🔔 Ring the Bell

Ring once. Hang up. We call you back.

Ring the Bell is a voice-first bridge between mindfulness practice inside incarceration and human connection after release. It works from almost any phone, with no app, no account, and no smartphone required.

Inspiration

Boundless Freedom Project (BFP) teaches mindfulness inside correctional institutions. Participants learn simple practices they can return to when life feels overwhelming.

But release creates a continuity problem.

The practices may still be familiar, but the structure around them suddenly disappears. The phone may change. Internet access may be unreliable. The community they practiced with is no longer physically around them. And the first days and weeks after coming home can be some of the most disorienting.

When we spoke with Boundless Freedom Project, one idea kept coming back:

Connection matters. No one should feel like they have to do this alone.

So instead of building another reentry app, we asked:

What if the bridge home could fit in your wallet and work from almost any phone?

That became Ring the Bell.

What it does

One ring starts the bridge

A participant calls the Ring the Bell number and hangs up. The system rejects the incoming call before answering, so it costs them nothing, even on a prepaid phone with no minutes left. Then it calls them back.

They hear a bell, choose English or Spanish, and then, before anything else, hear:

"You're not alone. If you need support, let us know. To talk to a live person, press 1."

From there they can:

  • return to one of their familiar mindfulness practices
  • talk to a real person from Boundless
  • reconnect with the Outside Sangha
  • press 5 — "Just came home?" for a gentler welcome: practice together, one-on-one support, or community
  • reach clear 911 and 988 information at any moment

No account. No password. No app.

A physical card carries continuity

Before release, a participant chooses three familiar practices and a language. Those choices are encoded into a short, permanent card number.

The card does not contain their name, incarceration history, practice history, or wellness profile.

From another phone, a replacement phone, or a borrowed phone, the same card brings back the same familiar starting point.

And if the card is lost? They can still call.

Personalization is optional. Access is not.

Familiar practices first

The system starts with practices the participant already knows, rather than introducing another wellness program at the moment of reentry. Participants can also reach all 21 of BFP's one-minute exercises (press 8 on the phone).

On the web version, someone can type what they remember, even something approximate like "the hug one" or a misspelled phrase, and Ring finds the right practice. Practice note cards let them review the steps, with their own three first.

There are no streaks, scores, leaderboards, or missed-day warnings. The goal is not to measure the person. The goal is to make returning easier.

Help now, even mid-practice

If a practice brings something up, pressing 9 works at any moment, including in the middle of a practice. Ring stops, says plainly:

  • 911 for immediate danger
  • 988 to call or text for a mental-health crisis

and offers to repeat the numbers. If it's too much, but not an emergency, a short grounding step helps them come back to the room: eyes open, three things you can see, feet on the floor, one slow breath.

Ring never places an emergency call for anyone. The person stays in control.

Human connection is one key away

At any point, a participant can ask for a person. If someone from Boundless is available, Ring connects the call live. If no one picks up, it never ends at a dead end. It creates a warm callback request with only what the team needs:

  • preferred language
  • safest time to call
  • whether voicemail is safe
  • the topic they want help with

The participant does not have to explain their entire story to enter the queue.

Requests move through a simple handoff flow:

Waiting → Assigned → Reaching out → Connected → Closed

When the connection is complete, the participant's contact details are erased. Boundless keeps the handoff, not an unnecessary long-term profile.

The Day 7 envelope

Before release, the participant can also write themselves a physical note: something you want the person you are seven days from now to remember. It goes into a sealed envelope.

When they call, Ring offers a moment to open it: "If you still have your Day 7 envelope, press 7."

The system never reads it. Never analyzes it. Never stores it.

The most personal part of Ring the Bell is deliberately something Ring the Bell knows nothing about.

How we built it

Ring the Bell is intentionally lightweight: Node.js with zero dependencies, Twilio Voice through its REST API (no SDK), and plain HTML, CSS, and JavaScript.

Voice-first architecture

The core phone engine is independent from the interface. The same stateless state machine can render its next step as:

  • TwiML for a real Twilio phone call
  • structured data for our browser phone simulator

That gives us a real phone experience, plus a reliable fallback for facilitators when venue Wi-Fi, cellular service, or telephony infrastructure becomes unreliable.

Missed-call callback

The incoming call receives <Reject reason="busy"/>, so it is never answered and never billed to the caller. The server then starts the callback through Twilio. Live one-on-one support uses <Dial> to a staff phone, with an automatic fallback to a callback request if no one answers.

The card is data, not an account

A participant chooses 3 practices from 21:

$$\binom{21}{3} = 1330$$

With two supported languages:

$$1330 \times 2 = 2660$$

possible preference combinations, which fit inside a short card code. A Damm check digit catches entry errors. We exhaustively tested every single-digit substitution and adjacent-digit swap across all valid cards, and every tested error was detected.

There is still always a no-card fallback.

Continuity stress testing

We did not only test the perfect flow. We deliberately tried to break it. Our live continuity benchmark includes:

  • lost phone
  • changed number
  • borrowed phone
  • no caller ID
  • no minutes
  • lost card
  • damaged card
  • mistyped card
  • Spanish
  • shared device
  • inability to speak privately
  • staff unavailable
  • crisis path

Across the current benchmark, 15 / 15 scenarios still reach a practice, a human path, or a safe fallback without locking the participant out.

That became one of our core design principles:

No wrong door. Nothing needed for personalization is required for access.

Privacy by design

We designed each feature around the minimum data it actually needs:

  • A default practice needs no identity information.
  • A familiar practice needs the card code.
  • A human callback temporarily needs a phone number and callback preference.

Ring does not require an account, email, location, incarceration history, a longitudinal mood history, or a practice history. Even the "you're not alone" count is a single number per day: no names, no phone numbers.

The participant-facing interface avoids revealing reentry context on a shared phone, and includes a Quick Exit. After relevant actions, Ring shows a simple privacy receipt explaining what information was used and what was never requested.

Safety boundaries

Ring the Bell is not an emergency service, therapist, or crisis-monitoring system. It clearly distinguishes:

  • 911 for immediate danger
  • 988 for mental-health crisis support
  • Boundless support, which is separate and may not be immediate

The voice crisis path does not silently create a Boundless case or automatically place an emergency call.

Challenges we faced

Designing for a keypad instead of a touchscreen

A touchscreen makes it easy to hide complexity behind menus. A phone keypad does not. Every additional choice costs attention.

After testing with BFP, we learned our early call read options too fast and ran them together. We restructured it: language first, then "you're not alone," then practices, with pauses between every option. We kept simplifying until the important actions were obvious: practice, person, community, safety.

Building for things going wrong

Real life does not preserve the perfect demo state. Phones get lost. Numbers change. Caller ID disappears. Cards get damaged. Staff may be unavailable. Audio may fail to autoplay. Venue Wi-Fi drops mid-demo.

Instead of treating those as edge cases, we treated them as requirements, and hardened each one so the line falls back gracefully instead of breaking.

Safety without pretending to be healthcare

We had to resist features that sounded technologically impressive but would have made the product less trustworthy. We deliberately did not build:

  • emotion detection
  • risk scoring
  • an AI therapist
  • recidivism prediction
  • passive monitoring
  • automatic family alerts
  • mood scoring

We even considered connecting callers directly to crisis lines, and chose instead to guide, not dial.

Ring should help someone reconnect, not evaluate them.

Staying honest about content

The practice names and one-line instructions come from BFP's own 21 One-Minute Mindfulness Exercises. The step-by-step guided wording and the Spanish translations are ours, and they are labeled as material for BFP to review, not as official clinical or therapeutic content.

What we learned

Co-design changed the product

Our early thinking focused on getting mindfulness exercises onto another device. BFP's feedback shifted the center of the project toward connection. That led to live one-on-one support, the "Just came home?" path, the Outside Sangha path, the "you're not alone" message, and making a person visible throughout the experience.

Constraints improved the design

Designing for a basic telephone forced us to remove unnecessary interaction. That simplicity ended up making the experience calmer.

Privacy is not just a compliance feature

For people coming from highly institutional environments, collecting less can itself be part of building trust. Our rule became:

Collect what the bridge needs, not everything the technology could collect.

Accessibility

Accessibility is the starting point of Ring the Bell, not an add-on: the core experience needs no screen at all. The phone line works entirely by voice and keypad, so it runs on any phone, including flip phones with no data, and callers never have to speak out loud, which helps people who are hard of speech or who have no privacy. Every option is read slowly, with pauses, in English or Spanish. On the web, we used Atkinson Hyperlegible, a typeface designed by the Braille Institute for low vision, with large 19px text and 44–64px touch targets. Every text color pair meets WCAG AA contrast in both light and dark mode. Buttons are labeled for screen readers, captions and status updates are announced live, keyboard focus is always visible, and the page language switches with English and Spanish. Every spoken line has a caption, every practice is also available as text, reduced-motion settings are respected, and there are no timers, streaks, or scores. Search forgives typos and half-remembered names, so people can find a practice even if they don't remember exactly what it was called.

What's next

The next step is not more features. It is deployment with Boundless Freedom Project:

  • review and approve the practice content
  • record practices in facilitators' and alumni's own voices
  • native-speaker review of the Spanish
  • configure real staff coverage and routing
  • run a small pilot with people transitioning from an inside BFP program to the outside community
  • measure whether human callback requests actually connect

Longer term, Ring could connect directly into BFP's existing reentry and community programs without becoming a separate ecosystem.

Why Ring the Bell

BFP already built the relationship. We are not trying to replace it.

Ring the Bell is designed to keep that relationship from disappearing at the gate.

Technology carries the bridge. People are what's on the other side.

Built With

Share this project:

Updates

Submission history