senior-bridge isn’t magic. It’s an SMS companion powered by a language model, and like any thoughtful friend, it can only sound personal if you tell it who it’s talking to.
Every signal in the profile — nickname, tone, the things they love, the things they don’t want to discuss, a passing detail about a grandchild — gets folded into the prompt that composes each message. More signal in, more soul out. Less signal in, more “Hi Margaret, how are you doing today?” out.
Most caregivers fill in the bare minimum, then wonder why the messages feel generic. This guide shows what a thoughtful setup looks like, using a fictional senior named Margaret as our worked example. Map your own loved one onto her as you read.
You can update any of this at any time. The bot’s voice will shift within a few days.
Same person. Same morning. Two very different texts.
❌ From a thin profile
Loves: family, walks. Style notes: (none).
Good morning, Margaret! ☀️ Hope you have a wonderful day today!
✅ From a rich profile
Loves: grandkids Lily and Owen, slow walks at Ault Park, the smell
of basil from her summer garden, calling her sister Joan in Cleveland on
Sundays. Style notes: speaks plainly, uses lots of exclamation points,
doesn’t want pity.
Morning Maggie!! Cardinals were singing their heads off here at sunrise — bet your feeder’s hopping. Today’s a good day for a slow loop around Ault Park if the hip’s behaving. Joan call coming up Sunday too — anything you want to ask her about?
The first message is friendly. The second sounds like a friend. The difference is the profile, not the model.
Each field below shows what most people write, what works better, and a one-sentence note on how the bot uses it.
The display name is what shows up in your dashboard. The nickname is what the bot actually calls them in messages — and this is the one that matters.
❌ What most people write
Display name: Margaret Olsen
Nickname: (blank)
✅ What works better
Display name: Margaret
Nickname: Maggie
💬 Why: The bot opens messages with the nickname. “Maggie” lands like a friend; “Margaret” lands like a dentist’s office. Use what her family actually calls her.
These are about delivery, not personality. Get them right, then move on.
+1 and area code.
e.g. +15135551234. Locks once verified.💬 Why: A 7:00 AM “good morning” sent at 4:00 AM is not a good morning.
A 5-digit US zip. Powers the daily weather context, location-aware research (interests, local events), and tiny grace notes like “supposed to rain this afternoon — keep that umbrella handy.”
❌ What most people write
(blank)
✅ What works better
45208 (Margaret’s zip in Cincinnati)
💬 Why: Without it, the bot can’t reference weather or anything local. Generic messages stay generic.
Four warm presets. Pick the one closest to her real personality. You’re not picking a brand voice, you’re picking the vibe of the friend writing the text.
✅ For Maggie: warm playful. She uses lots of exclamation points and doesn’t want pity. The bot will match her energy.
💬 Why: Tone shapes sentence length, word choice, and how much “small talk” the bot allows itself before getting to the point.
Catholic, Protestant, Jewish, Muslim, Buddhist, spiritual (non-denominational), or None. None is a perfectly valid choice — the bot will simply skip all religious content.
When set, faith influences the optional morning spiritual element (a brief verse, teaching, or reflection) and any holiday wishes. The bot never substitutes one tradition’s content for another — a Buddhist senior will never receive a bible verse.
✅ For Maggie: Catholic / Christian. She attends Sunday mass at St. Cecilia’s and loves morning verses. Toggle on “Include a morning spiritual element” and the bot will weave a short verse into roughly half her morning messages.
💬 Why: A 73-year-old who’s been Catholic her whole life will notice if her morning text suddenly quotes Rumi. Pick the right tradition, or pick None.
This is the single biggest lever for personalization. Comma-separated. The bot reads this list every time it composes a message and pulls from it whenever it can do so naturally.
Generic Loves give you generic messages. Specific, sensory, story-shaped Loves give you messages that feel like a friend who knows them.
❌ What most people write
family, walks, gardening
✅ What works better
her grandkids especially Lily and Owen, slow walks at Ault Park, the
smell of basil in her summer garden, Catholic radio in the kitchen,
calling her sister Joan in Cleveland on Sundays, watching the cardinals
at her bird feeder
💬 Why: “Family” is a noun. “Her grandkids especially Lily and Owen” is a scene the bot can write into. Names, places, sensory details — those are gold. Aim for 5+ specifics.
Equally important. Comma-separated. Things the bot will never bring up.
❌ What most people write
(blank)
✅ What works better
politics; her son David’s recent divorce (still tender); heavy
medical detail unless she brings it up
💬 Why: A bot that doesn’t know what to dodge will eventually wander somewhere painful. Even one Avoid topic dramatically improves safety. The senior can always raise the topic herself; the bot just won’t initiate.
Voice calibration. Where Loves tells the bot what to talk about, Style notes tells it how she sounds. Specific phrasing matters here — write the way you’d describe her to a new friend.
❌ What most people write
friendly, warm
✅ What works better
Speaks plainly, uses lots of exclamation points, doesn’t want pity.
Recovering from hip replacement in November and is determined to walk
Ault Park again. Match her warmth and energy.
💬 Why: “Doesn’t want pity” alone changes the bot’s behavior across hundreds of messages. The bot will check in on the hip without making it the center of every conversation.
Background the bot reads but never recites verbatim. This is the biographical scaffold — who’s in her life, what’s recent, what shapes how she’s feeling this season.
✅ For Maggie:
Widowed in 2019. Three kids: David (Cincinnati, going through a
divorce), Anne (Lily and Owen’s mom, lives 20 min away), and Mark
(Seattle, calls every other Sunday). One sister, Joan, in Cleveland —
they talk most weekends. Catholic, life-long, attends St. Cecilia’s. Hip
replacement Nov 2025 — recovery going well, walking with a cane
indoors.
💬 Why: The bot uses this to anchor references — when Anne calls, the bot knows that’s the daughter, not a friend. When Maggie mentions “Joan,” the bot knows that’s her sister in Cleveland, not someone new.
Personal Interests is a premium / trial feature. Up to 5 topics, max 2 daily. The bot researches each one in the background and weaves a fresh tidbit into messages — a Reds box score, a saint’s feast day, a backyard bird sighting in her zip code.
This is where setup pays off most visibly. The difference between a good interest entry and a great one is whether it gives the bot a story angle, not just a topic.
❌ Generic: baseball
🟡 Better: Cincinnati Reds
✅ Best: Cincinnati Reds — she’s followed them since the Big Red Machine days, has a soft spot for Joey Votto.
❌ Generic: birds
✅ Best: backyard birds in 45208 — cardinals, chickadees, the occasional cooper’s hawk. She has two feeders on the back porch.
❌ Generic: local news
✅ Best: Cincinnati neighborhood news around Hyde Park / Mt. Lookout — small stuff, not crime or politics.
❌ Generic: music
✅ Best: Catholic radio in the kitchen — Sacred Heart Radio Cincinnati. Old hymns on Sunday mornings.
✅ Best: Catholic Saint of the Day — she keeps a small daily devotional and likes hearing whose feast day it is. (cadence: daily)
Reminders show up in the senior’s local timezone at the time you set, in the senior’s voice — not as a calendar alert. Title, description, and time all feed the bot.
❌ What most people write
Title: Take pills
Description: (blank)
Time: 9:00 AM
✅ What works better
Title: Synthroid 50mcg
Description: Empty stomach, 30 min before breakfast. Bottle’s on the
kitchen windowsill.
Time: 7:30 AM
The detail in the description doesn’t get repeated word-for-word — but it lets the bot phrase the nudge naturally:
Morning Maggie! Synthroid time — windowsill bottle, 30 min before breakfast. ☀️
💬 Why: “Take pills” is what a robot says. “Synthroid bottle on the windowsill” is what a friend says.
Keep it boring. Most seniors do best with:
If a senior is an early riser or a night owl, shift the windows by 30–60 minutes. Don’t push morning earlier than 7:00 AM unless they specifically want it.
If the senior is self-managing their own profile, they can name a single emergency contact. This contact does not see day-to-day messages or get added to the dashboard. They are reached only when senior-bridge detects a serious safety signal (talk of self-harm, a medical emergency, sustained distress) — at which point senior-bridge sends them a brief text and email summary so they can check in.
It’s a safety net, not a feed. One paragraph in the contact’s pocket for the rare day they need it.
Run through this list before flipping Enabled on. Most “the bot feels generic” complaints come from skipping several of these.
If you can tick 13+ of these, you’re ready. Flip Enabled on.
This isn’t magic. The more you tell us, the better we can be.
You can update any field at any time, and the bot’s voice will shift within a few days. If something feels off after launch — too chatty, too quiet, brought up the wrong topic — open the profile, refine the field, save. That’s it.
Questions, feedback, or weird messages we should look at? Reply to
any senior-bridge text with /help or email
[email protected].
Welcome aboard.