The visible cost.
Clinical, content, GTM, and ops teams lose real hours every week to repetitive manual workflows that quietly compound across the org.
Built for Hello Heart
I help operationally complex companies move high-value AI ideas from discussion to daily use — without the complexity tax. This is a microsite built to show how I'd do that for Hello Heart.
A microsite built specifically for the Senior AI Builder role at Hello Heart.
The cost of friction
Great teams shouldn't lose hours to work software could do.
Every growing health company hits the same wall: more workflows, more tools, more handoffs. AI promises leverage, but most of that promise gets stuck in slides, pilots, and good intentions.
The frustrating part.
Obvious automation opportunities stall in meetings, get scoped into oblivion, or live in pilot mode forever — while the work keeps piling up.
The belief.
Internal AI should create real leverage for the people doing the work. If it adds steps, dashboards, or anxiety, it's not working.
Meet your guide
A builder who lives where software meets operations.
I'm Jimmy — a builder who lives at the intersection of software, operations, and the messy reality of how teams actually work. I understand both sides of the equation: the engineering required to ship internal AI tools, and the human workflows they're meant to improve. My job is to translate business problems into small pieces of working software that people quietly come to rely on.
- Full-stack web + automation builder
- Cross-functional translator across ops, GTM, and engineering
- Ships small, measures fast, iterates honestly
- Regulated-environment aware — governance-friendly by default
The operating model
How I'd approach it at Hello Heart.
A simple, repeatable loop for finding the right problems and shipping AI tools that stick.
1
Find the workflow friction.
Sit with each team, map their repetitive work, and surface where AI creates real time-back or risk reduction.
2
Prototype fast with guardrails.
Build the smallest useful version in days, not quarters — scoped, observable, and safe for a regulated environment.
3
Launch, measure, document.
Roll out to a small group, track adoption and time saved, and write down what worked so the next team gets it cheaper.
4
Scale reusable patterns.
Turn one-off wins into shared components — auth, prompts, evals, UI — so the org's AI capability compounds.
Concept tools for Hello Heart
Three internal AI products I'd build first.
These aren't generic side projects — they're examples of the kinds of internal AI tools I'd build for Hello Heart's clinical, content, ops, and GTM teams.
Workflow Intake & ROI Scorer
Problem it solves — helps leadership ship AI where it matters most, not where it's loudest.
Every team has automation ideas; very few of them are equally valuable. A structured intake turns scattered requests into a ranked, defensible pipeline so leadership invests in the workflows with the clearest member, clinical, or operational impact.
- Lightweight intake form any team can submit in minutes
- AI-ranked priority score: effort saved × frequency × risk
- Auto-tagging by function (clinical, GTM, content, ops, eng)
- Leadership dashboard showing pipeline, status, and realized ROI
How it works
A team submits a workflow they want automated — frequency, time spent, systems involved, risk profile. An LLM normalizes the description, asks clarifying questions, and produces a structured score against shared criteria. Leadership sees a live ranked queue with estimated hours saved per quarter. Once shipped, the tool tracks actual usage against the original estimate, closing the loop between promise and proof.
Regulated Content Co-Pilot
Problem it solves — accelerates member-facing and clinical-adjacent drafting without bypassing review.
Trust is the product. A co-pilot that bakes in tone, citation standards, and the existing review chain lets content, clinical, and marketing teams move faster while keeping governance intact — not because they remembered to, but because the tool quietly enforces it.
- Brand and clinical tone presets, by content type
- Source-citation prompts with required-evidence checks
- Built-in review routing (writer → clinical → comms)
- Audit trail of edits, prompts, and approvals
How it works
A writer picks a content type — member nudge, blog post, employer-facing one-pager. The co-pilot loads the relevant tone, structure, and citation requirements, then drafts against approved source material rather than the open internet. Required review steps appear inline; nothing publishes until each role has signed off. Every draft, prompt, and approval is logged, so compliance gets a real audit trail instead of a Slack thread.
GTM Operations Copilot
Problem it solves — removes manual CRM overhead from revenue and customer success teams.
Employer and health-plan relationships are too important to lose inside a CRM. A GTM copilot lets account teams stay in front of partners while the system quietly handles notes, follow-ups, and account-health signals in the background.
- Call transcription + structured summary with next steps
- Drafted follow-up emails matched to relationship stage
- Automatic CRM field updates (stage, contacts, risk signals)
- Account-risk surfacing based on engagement and sentiment
How it works
After a partner call, the copilot generates a structured summary — decisions, blockers, owners, dates — and drafts a follow-up email in the rep's voice. CRM fields update automatically, with humans reviewing only what changed. In the background, the system watches engagement patterns across accounts and flags ones drifting toward risk before they go quiet. The rep stays in conversation; the busywork stays out of their week.
What success looks like
Visible, defensible operational gains.
Not vibes — observable wins across the teams that keep Hello Heart running.
0%
Reduction in manual hours
on prioritized workflows in year one
0
Faster prototype-to-pilot
versus traditional internal tooling cycles
0%+
30-day tool retention
among intended internal users
Less manual effort
Hours back across clinical, content, GTM, and ops.
Faster internal execution
Shorter cycles from idea to shipped tool.
Better AI adoption
Tools people actually open in week three, not just week one.
Reusable internal systems
Each project lowers the cost of the next.
Measurable productivity gains
Time saved is tracked, not assumed.
Safer AI in a regulated environment
Governance built in, not bolted on.
Stakes
What's at stake if friction wins.
None of this is catastrophic on day one. It just quietly limits how much great work a great team can do.
- Strong teams stay stuck in repetitive work they're overqualified for.
- Internal AI efforts stay fragmented across tools, prompts, and tabs.
- High-ROI opportunities get talked about but never shipped.
- Knowledge stays siloed; adoption lags experimentation, and the org learns slower than it should.
Operating principles
How I think about internal AI.
Boring wins over flashy.
The best internal tools feel inevitable, not impressive. If a teammate forgets they're using AI, it's working.
Adoption is a design problem.
A tool nobody opens has zero ROI. I design for the second and third week of use, not the demo.
Ship the smallest useful thing.
Small surfaces, narrow scope, real users. Scope grows from usage, not from speculation.
Governance is a feature, not a tax.
In healthcare, guardrails are part of the product. Built in early, they make AI move faster — not slower.
Next step
Let's build the tools your teams will actually use.
Hello Heart already has the mission, the members, and the team. The opportunity is to give that team quiet, well-built internal AI that compounds — and I'd like to help build it.