BHC AI Roadmap

1. Executive Summary 2. Current AI Readiness Snapshot Tools in Use Key Observations Gaps and Challenges 3. AI Opportunities Identified Key Opportunity Areas 1. Automated Referrals & Participant Onboarding 2. Dynamic Reporting & Data Insights 3. Meeting & Call Summarisation 4. Eligibility & Scheduling Automation 5. Data Quality Assurance Agents 6. Content Creation & Stakeholder Engagement 7. Tender & Proposal Support 4. Prioritisation Framework Explained 1. Effort Score 2. Ambiguity Score 3. Benefit Score ✅ Priority Score Formula 🎯 Use Case Categories Why We Prioritise 5. Use Case Framework Table 6. Use Case Descriptions 🟢 Quick Wins (0–3 months)✅ 1. AI-Powered Content Creation ✅ 2. Document Generation ✅ 3. Case Studies and White Papers ✅ 4. Admin Task Reduction ✅ 5. Referral Automation ✅ 6. Dynamic Dashboards ✅ 7. Python-Driven Analysis 🟡 Medium Term (3–12 months)✅ 1. Auto-Follow-Up for Drop-offs ✅ 2. Document Delivery via API ✅ 3. HubSpot Outreach Automation ✅ 4. Online Registration Workflow ✅ 5. Call Analysis with Transcripts ✅ 6. Survey Engagement Tracking ✅ 7. Health Coach Scheduling ✅ 8. Funder Reports ✅ 9. Internal Reporting Automation ✅ 10. Survey Monitoring ✅ 11. Data Discrepancy Detection ✅ 12. Follow-Up Engagement Triggers ✅ 13. KPI-Based Reporting ✅ 14. Communication Preferences Long-Term Use Cases (12–24 months)7. Assistant Evaluation & Platform Strategy 🧠 Assistant Platform Evaluation 🔍 Evaluation Criteria ✅ Platform Selection: Microsoft Copilot 365 🛠 Core Integration Map 🗂 SharePoint Setup Recommendations 🔐 Licensing & Access Guidelines 📣 Governance & Enablement What about other AI assistants ?8. Implementation Roadmap 🎯 Purpose 🛍️ Buy Track: Start with Value, Not Complexity 🧠 Build Track: Design Agents That Work for You 🧩 Why This Sequence Matters 📈 Summary: This Is Strategic Enablement 📆 3-Phase Rollout Plan 🗓️ Gantt-Style Milestone Table 📌 Execution Notes 9. Governance & Change Management 👥 Lightweight AI Working Group 📋 Practical Policies & Controls 🚀 Change Enablement Plan 📊 Light Metrics to Track 10. Next Steps & How We Can Help ✅ Immediate Next Steps 🙌 How We Can Support 🧠 Copilot Training & Prompt Coaching 🤖 Agent & Workflow Development 🔄 Ongoing Support & Light Retainer 🏁 Let’s Get Started

1. Executive Summary

Better Health Company is embarking on a strategic journey to integrate AI across its operations to improve efficiency, streamline participant onboarding, enhance data handling, and elevate the quality of reporting and decision-making. This AI Strategy Roadmap was developed through a comprehensive workshop series involving leadership and frontline staff, combined with a prioritised use case evaluation framework.

The focus of this roadmap is practical enablement. It reflects real-world pain points—like manual reporting, duplicated data entry, and slow participant onboarding—and transforms them into AI-enhanced solutions grounded in current capabilities. Immediate opportunities include automating referrals, improving data quality checks, and leveraging Microsoft Copilot and HubSpot AI for reporting and stakeholder engagement.

By sequencing these opportunities into Quick Wins, Medium-Term initiatives, and Long-Term transformations, Better Health Company can scale its AI maturity with confidence, ensuring all changes align with privacy, governance, and user readiness requirements.

2. Current AI Readiness Snapshot

Better Health Company is in the early adoption phase of AI, with some promising use cases already trialled by individuals across the team. The organisation shows strong curiosity and willingness to explore AI, especially where it reduces manual admin, streamlines reporting, or improves participant experience.

Tools in Use

Microsoft Copilot 365: Four licenses in use for proposal writing, email summarisation, and meeting thread insights.

ChatGPT (Free and Plus): Used for ideation, frameworks, research, email drafting, policy templates, and instructional support.

Murph AI: Used for e-learning narration.

Dialpad: Contains potential for call transcriptions and sentiment analysis, but currently not activated for coaching due to privacy concerns.

HubSpot: Core CRM system; some exploration with Breeze AI and form automation underway.

Key Observations

Copilot is underused beyond RFQ responses due to limited training and awareness of its full capabilities (e.g. Notebooks, Agents, Excel/PBI integration).

HubSpot is actively being used but not fully automated—AI connectors and form-based workflows are in planning stages.

Voice/Data Integration: Coaching calls are recorded manually; opportunities for automated transcription and note summarisation are recognised but not yet deployed.

Automation Appetite: The team recognises high-friction manual processes in reporting, onboarding, and referrals, and is eager to replace these with smart automations.

Gaps and Challenges

Data Fragmentation: Key data sits across disconnected systems (Better Health Data, Excel, HubSpot), slowing insights and reporting.

Privacy and Governance: Concerns around where AI tools store and process sensitive data—particularly participant information—limit current use.

Skill Confidence: While curiosity is high, prompt fluency and agent setup knowledge vary widely, especially beyond the innovation team.

Licensing Coverage: Only a small subset of the team currently has access to full Copilot or ChatGPT Plus functionality.

3. AI Opportunities Identified

The AI strategy workshops revealed a strong alignment between team frustrations and areas where AI can deliver immediate impact. Across departments, the most compelling opportunities focus on eliminating repetitive tasks, improving data quality, and streamlining participant-facing workflows.

Key Opportunity Areas

1. Automated Referrals & Participant Onboarding

Manual handling of faxed or emailed referral forms, combined with human eligibility checks and multiple follow-ups, makes this process time-consuming and prone to delays. Automating intake—particularly where form formats are standardised—was seen as a high-impact opportunity.

2. Dynamic Reporting & Data Insights

The reporting function currently requires time-intensive Excel manipulation and static data exports. There is strong interest in using Copilot in Excel and Power BI for:

Real-time dashboards

Dynamic demographic filtering

Survey and program progression tracking

Funder-specific and internal analytics

3. Meeting & Call Summarisation

Sales, stakeholder engagement, and coaching calls often involve valuable insights, but these aren’t always captured in a usable format. Tools like MeetGeek, Copilot for Teams, and potential Dialpad transcription can automate:

Summary notes

Follow-up actions

Sentiment tracking

Coaching QA compliance

4. Eligibility & Scheduling Automation

Currently, eligibility rules are manually applied post-signup, with limited logic built into HubSpot forms. Automating:

BMI eligibility checks

Program matching

Appointment booking (via Acuity or Calendly)

could transform the user journey from signup to engagement, reducing drop-offs and delays.

5. Data Quality Assurance Agents

Gaps or inconsistencies in Better Health Data (e.g. call notes missing attendance, illogical height/weight changes) impact reporting quality. AI agents could be trained to flag anomalies, nudge for missing data, and provide QA assistance across participant records.

6. Content Creation & Stakeholder Engagement

AI tools can support campaign content creation, stakeholder identification, and newsletter generation. Opportunities include:

Automating LinkedIn/email outreach lists

Local clinic identification via mapping tools

Generating brand-aligned newsletters from blogs or performance data

7. Tender & Proposal Support

Copilot’s Notebook and Agent tools have been successfully used to streamline tender development—by referencing previous submissions, reading RFQs, and generating compliant responses. This could be scaled across the team with proper training and shared setup.

4. Prioritisation Framework Explained

To ensure Better Health Company focuses on AI initiatives that are not only impactful but also achievable, we applied a structured prioritisation framework during the strategy workshops.

Each use case was scored based on three key criteria:

1. Effort Score

Represents the technical and operational complexity of implementation.

Factors: system integration, data availability, team capacity.

Lower score = easier to implement.

2. Ambiguity Score

Measures how clearly the use case is defined and understood.

Includes confidence in the solution path and expected outcome.

Lower score = more clearly defined.

3. Benefit Score

Estimates the potential business impact.

Factors: time saved, revenue enabled, compliance improved, experience enhanced.

Higher score = more valuable.

✅ Priority Score Formula

To compare all use cases fairly, the following weighted formula is used:

Priority Score=Benefit(Effort+Ambiguity2)×5\text{Priority Score} = \frac{\text{Benefit}}{\left(\frac{\text{Effort} + \text{Ambiguity}}{2}\right)} \times 5

Priority Score=(2Effort+Ambiguity)Benefit×5

This approach rewards use cases that are high-benefit and low-complexity—perfect candidates for early wins.

🎯 Use Case Categories

Priority Score	Category	Recommendation
10 and above	Quick Wins	Prioritise immediately (0–3 months)
6–10	Medium Term	Prepare and build readiness (3–12 months)
Below 6	Long Term	Reassess after foundations are laid (12–24 months)

Why We Prioritise

Reduces Risk: By validating small wins first, we de-risk larger investments.

Delivers ROI Fast: Quick results build confidence and funding momentum.

Builds Change Readiness: Helps team adopt AI gradually with tangible impact.

5. Use Case Framework Table

Category	Use Case	Description	Solution Type	Suggested tool	Comments	Efforts	Ambiguity	Benefits	Priority Score	Recommendation
Marketing & Business Development	AI-Powered Content Creation	Auto-generate blogs, emails, and social content with branding guidelines.	Assistive agent / Automation Agent	Copilot	Can start with a Copilot agent and move to automation later. 1. Use Copilot365 to create content 2. Create the following Copilot agents - Brand Tone evaluator, Blog writer, Linkedin writer 3. Create a Brand kit to go into Copilot create 4. Automate content/social creation using make.com workflow	1	2	5	16.67	Quick Wins (0-3 months)
Business Operations & Admin	Document Generation	Use AI to auto-create presentations, training materials, and summaries from inputs.	Assistant	Copilot	Sem-automation and efficiency can be achieved using Copilot. Implement Copilot365 and train team on proper usage	1	2	4	13.33	Quick Wins (0-3 months)
Marketing & Business Development	Case Studies and White Papers	Generate personalised case studies for different stakeholders and health partners.	Assistive agent / Automation Agent	Copilot	Can start with a Copilot agent and move to automation later	1	2	4	13.33	Quick Wins (0-3 months)
Business Operations & Admin	Admin Task Reduction	Use AI agents for email drafting, calendar booking, and repetitive internal tasks.	Assistant	Copilot	Requires training and testing	1	2	4	13.33	Quick Wins (0-3 months)
Participant Onboarding & Program Delivery	Referral Automation	Extract data from emails/faxes and populate referral forms automatically.	Automation Agent	Make	1. Setup make automation that will take referral forms received on the email or uploaded forms and populate the referrals form automatically in Hubspot/BHD	2	1	3	10.00	Quick Wins (0-3 months)
Reporting & Data Analysis	Dynamic Dashboards	Replace static reports with live dashboards (e.g., Power BI) for real-time visual insights.	Tool	Power BI	Power BI	1	2	3	10.00	Quick Wins (0-3 months)
Reporting & Data Analysis	Python-Driven Analysis	Use Python for deeper statistical and behavioural analysis across datasets.	Assistant	Copilot	Short term - try and test copilot for excel and evaluate the results.	1	2	3	10.00	Quick Wins (0-3 months)
Participant Onboarding & Program Delivery	Auto-Follow-Up for Drop-offs	Follow up with participants who begin but don't complete registration.	Automation Agent	Hubspot/Make	1. Setup a make / hubspot automation	3	2	4	8.00	Medium Term (3-12 months)
Participant Onboarding & Program Delivery	Document Delivery via API	Replace manual mail with automated document generation and delivery. GP letters via automation.	Feature	BHD	Develop feature in BHD. If postal - click send print and post	3	2	4	8.00	Medium Term (3-12 months)
Marketing & Business Development	HubSpot Outreach Automation	Automate targeted outreach using buyer intent data and tools like Clay/HeyReach.	Tool	Heyreach	Requires further discussion	3	2	4	8.00	Medium Term (3-12 months)
Participant Onboarding & Program Delivery	Online Registration Workflow	Create seamless onboarding with automated eligibility checks (e.g., BMI percentile).	Automation Agent/Functionality	Make/Hubspot	ㅤ	3	4	5	7.14	Medium Term (3-12 months)
Quality Assurance	Call Analysis with Transcripts	Transcribe calls and use AI to evaluate adherence to frameworks and conversational quality.	Automation Agent	Custom	ㅤ	3	3	4	6.67	Medium Term (3-12 months)
Quality Assurance	Survey Engagement Tracking	Monitor completion rates, prompt non-responders, and ensure data coverage.	Feature/Automation Agent	BHD/Make/Custom	ㅤ	4	2	4	6.67	Medium Term (3-12 months)
Participant Onboarding & Program Delivery	Health Coach Scheduling	Use AI to offer appointment slots based on coach availability and participant preferences.	Automation Agent	Make/Hubspot	Difficult to achieve with AI assistant - should explore a workflow that redirects to booking system	4	4	5	6.25	Medium Term (3-12 months)
Reporting & Data Analysis	Funder Reports	Transform raw data into formatted reports for funders, aligned with funding criteria.	Assistant	Copilot	Develop standard report format. Start with Copilot - once we figure out the right prompting logic - can be automated	1	4	3	6.00	Medium Term (3-12 months)
Reporting & Data Analysis	Internal Reporting Automation	Automatically generate internal reports to inform strategy and decision-making, reducing reliance on manual spreadsheet work.	Tool	Power BI	Start with Power BI	3	4	4	5.71	Medium Term (3-12 months)
Reporting & Data Analysis	Survey Monitoring	Automatically check for survey completions and send reminders (based on coaching calls)	Automation Agent	Make	ㅤ	3	1	2	5.00	Medium Term (3-12 months)
Reporting & Data Analysis	Data Discrepancy Detection	Flag mismatches in participant data (attendance vs engagement/activity).	Automation Agent	Custom	ㅤ	4	4	4	5.00	Medium Term (3-12 months)
Participant Engagement	Follow-Up Engagement Triggers	Auto-message participants who haven‚Äôt attended, engaged, or booked after registration.	Automation Agent/Feature	Hubspot/make ?	Can it be done via hubspot or make ?	3	3	3	5.00	Medium Term (3-12 months)
Marketing & Business Development	KPI-Based Reporting	Use AI to generate reports aligned with key performance indicators for business growth.	Tool	ㅤ	Can Power BI do it ? Start with Powerbi and revisit 3 months later	3	3	3	5.00	Medium Term (3-12 months)
Business Operations & Admin	Communication Preferences	Auto-send resources (e.g., factsheets) based on notes from coaching sessions.	Automation Agent	Make	Need to discover the link between resources and meeting triggers	3	3	3	5.00	Medium Term (3-12 months)
Participant Engagement	Intelligent Booking Assistant	AI assistant to recommend best appointment times based on user and coach schedules.	Feature	Custom	Difficult to achieve with AI assistant - should explore a workflow that redirects to booking system	5	2	3	4.29	Long Term (12-24 months)
Business Operations & Admin	Invoice Processing Automation	Streamline manual invoice entry and reconciliation processes.	Automation Agent	ㅤ	Requires indepth exploration - only benefit one person	4	3	3	4.29	Long Term (12-24 months)
Participant Onboarding & Program Delivery	Content-Based Resources	Suggest factsheets/resources based on discussion during coaching calls.	Automation Agent	Custom	ㅤ	3	4	3	4.29	Long Term (12-24 months)
Quality Assurance	Coaching Framework Alignment	Flag deviations from program guidelines during delivery.	Automation Agent	Custom	ㅤ	4	4	3	3.75	Long Term (12-24 months)
Participant Engagement	Coaching Note Generation	Record coaching calls and generate structured notes for review and quality control.	Automation Agent	Make/Custom?	ㅤ	5	2	2	2.86	Long Term (12-24 months)
Marketing & Business Development	Avatar Explainer Videos	Use AI-generated characters for visual storytelling and program explanation.	Tool	nuggts	ㅤ	1	3	1	2.50	Long Term (12-24 months)
Participant Onboarding & Program Delivery	Voucher/Mailout Automation	Send milestone rewards (e.g., vouchers) and	Automation Agent	Make/BHD	ㅤ	3	2	1	2.00	Long Term (12-24 months)
Reporting & Data Analysis	Weekly Insight Snapshots	Automate weekly data summaries to highlight trends and key metrics.	Automation Agent	Custom	Revisit it later	5	3	1	1.25	Long Term (12-24 months)

6. Use Case Descriptions

🟢 Quick Wins (0–3 months)

✅ 1. AI-Powered Content Creation

Function: Marketing / Communications

Solution Type: Assistive Agent / Automation Agent

Tool: Copilot + Make.com

🧩 Problem

Marketing and stakeholder teams spend excessive time creating consistent blogs, social posts, emails, and branded content. Manual creation leads to inconsistency, inefficiency, and underutilised internal insights.

💡 Solution

Use Copilot 365 to assist with content drafting, then scale to automation. Set up Copilot agents for brand tone checks, blog writing, and LinkedIn content generation. Automate posting using Make.com workflows for channels like HubSpot, Mailchimp, or LinkedIn.

📌 Implementation Steps

Create a Brand Kit inside Copilot (logo, colour, tone, tagline, fonts)

Deploy Copilot agents for:

Brand Tone Evaluator
Blog Writer
LinkedIn Writer

Train users to generate content from internal reports or notes

Automate distribution through Make.com workflows

🎯 Impact

Saves 4–6 hours/week in content production

Improves brand consistency across channels

Increases frequency of outreach without additional headcount

✅ 2. Document Generation

Function: Internal Comms / Training

Solution Type: Assistant

Tool: Microsoft Copilot

🧩 Problem

Teams manually create presentations, internal summaries, and training packs—often duplicating effort and inconsistently applying brand guidelines.

💡 Solution

Use Copilot to generate presentation decks, proposal outlines, and internal knowledge packs directly from existing documents or meeting summaries.

📌 Implementation Steps

Train staff on using Word and PowerPoint Copilot

Define prompt libraries for training decks, capability overviews, etc.

Build examples and templates to kick-start usage

Create Copilot prompts that extract summaries from meeting notes or reports

🎯 Impact

Reduces manual formatting and writing

Speeds up training material development

Improves consistency in internal documentation

✅ 3. Case Studies and White Papers

Function: Marketing / BD

Solution Type: Assistive Agent / Automation Agent

Tool: Copilot

🧩 Problem

Creating stakeholder-specific case studies is time-intensive and often delayed, limiting timely communication of success stories or pilots.

💡 Solution

Use Copilot to auto-generate stakeholder-aligned case studies from internal reports, client feedback, or project outcomes. Start with assistive mode and later automate from tagged source content.

📌 Implementation Steps

Tag key outcomes and project insights in SharePoint or CRM

Use Copilot to summarise per stakeholder type (e.g., PHN, GP, insurer)

Create visual templates for white paper and one-pager output

Build prompt library (“Generate a PHN-focused outcome summary for X project”)

🎯 Impact

Increases marketing agility

Enhances partner engagement

Reduces content bottlenecks

✅ 4. Admin Task Reduction

Function: Business Support

Solution Type: Assistant

Tool: Microsoft Copilot

🧩 Problem

Admin teams spend time on repetitive tasks such as email drafting, calendar booking, meeting summaries, and inbox triage.

💡 Solution

Deploy Copilot to reduce admin load through email summarisation, meeting follow-up generation, and proactive scheduling assistance.

📌 Implementation Steps

Enable Outlook Copilot and train on summarisation / response prompts

Test prompt libraries for:

Meeting follow-up
Calendar invite generation
Draft email replies

Roll out usage guidelines to support staff

🎯 Impact

Reduces email load and scheduling friction

Frees up admin team for higher-value tasks

Accelerates response time to internal and external queries

✅ 5. Referral Automation

Function: Program Intake

Solution Type: Automation Agent

Tool: Make.com + HubSpot

🧩 Problem

Referrals arrive via email or fax and require manual data entry into systems like HubSpot or Better Health Data, creating delay and risk of error.

💡 Solution

Automate the process of capturing structured data from referral forms (PDF or email) and populating participant records.

📌 Implementation Steps

Identify common form formats and data fields

Set up OCR workflows via Make.com to extract data

Map and sync into HubSpot or BHD

Add exception handling for incomplete or irregular forms

🎯 Impact

Cuts referral intake time from hours to minutes

Improves data accuracy

Reduces admin burden on frontline teams

✅ 6. Dynamic Dashboards

Function: Internal Reporting

Solution Type: Tool

Tool: Power BI

🧩 Problem

Program performance is tracked using static Excel reports, leading to delays in insights and reactive decision-making.

💡 Solution

Create real-time dashboards in Power BI linked to Better Health Data to enable live filtering, trend analysis, and cross-program insights.

📌 Implementation Steps

Define key KPIs and dimensions for filtering (e.g., demographic, region, outcome)

Set up Power BI data connector or CSV sync

Build reusable visual templates

Share dashboards with relevant managers or stakeholders

🎯 Impact

Improves speed and quality of decisions

Enables self-service analytics

Reduces ad hoc report requests

✅ 7. Python-Driven Analysis

Function: Strategy / Data

Solution Type: Assistant

Tool: Copilot + Excel / Notebooks

🧩 Problem

Deeper data analysis (e.g., behavioural trends, statistical relationships) is limited by Excel alone and requires specialist support.

💡 Solution

Use Copilot Notebooks or Excel Copilot to test Python queries on data sets for rapid exploration of participant behaviour and outcome trends.

📌 Implementation Steps

Identify priority questions (e.g., “Which participant types are more likely to complete?”)

Use Copilot to run simple regressions or cohort analysis

Review outputs with data analyst for validation

Train key staff on how to iterate with Python code via Copilot

🎯 Impact

Unlocks deeper program insight

Reduces bottlenecks on data team

Builds internal data literacy

🟡 Medium Term (3–12 months)

✅ 1. Auto-Follow-Up for Drop-offs

Function: Participant Onboarding & Program Delivery

Solution Type: Automation Agent

Tool: HubSpot / Make

🧩 Problem

Participants who begin the onboarding process but fail to complete registration often drop off unnoticed, reducing conversion rates and program impact.

💡 Solution

Set up an automated workflow in HubSpot or Make to detect drop-offs (e.g., form incomplete, no booking within X days) and trigger personalised follow-up messages.

📌 Implementation Steps

Define what constitutes a “drop-off” in HubSpot

Create smart list filters for inactive leads

Use Make to trigger reminder messages or escalate to phone follow-up

Track conversion and engagement results

🎯 Impact

Boosts participant activation

Reduces manual follow-up workload

Recovers value from lost leads

✅ 2. Document Delivery via API

Function: Program Delivery / Admin

Solution Type: Feature

Tool: Better Health Data

🧩 Problem

Clinically required letters (e.g., GP summaries) are currently created and sent manually, often by post or PDF email, which is inefficient and prone to delay.

💡 Solution

Develop a BHD-integrated API feature to generate and send standardised documents automatically via email or ClickSend print-post service.

📌 Implementation Steps

Build template letter format into BHD

Map triggers for generation (e.g., program completion)

Integrate ClickSend for postal delivery or secure email

Add logging of communications for compliance

🎯 Impact

Reduces admin time per participant

Ensures timely delivery to stakeholders

Improves documentation consistency

✅ 3. HubSpot Outreach Automation

Function: Marketing & Business Development

Solution Type: Tool

Tool: HeyReach / Clay

🧩 Problem

Outreach to new health partners and stakeholders is manual and ad hoc, limiting reach and missing opportunities from buyer intent signals.

💡 Solution

Use outreach automation tools (e.g., Clay, HeyReach) to automate contact list generation, message sequencing, and follow-ups based on intent or segmentation.

📌 Implementation Steps

Identify target segments (e.g., PHNs, clinics, insurers)

Integrate CRM with Clay/HeyReach tools

Create multi-touch outreach flows

Set tracking for replies, interest, and engagement

🎯 Impact

Expands outreach capacity

Improves consistency and targeting

Enables scalable BD activity

✅ 4. Online Registration Workflow

Function: Participant Onboarding

Solution Type: Automation Agent / Workflow

Tool: HubSpot + Make

🧩 Problem

Eligibility screening (e.g., BMI checks) is handled after registration, adding manual work and friction.

💡 Solution

Build a smarter registration flow that includes automated eligibility checks (e.g., age, BMI) and redirects ineligible users or routes them to alternate programs.

📌 Implementation Steps

Define rules and eligibility logic

Build logic into HubSpot forms or Make workflows

Create clear messages for pass/fail outcomes

Integrate with Acuity for bookings if eligible

🎯 Impact

Saves triage time

Reduces ineligible referrals

Streamlines onboarding

✅ 5. Call Analysis with Transcripts

Function: Quality Assurance

Solution Type: Automation Agent

Tool: Custom transcription pipeline

🧩 Problem

Manual QA of coaching calls is slow, inconsistent, and rarely covers more than a sample of interactions.

💡 Solution

Transcribe calls and use AI to evaluate adherence to frameworks, tone, duration, and conversational quality.

📌 Implementation Steps

Choose transcription tool (e.g., Dialpad, Whisper API)

Train AI model on what “good” looks like in calls

Set scoring system or flag deviations (e.g., missing call close, goal-setting absent)

Report summary to team leads

🎯 Impact

Improves QA coverage

Enhances coaching quality

Supports compliance and audit

✅ 6. Survey Engagement Tracking

Function: Quality Assurance

Solution Type: Feature / Automation Agent

Tool: BHD / Make / Custom

🧩 Problem

Survey completion rates can be low, and non-responders often fall through the cracks, reducing data quality for reporting and analysis.

💡 Solution

Implement survey tracking workflows to monitor non-completion and auto-remind participants. Include re-attempt triggers and opt-out flags.

📌 Implementation Steps

Identify all survey points and expected timelines

Track completions in BHD or CRM

Create alert and reminder workflows via Make

Add feedback loop to coach dashboards

🎯 Impact

Improves data coverage for funders

Increases participant engagement

Supports longitudinal program evaluation

✅ 7. Health Coach Scheduling

Function: Participant Onboarding & Program Delivery

Solution Type: Automation Agent

Tool: Make / HubSpot

🧩 Problem

Scheduling is handled manually or by coaches, causing delays and inconsistent availability for participants.

💡 Solution

Use an AI-driven or rules-based system to propose available time slots based on coach calendars and participant preferences.

📌 Implementation Steps

Connect calendar availability with HubSpot (or Acuity)

Set rules for optimal slot matching

Present options to participants dynamically

Include fallback or escalation logic

🎯 Impact

Reduces scheduling admin

Improves show rates

Enhances participant control

✅ 8. Funder Reports

Function: Reporting & Data Analysis

Solution Type: Assistant

Tool: Copilot

🧩 Problem

Preparing customised funder reports from raw program data takes hours per report and varies in format.

💡 Solution

Create a Copilot-assisted workflow that transforms program data into pre-structured reporting packs aligned to funder criteria.

📌 Implementation Steps

Define templates for key funders (e.g., PHNs, insurers)

Use Copilot to summarise metrics and add narrative

Automate population from Excel / Power BI dashboards

Generate outputs in Word or PowerPoint

🎯 Impact

Reduces manual report prep

Improves consistency across funders

Allows faster turnaround on requests

✅ 9. Internal Reporting Automation

Function: Strategy / Data Analysis

Solution Type: Tool

Tool: Power BI

🧩 Problem

Internal strategy relies on manually compiled spreadsheet reports, making it harder to act on trends and gaps in real time.

💡 Solution

Replace spreadsheet-based reporting with automated Power BI dashboards tailored to internal KPIs and decision-making timelines.

📌 Implementation Steps

Identify core metrics (e.g., completions, dropout rates, demographics)

Link source data via Excel/CSV connector

Build standardised views for execs, ops, and program leads

Enable drill-down and filter capabilities

🎯 Impact

Enables data-driven decision-making

Reduces reporting workload

Supports agile strategic pivots

✅ 10. Survey Monitoring

Function: Reporting & Data Analysis

Solution Type: Automation Agent

Tool: Make

🧩 Problem

Survey completions aren’t actively monitored, resulting in missed data and inconsistent participant engagement.

💡 Solution

Automate survey monitoring by checking participant activity logs and sending reminders or coach alerts when surveys remain incomplete.

📌 Implementation Steps

Define key survey points (e.g., week 4, week 8, post-program)

Set up Make workflow to check for missing submissions

Send automated reminders or generate task for follow-up

Track response rates over time

🎯 Impact

Increases data capture

Reduces manual tracking by coaches

Improves funder report quality

✅ 11. Data Discrepancy Detection

Function: Reporting & QA

Solution Type: Automation Agent

Tool: Custom

🧩 Problem

Mismatches often exist between attendance records and engagement data, skewing program evaluation and reporting.

💡 Solution

Implement logic-based discrepancy checks to flag inconsistencies between expected activity (e.g., booked call) and actual system data (e.g., no attendance).

📌 Implementation Steps

Define discrepancy types (e.g., call logged but no notes, survey scheduled but missing)

Build a rules engine in BHD or Make

Generate weekly report or alert for flagged records

Escalate for resolution by QA or admin

🎯 Impact

Improves data accuracy

Supports accountability

Strengthens evidence base for funding

✅ 12. Follow-Up Engagement Triggers

Function: Participant Engagement

Solution Type: Automation Agent / Feature

Tool: HubSpot / Make

🧩 Problem

Participants who register but fail to attend or engage are not followed up in a structured, consistent way.

💡 Solution

Trigger automated follow-up messages (email/SMS) based on inactivity indicators such as no bookings or no call engagement within 7–14 days.

📌 Implementation Steps

Define inactivity window (e.g., 10 days with no actions)

Set smart filters in HubSpot or Make

Send engagement nudges with dynamic messaging

Track conversion to activity

🎯 Impact

Boosts program retention

Supports timely reactivation

Reduces participant drop-off

✅ 13. KPI-Based Reporting

Function: Strategy / Marketing

Solution Type: Tool

Tool: Power BI or Copilot

🧩 Problem

KPIs for business development and executive insights are reported inconsistently, often requiring manual data collection and formatting.

💡 Solution

Create an AI-enabled dashboard or Copilot template that regularly updates and visualises progress on key metrics (e.g., partner engagement, pipeline conversion).

📌 Implementation Steps

Identify key business KPIs

Build data model using Power BI

Set visual templates and sharing schedule

Optionally, layer Copilot to generate monthly summary slides or emails

🎯 Impact

Improves visibility across teams

Enables faster strategic alignment

Reduces reporting admin

✅ 14. Communication Preferences

Function: Business Ops / Admin

Solution Type: Automation Agent

Tool: Make

🧩 Problem

Participants receive generic follow-ups even when their history or session notes suggest a tailored message or resource would be more effective.

💡 Solution

Use coaching session tags or notes to trigger the delivery of tailored resources or follow-up content automatically (e.g., stress tips, diet resources).

📌 Implementation Steps

Tag notes using Copilot or manual fields

Build Make.com workflow to match triggers to messages

Deliver resource by email or SMS

Monitor resource click/open rates

🎯 Impact

Enhances participant experience

Increases perceived support

Reduces manual comms tasks

Long-Term Use Cases (12–24 months)

These use cases represent high-potential AI opportunities that require further validation or foundational systems to be in place before execution. They typically involve complex integration, deeper automation, or advanced analytics and may span multiple departments.

At this stage, they are not described in full detail. Instead, they will be tracked as part of an ongoing AI opportunity pipeline and re-evaluated after the successful delivery of Quick Wins and Medium-Term projects.

These longer-horizon use cases can be monitored in a Use Case Tracker and prioritised for design sprints or pilot programs as part of future roadmap updates.

7. Assistant Evaluation & Platform Strategy

This section outlines the foundational platforms selected for AI rollout at Better Health Company, including the evaluation process, assistant selection rationale, and how systems like SharePoint, HubSpot, and Make.com will integrate to support scalable, secure automation.

🧠 Assistant Platform Evaluation

Before platform selection, we reviewed five leading Gen AI options for business use:

Microsoft Copilot 365

OpenAI ChatGPT

Claude (Anthropic)

Meta LLaMA

Grok (xAI / Twitter/X)

🔍 Evaluation Criteria

Category	Focus
Security & Privacy	Enterprise-grade compliance, tenant isolation, GDPR/SOC2 readiness
Integration	Seamless compatibility with Microsoft 365, SharePoint, CRM
User Readiness	Familiar interfaces, low learning curve
Prompt Capability	High context awareness, accuracy in structured workflows
Governance	Admin roles, auditability, role-based agent access
Ecosystem Fit	Expandability across Copilot Studio, Make, Power BI, SharePoint

✅ Platform Selection: Microsoft Copilot 365

Microsoft Copilot was selected as the primary assistant platform due to its:

Native integration across Outlook, Word, Excel, Teams, SharePoint

Enterprise-grade security and tenant boundary enforcement

Context-aware assistance that draws on organisational documents and conversations

Scalability to agents via Copilot Studio and SharePoint

This approach aligns with the Dumb Monkey AI Canvas: start with assistants → evolve into agents → expand automation.

🛠 Core Integration Map

System	Integration Role	Use Case Examples
SharePoint	Source of truth for Copilot + agent training	Knowledge assistants, document summarisation
HubSpot CRM	Workflow and contact sync for AI triggers	Referral automation, follow-up agents
Teams & Outlook	Source for meeting context and follow-ups	Action generation, stakeholder comms
Power BI	Live dashboard data for Copilot + reporting	Funders reports, internal KPIs
Make / Power Automate	No-code integration layer	Triggered resource delivery, survey checks

🗂 SharePoint Setup Recommendations

To maximise assistant accuracy and functionality:

Organise by department, program, and asset type

Use consistent metadata tagging: category, date, region, version

Create a “Copilot-Ready Folder” with curated, validated source content

Avoid storing program-critical material on personal drives or non-M365 platforms

🔐 Licensing & Access Guidelines

Item	Recommendation
Microsoft Copilot 365	Activate first for operations, coaching, reporting teams
Copilot Studio	Use for internal assistants trained on SharePoint content
Make.com / Power Automate	Configure for task-based automation
Data Security	Enforce role-based access, especially for participant data
Formats	Standardise on DOCX, XLSX, and searchable PDFs

📣 Governance & Enablement

Appoint AI Champions by function (ops, marketing, QA, reporting)

Establish an AI Working Group to govern roadmap, adoption, and agent feedback

Run quarterly reviews on usage patterns, Copilot training needs, and automation success metrics

What about other AI assistants ?

Tool	Summary
ChatGPT	Fantastic generalist, but requires setup, governance, and secure wrapper to use safely in enterprise
Claude	Excellent for long documents and reasoning, but lacks ecosystem integrations
Meta LLaMA	Free and open, but not yet enterprise-ready for structured business tasks
Grok	Social media-oriented and not fit for enterprise workflow environments

8. Implementation Roadmap

🎯 Purpose

This roadmap lays out the phased execution plan across Quick Wins, Medium-Term, and Long-Term AI initiatives, ensuring early wins are captured while building a sustainable AI foundation.

It follows the Dumb Monkey AI Canvas™ — a proven framework for scaling Gen AI in real-world businesses.

The canvas splits AI adoption into two complementary tracks:

🛍️ Buy Track: Start with Value, Not Complexity

Layer	Examples	Why Start Here
AI Assistants	Microsoft Copilot 365, ChatGPT	Off-the-shelf productivity tools that deliver instant value
AI Tools	Make.com, Power BI, MeetGeek	Targeted utilities that solve niche problems with minimal setup

✅ Better Health Company starts here, using 14 Quick Wins to rapidly automate admin, boost engagement, and streamline reporting.

🧠 Build Track: Design Agents That Work for You

Layer	What They Do	When to Use
Assistive Agents	Provide structured answers using internal data and rules	After Copilot and tools are embedded
Automation Agents	Trigger workflows, sync CRMs, move files based on rules	When data structures and workflows are stable
Autonomous Agents	Make decisions or run multi-step tasks with minimal human input	Once AI maturity, QA, and confidence are high

🚀 Your roadmap includes 14 Medium-Term use cases in this Build Track — designed to unlock scale through intelligent agents.

🧩 Why This Sequence Matters

Reduces risk: You validate impact with assistants before building advanced systems.

Improves accuracy: Iterating prompts and workflows early means fewer agent errors later.

Delivers fast ROI: You capture value from day one, proving the case for investment.

Avoids waste: No complex builds before the business is ready for them.

📈 Summary: This Is Strategic Enablement

By following the Dumb Monkey AI Canvas:

You start small, fast, and focused

You build momentum by solving real team problems

You grow confidence with each layer of maturity

This isn’t “AI transformation.” It’s smarter business.

📆 3-Phase Rollout Plan

Phase	Timeline	Use Case Category	Focus
Phase 1: Quick Wins & Enablement	Months 1–3	14 Quick Wins	Deploy Copilot assistants, automate referrals, improve reporting UX
Phase 2: Agent Design & Automation	Months 4–12	14 Medium-Term Use Cases	Implement structured QA, follow-up logic, health coach automation
Phase 3: AI-led Intelligence	Months 13–24	Long-Term AI Agents	Explore predictive analysis, journey tracking, and autonomous flagging

🗓️ Gantt-Style Milestone Table

Month Range	Key Activities
Month 1–2	✅ Copilot license rollout✅ SharePoint structure setup✅ AI content and email generation live
Month 3–4	🎯 Referral automation live📤 GP letter workflow built🧠 Survey reminder and QA agents scoped
Months 5–6	🔁 Drop-off detection automation💬 Meeting summary Copilot flows📊 Stakeholder pack assistant ready
Months 7–9	🧾 Internal reporting assistant📡 Call QA agent pilot📈 HubSpot outreach automation live
Months 10–12	🧠 Inactivity flag agent📋 Discrepancy detector prototyped📤 KPI-based reporting dashboard
Months 13–18	🚦 Risk scoring logic design📊 Journey analytics prep🧱 Health outcome forecasting scoping
Months 19–24	🧠 Smart engagement recommender📥 Contextual content generation🔎 Continuous program optimiser

📌 Execution Notes

Sequence is deliberate: Build confidence before complexity.

Dependencies: Later phases depend on CRM cleanup, consistent SharePoint access, and validated prompt design.

Review rhythm: Run roadmap retros every 3 months to update priorities and incorporate feedback from teams.

9. Governance & Change Management

For smaller, fast-moving teams like Better Health Company, AI governance should be simple, clear, and embedded into existing ways of working — not layered on top.

This section outlines a light-touch framework to ensure AI is used responsibly, securely, and in ways that drive real value.

👥 Lightweight AI Working Group

A small AI Working Group will help coordinate progress, resolve blockers, and adapt the roadmap over time — without adding bureaucracy.

Role	Responsibility
AI Champion (1–2 people)	Leads rollout, shares learnings, gathers team feedback
Function Leads	Represent real use cases, suggest new opportunities
Exec Sponsor	Ensures alignment with business goals and unblocks cross-team issues

💡 Tip: AI Champion doesn’t need to be technical — just curious, committed, and comfortable exploring Copilot and automation tools.

📋 Practical Policies & Controls

Area	Light Governance Approach
AI Usage	Agree on where AI is safe to use (e.g., internal docs, content), and where human review is still needed (e.g., clinical claims)
File Storage	Store all source content (for Copilot/agents) in SharePoint or CRM — avoid personal Google Drives or email folders
Feedback Loop	Use a shared form or Teams channel for “What Worked / What Broke / What We Wish” AI feedback
Review Rhythm	Every 3 months, revisit use cases, prompt performance, and system gaps with the AI Champion group

🚀 Change Enablement Plan

Stage	Activities
Awareness	Short team-wide walkthrough of the roadmap and early use cases
Training	Role-specific micro-sessions (30 mins) on Copilot, prompting, dashboards
Support	Copilot prompt templates, shared FAQ doc, peer support in Teams
Recognition	Celebrate “AI Wins” at team meetings — highlight time saved, improvements, or smart ideas

📊 Light Metrics to Track

Keep success measures simple and visible to everyone:

Area	Example Metric
Adoption	% of team using Copilot weekly, # of agent-triggered actions per week
Impact	Hours saved, documents generated, referrals automated
Feedback	# of new ideas submitted, AI bugs flagged, prompts improved
Confidence	Self-rated comfort using AI (surveyed quarterly)

10. Next Steps & How We Can Help

This section outlines the immediate actions Better Health Company can take to begin implementing AI, along with how we can provide hands-on, flexible support to drive real results — without adding overhead.

✅ Immediate Next Steps

Step	Description
1. Assign AI Champion	Nominate 1–2 team members to lead exploration and coordinate use case pilots
2. Confirm Copilot Access	Review current Microsoft licensing and enable Copilot for core users (e.g., ops, reporting, admin)
3. Select First 2–3 Quick Wins	Begin with low-risk, high-impact use cases (e.g., onboarding emails, content generation, stakeholder packs)
4. Set Up SharePoint Structure	Organise files into accessible folders for Copilot and future agents (e.g., docs, reports, templates)
5. Create Feedback Channel	Start a simple Teams or email thread for sharing wins, issues, and ideas for future automation

🙌 How We Can Support

We offer support tailored to smaller, action-oriented teams — with a focus on enablement, not dependency.

🧠 Copilot Training & Prompt Coaching

2-week starter program

Role-specific training on Copilot (e.g., Outlook, Word, Excel)

Prompt libraries and cheat sheets

Lightweight “Prompt Practice” sessions for team adoption

🤖 Agent & Workflow Development

Build custom Copilot agents using SharePoint content

Design automation flows in Make or Power Automate

Focus on quick-return automations: email follow-ups, CRM syncs, content prep

🔄 Ongoing Support & Light Retainer

Monthly check-in to review AI usage, unblock issues, prioritise next steps

Roadmap updates based on business changes

Optional shared dashboard to track adoption, value, and feedback

🏁 Let’s Get Started

We recommend scheduling a Kick-Off Call to:

Finalise the first 2–3 use cases

Confirm who will pilot Copilot

Set up file access and automation tools

Assign ownership for feedback and progress