale does not override patient choiceNodocumented
Not independently replicated
To continuously train, refine, and improve the accuracy, safety, and performance of the models and algorithms underlying Chronilogix AI Source

Privacy III.B and Terms 4.3 and 5.2 use aggregated, de-identified user content to train models under a perpetual license. No opt-out is documented, only a conditional right to object. The site says 'Member data is never used to train our models', which conflicts, so the binding documents govern.

11. Patient can run it or take it elsewhereNonot-documented
Not independently replicated
Quotation not recorded; evidence repair pending.

Vendor-hosted app. Portability is by email request only, and no way to run it yourself or move the function is documented.

12. No advertising or data businessNodocumented
Not independently replicated
to serve our advertisements to you through third-party platforms such as Facebook or Google Source

Privacy III.E and X document retargeting ads, and IV.B shares aggregated and de-identified data with analytics and advertising providers. The ads promote Chronilogix, not third parties. The policy says it does not sell personal data.

Initial coding: Claude Sonnet helpers, compiled by Claude Opus 5.5 (Anthropic). Selective review and corrections: GPT-6 Astra (OpenAI). Editor: Hugo Campos. Acceptance of corrections is not product verification.

Checklist rules · Independent review and limitations · Later evidence follow-up

Earlier v1.3 assessment and control finding

Historical assessment under different rules. It does not set the current checklist zone.

MixedDocuments only, not tested.

Agency finding

Mixed

No position: model provider not disclosed

Unresolved, toward the middle of the institutional to patient-directed line on current documents. The patient opens the app, chooses a coach and sets goals, which places the direct consumer function above the institutional end. The rails that could pull it toward that end are documented but not tied to this deployment: an operator-fixed Motivational Interviewing method whose own implementation guide has the deployer identify 'the change behavior' and track conversion, refill and retention, buyer marketing for claims mitigation and adherence, a medication-adherence thread built into the diabetes coach, and a validation layer that rewrites replies by unpublished criteria. The output license limits use to individual self-reflection with unclear scope for sharing with a clinician. Operator and model developer disclosure: the operator is named, the model developer is not, so no position or candidate set is shown (model provider not disclosed).

Evidence
Documentary review. Public sources only, no hands-on testing by HugoScore.
Review status
Human adjudication: Hugo Campos settled named questions (what was reviewed, the disclosure results, any coder disagreement). Not verification of the service's claims.
Coded by
Claude Opus 5.5 (Anthropic), GPT-6 Astra (OpenAI) Conflict check open: the model developer is not known.
  • Act Goal setting and step plans between visits documented, not tested

    The patient can turn a broad aim into small, scheduled steps, track progress and adjust the plan with a coach available at any hour, documented in the App Store description and product page. Comparator: the patient's realistic alternative in this use is no between-visit coaching or a human coach with limited availability; the comparator itself rests on vendor statements. Operation untested.

  • Refuse or leave Training use contradicted within the notice documented, not tested

    The Terms (4.3, 5.2) and Privacy Policy (III.B, V) take the user's explicit consent and a perpetual license to use anonymized, aggregated User Content to train Chronilogix AI and other Chronilogix products. The footer printed on those same Terms and Privacy pages says 'Member data is never used to train our models', and the Product page and FAQ say conversations are never 'used to improve our models'. No reachable clarification says whether 'member' excludes direct users. A patient reading the notice at entry cannot tell whether their conversations train the models. The underlying practice remains unknown. Blocks informed entry (refuse or leave).

The question we ask Who does Chronilogix serve in this deployment?
Control Public descriptions of Roni and Millie, consumer app listing, institutional sales pages and published policies. No authenticated or partner deployment tested.
Agency read Mixed
Vendor
Chronilogix, Inc.
Who it serves
Hybrid: patient-facing coaching with institution-directed commercial deployments and a consumer app
Primary User
Patients and members using AI coaching; employers, plans and other partners purchase deployments
Control Model
Vendor-hosted; consumer app and partner-embedded offerings
Patient Impact
Coaching can shape self-management goals, medication-related conversations and care-seeking decisions.
Profile Status
Document-based draft
Last assessed
Oct 30, 2026
Earlier review confidence
Medium confidence in the agency finding, low for control. Documentary evidence only.
Earlier review model
Claude Opus 5.5 (Anthropic), coder 1; GPT-6 Astra (OpenAI), coder 2
Earlier human review record
Hugo Campos settled named questions on October 30, 2026: what was reviewed and the disclosure results. He has not verified the service's claims.

Earlier evidence and assessment

The sections below preserve the earlier review. Its classifications and provenance are historical; the checklist above gives the current draft classification.

Earlier finding · No position: model provider not disclosed

Method v1.3. The control line is evidence for how the agency finding arose, not a quality score.

Mixed

Patient-facing support and adjustable goals coexist with institutional purchasing priorities and unresolved disclosure conflicts. No numerical ranking is assigned.

Patient agency

How this tool changes agency

Expands agency when

Potential support for reflection, personally meaningful goals and between-visit self-management.

Limits agency when

Patient control of sponsor goals, AI interpretations and sharing is unverified; inconsistent disclosures complicate informed choice.

Patient agency assessment

Who sets and changes the goal?

Partial

The consumer listing describes adjustable goals; authority when patient and sponsor priorities differ is unverified.

Can patients tell AI is involved?

Yes

Public materials identify AI coaches; disclosure inside partner deployments was not tested.

Can patients meaningfully choose?

Partial

Consumer choice is described; refusal without institutional consequences is not established.

Can patients correct or challenge the output?

Partial

Data correction rights are stated, but correction of AI interpretations and shared summaries was not tested.

Does it help patients understand or act?

Potentially

Goal setting and reflection may support action; advocacy against institutional decisions is not established.

Can patients evaluate the conditions of use?

Partial

The FAQ and privacy policy disagree about model improvement using conversations. Actual training practice is unverified.

Text findings

What happens to patient data?

Conflicting disclosures

The FAQ excludes conversation use for model improvement; privacy policy III.B permits training with primarily de-identified content. V retains anonymized improvement data after deletion. These are policy statements, not an audit of practice.

Who is left out or burdened?

Not independently evaluated

Cultural tailoring and broader access are claimed. Accessibility, affordability in each deployment and subgroup outcomes are unverified.

Are the clinical boundaries clear?

Partial

Coaching-replacement marketing requires clarification alongside self-help-only terms. Human coaching and licensed clinical care are not interchangeable; escalation performance was not tested.

Who defined what good looks like?

Vendor/scientific leadership visible

Patient partnership in defining evaluation outcomes was not established in the reviewed public sources.

Earlier review provenance

Criteria

HugoScore method v1.3, published October 30, 2026 (hugoscore.org/framework).

Reviewer

AI coders working from public sources under the v1.3 review procedure, with adjudication by Hugo Campos

AI / model

Claude Opus 5.5 (Anthropic), coder 1; GPT-6 Astra (OpenAI), coder 2

Human review

Hugo Campos settled named questions on October 30, 2026: what was reviewed and the disclosure results. He has not verified the service's claims.

Review date

2026-10-30

Limitations

Documentary review of public sources only. No account, interaction testing, health-data upload, security audit or vendor interview. Vendor statements are not independent validation.

Earlier review method

HugoScore method v1.3 documentary review of public sources, listed in the report. Coded independently by two AI coders. No account, interaction testing or health-data upload.

Document-based draft · Medium confidence in the agency finding, low for control. Documentary evidence only.