Founding pilot, five partners
Coaching breaks down between sessions. We fix that.
Tiny Shifts records your coaching sessions, extracts personalized nudges from what your client actually said, and delivers them throughout the week. Every message comes from you. About five minutes of your time per session.
One hour with you. 167 without. One cell is one hour.
A nudge from your session
To: [email protected] • From: Coach Adams
Hey Sarah,
Remember what we talked about with John. Before your 1:1 today, take one breath and ask yourself: what does he actually need from me right now?
You mentioned wanting to lead with calm instead of reactivity. Notice: when did you pull that off this week?
Coach Adams
Three steps. Zero change to your workflow.
Your session happens normally. We handle everything after.
-
01
Your session happens
A silent recording companion joins your Zoom, Meet, or Teams call. It records and transcribes automatically.
-
02
AI extracts personalized nudges
Eight to fifteen nudges are pulled from what your client actually said: their commitments, their language, their breakthroughs. Not generic templates.
-
03
You review, clients receive
Approve, edit, or remove any nudge. Nothing reaches your client without your sign-off. Messages come from you, on your schedule.
What Tiny Shifts is not
We built this to extend your coaching, not replace it.
Not a chatbot
Your clients never interact with AI. Every message they receive comes from you, in your voice, with your name.
Not generic templates
Every nudge is extracted from your actual session. Your client's own words, their specific commitments, their language.
Not a replacement
You are the coach. Tiny Shifts makes sure the work you do in session keeps working during the 167 hours between sessions.
Everything you need. Nothing you do not.
Built for coaches by a team that includes one.
Session-specific extraction
It draws prompts from what your client said, not from the session in general: the commitments they made, the realisation they had, the thing they said they would try. Each one quotes their own language back.
Full review control
You see every prompt before anyone else. Edit the wording, change the timing, delete what does not sound like you. Nothing has ever been sent without a coach approving it, and that is not a setting.
Email and SMS delivery
Email, text, or both, on the cadence you set, from your name and address. Your client installs nothing and logs into nothing.
Client replies come back
One tap to say a prompt helped, or that now is not the time. It is the only signal anyone gets that a prompt landed, and it tells you which to send more of.
Session prep
A brief before each session: where you left off, what was delivered, what they engaged with, the themes that keep recurring. Emailed ahead of the meeting.
Client timeline
The whole arc of one client across every session: commitments, recurring themes, what shifted and when. One view instead of your notes plus your memory.
Client memory that builds
What the system has understood about each person, in plain language, growing session over session. Correct it, pin what matters, delete anything. Not a black box you are asked to trust.
What needs attention
Quiet flags, not a dashboard: a goal open for weeks with nothing moving, a theme raised repeatedly and never addressed, a client who has gone quiet. Deterministic rules, not a guess.
Speaker matching
It works out who said what and matches voices to your client list. Where it is unsure it asks you rather than guessing, because attributing one client's words to another is the one mistake that matters.
Calendar auto-join
Connect Google Calendar once and sessions are picked up automatically. Video meetings only, and exact-name matching, so an ambiguous match is skipped rather than risked.
Or skip the bot entirely
Not comfortable with a bot in the room? Upload a recording or paste a transcript instead. Same pipeline from there on.
Client portal
Optional. Clients who want to can log in and see what they have received, rather than only meeting it in their inbox.
Your brand, not ours
Your name, your logo, your sending address. Clients see your practice on everything. Tiny Shifts stays out of the way entirely.
Export and delete
Everything for one client, out in a single action, any time. Deletion means deletion. This should be table stakes and often is not.
We went looking for evidence. Some of it did not survive.
Before designing the extraction we ran two research passes over the literature, then tried to refute our own findings: three independent votes per claim, two refutations to kill it. Twenty-five claims were tested. Five were discarded. Here is what held, including the inconvenient parts.
Between-session work is one of the few things with causal support
When the same therapy is compared with and without assigned homework, the version with it does better. What was tested is assign and review as a package, and the reviewing half counts. That is why Tiny Shifts does not stop at sending; it brings the thread back to you before the next session.
The relationship is the largest measured factor, and it survives technology
Nothing measured matters more to outcomes than the working relationship, and it holds up when contact is technology-mediated. Which makes one question the highest-stakes one in the product: does a prompt strengthen the bond with you, or quietly stand in for it? We build on the first answer.
The coaching evidence base is thinner than the industry quotes
The widely cited coaching effect comes largely from studies with no control group, which roughly triple the apparent result. Most rely purely on self-report and almost none measure anything after the last session. We use the controlled figure, not the flattering one.
Fatigue is not the danger. Irrelevance is.
Reminder interventions perform far worse at scale than published trials suggest. The failure mode is not too many messages, it is generic ones. That finding is why extraction is tied to what a specific client actually said, and why four prompts that land beat fifteen that do not.
Stated plainly: the 0.48 finding comes from clinical therapy, not coaching, and transporting it across is unsupported. Only one of these findings rests on a causal design at all. Nothing here licenses a claim that using Tiny Shifts improves outcomes by any particular amount, which is exactly what the pilot exists to find out.
How the pilot works
We are selecting five founding partners for a hands-on, 60-day pilot. You get full access and direct attention from the founders. We get your honest feedback.
Apply
Tell us about your practice. Two-minute form.
Onboard
30-minute call with our founders. We set everything up for you.
Use
Full platform access for 60 days. All features included. Direct support line.
Shape
Your feedback directly shapes what we build next. Check-in at day 30, debrief at day 60.
What pilot partners receive:
- Full platform access, every feature, no restrictions
- Concierge onboarding with both founders
- Elisha's coaching reinforcement framework recording
- Starter prompt library matched to your coaching niche
- Direct technical support from the builder
- Founding partner rates locked after pilot
Meet the team
The methodology comes from clinical practice. The platform comes from production engineering. Both are in the room.
Elisha Goldstein, PhD
Coaching & Methodology
20 years of clinical practice. Five published books on mindfulness-based behavior change. Uses Tiny Shifts with his own clients every week. The methodology behind the platform comes from his work.
Nicholas Crabill
Platform & Technology
Built the entire platform. Handles technical operations, onboarding support, and all platform development. Your direct line for anything technical during the pilot.
Nuri Djavit
Go-to-Market & Strategy
Shapes the business strategy and market approach. Ensures the pilot serves both partners and the product roadmap.
"When our system went offline temporarily, multiple clients reached out asking what happened to their nudges. They'd already built a dependency on the reinforcement after just a few weeks."
Early signal from our first users
Involuntary product removal, the strongest form of validation
Hey this is our client reviews
TemporaryPlaceholder reviews until the real ones are in. None of these is a real client.
“Placeholder review. A sentence or two about what changed between sessions once the nudges started arriving.”
“Placeholder review. Something specific the coach noticed: a client who did the thing they said they would, on the day they said it.”
“Placeholder review. Something about the messages sounding like the coach, not like software.”
Privacy and security
In progress: compliance infrastructure expanding for the therapy marketEncrypted
Data encrypted in transit and at rest
No AI training
Your data is never used for model training
Auto-delete
Recordings deleted after extraction
Coaching-safe
Built for coaching. HIPAA path planned for therapy expansion.
Common questions
What is a design partner?
You get full platform access and direct founder attention: onboarding, support, and strategy. We get your honest feedback to shape the product. It's a partnership, not a transaction. You're helping build the tool alongside us.
What do I need to get started?
Zoom, Google Meet, or Microsoft Teams for your coaching sessions. That's it. We handle everything else during a 30-minute onboarding call.
How long is the pilot?
60 days. We check in at day 30 and do a full debrief at day 60 where we review data, share what we learned, and discuss whether to continue at founding partner rates.
Do my clients know it's AI?
Messages come from you, in your voice, with your name and branding. With white-label enabled, clients see your practice name on everything. How you explain the process to clients is up to you. We recommend transparency, but the tool itself is invisible.
What happens after the pilot?
Founding partners who continue get their pilot rates locked in. Standard pricing after launch will be higher. You also get the opportunity to co-author the first case study if the pilot produces measurable outcomes.
Is client data safe?
Session recordings are automatically deleted after nudges are extracted. All data is encrypted in transit and at rest. We never use your data for AI model training. We're building toward full HIPAA compliance for therapy market expansion.
How much of my time does this take?
About five minutes per session to review and approve nudges. Your session itself happens normally with zero workflow changes. The recording companion joins silently and handles everything after.
Join the founding pilot
Five spots. 60 days. Direct access to the founders. Tell us about your practice.