The Self-Efficacy for Diabetes Management (SEDM) scale will be used to measure diabetes self-efficacy at baseline. The SEDM has scale from 1-10 where 1 denotes 'not at all sure' and 10 denotes 'completely sure' where higher scores mean better outcomes.

Minimal clinically important difference is 10%.

• transition readiness measured by the Readiness Assessment of Emerging Adults with Type 1 Diabetes Diagnosed in Youth (READDY) tool at 12 months; READDY is a transition readiness assessment for emerging adults with diabetes diagnosed in youth. The READDY tool assesses diabetes-related knowledge or skill items by querying respondents on 42 total items split into 5 domains: knowledge, navigation, health behaviors, and insulin pump skills. Respondents answer on a Likert scale from "yes, I can do this" scored 5 to "Haven't thought about it" scored 1; confidence level is evaluated in each domain with a higher score indicating more confidence

• transition readiness measured by the Readiness Assessment of Emerging Adults with Type 1 Diabetes Diagnosed in Youth (READDY) tool at 6 months; READDY is a transition readiness assessment for emerging adults with diabetes diagnosed in youth. The READDY tool assesses diabetes-related knowledge or skill items by querying respondents on 42 total items split into 5 domains: knowledge, navigation, health behaviors, and insulin pump skills. Respondents answer on a Likert scale from "yes, I can do this" scored 5 to "Haven't thought about it" scored 1; confidence level is evaluated in each domain with a higher score indicating more confidence

• transition readiness measured by the Readiness Assessment of Emerging Adults with Type 1 Diabetes Diagnosed in Youth (READDY) tool at baseline. READDY is a transition readiness assessment for emerging adults with diabetes diagnosed in youth. The READDY tool assesses diabetes-related knowledge or skill items by querying respondents on 42 total items split into 5 domains: knowledge, navigation, health behaviors, and insulin pump skills. Respondents answer on a Likert scale from "yes, I can do this" scored 5 to "Haven't thought about it" scored 1; confidence level is evaluated in each domain with a higher score indicating more confidence

• perceived stigma of living with T1D using the Barriers to Diabetes Adherence in Adolescence questionnaire (BDA) Stigma Subscale; in our study, stigma will be defined as an affirmative response to at least one of three key items on this subscale. The scale ranges from "not at all true" (1) to "completely true" (5). Where a score greater than 2 to at least 1 of 3 key items indicates stigma.

• perceived stigma of living with T1D using the Barriers to Diabetes Adherence in Adolescence questionnaire (BDA) Stigma Subscale; in our study, stigma will be defined as an affirmative response to at least one of three key items on this subscale. The scale ranges from "not at all true" (1) to "completely true" (5). Where a score greater than 2 to at least 1 of 3 key items indicates stigma.

• perceived stigma of living with T1D using the Barriers to Diabetes Adherence in Adolescence questionnaire (BDA) Stigma Subscale; in our study, stigma will be defined as an affirmative response to at least one of three key items on this subscale. The scale ranges from "not at all true" (1) to "completely true" (5). Where a score greater than 2 to at least 1 of 3 key items indicates stigma.

• A1c will be self-reported in the following categories of values: < 7%, 7 to < 8%, 8 to < 9%, >9% where higher percentage indicates poorer diabetes management

A1c will be self-reported in the following categories of values: < 7%, 7 to < 8%, 8 to < 9%, >9% where higher percentage indicates poorer diabetes management

A1c will be self-reported in the following categories of values: < 7%, 7 to < 8%, 8 to < 9%, >9% where higher percentage indicates poorer diabetes management

• Health administrative data from ICES in Ontario and RAMQ in Quebec will be collected during the 12 months of intervention. ICES is an independent, not-for-profit research institute made up of a community of research, data and clinical experts in Ontario. RAMQ manages, administers and publishes health data in the province of Quebec.

• Health administrative data from ICES in Ontario and RAMQ in Quebec will be collected during the 24 months prior to enrolment. ICES is an independent, not-for-profit research institute made up of a community of research, data and clinical experts in Ontario. RAMQ manages, administers and publishes health data in the province of Quebec.

• Health administrative data from ICES in Ontario and RAMQ in Quebec will be collected during the 12 months of intervention. ICES is an independent, not-for-profit research institute made up of a community of research, data and clinical experts in Ontario. RAMQ manages, administers and publishes health data in the province of Quebec.

• Health administrative data from ICES in Ontario and RAMQ in Quebec will be collected during the 24 months prior to enrolment. ICES is an independent, not-for-profit research institute made up of a community of research, data and clinical experts in Ontario. RAMQ manages, administers and publishes health data in the province of Quebec.

• Costs of implementation will be collected to ascertain the cost of developing and using this intervention in a hospital setting in Ontario and Quebec; variables that will be collected for this analysis will include: salaries for staff working on the design, development and management of the intervention for the entire intervention duration; salary of research coordinators working to onboard participants to the intervention; cost of digital services and office equipment needed to develop and manage the intervention (computers, hosting service, servers); cost of REDcap servers and hospital IT support for REDCap integration with MEMOTEXT platform
• Aggregate and direct medical costs associated with the intervention and its implementation will be measured for the 12 months of the intervention duration using study and administrative data

• Analyses for the preliminary safety and performance KiT LLM chatbot testing will include descriptively describing whether the LLM Evaluator acts more conservatively than human raters when scoring answers, and if the answers generated by KiT LLM are acceptable based on the graded criteria.
• After the trial is completed, the investigators will analyze conversation logs between participants and KiT LLM to derive effective engagement with the bot. This will include analyzing the conversation between participants and KiT LLM (eg, duration, breadth, and depth of bot usage).
• The performance of the LLM Evaluator will also be assessed by evaluating conversations flagged and any potentially sensitive conversations missed. This evaluation will provide insights into how KiT LLM was used in a real-world setting and opportunities to improve its performance over time.

Embedded process evaluation using real-time engagement metrics looking at which types of messages are the most and least engaged with Concurrent process evaluation allows for real-time intervention modification to increase and improve user engagement and content delivery

Embedded process evaluation using real-time engagement metrics (i.e. number of participants who are engaging with intervention messages on and which types of messages are the most and least engaged with) Concurrent process evaluation allows for real-time intervention modification to increase and improve user engagement and content delivery

The investigators will identify the potential paths and corresponding combination of factors that are associated with trial outcomes using Coincidence Analysis.