Company overview
Wysa
B2B2C · Clinical AI · Founded 2015
India / UK / US
$30M
total raised
~$20M ARR
estimated revenue
5M+ users
NHS-listed · CBT/DBT model
Woebot
B2B Clinical DTx · Founded 2017
San Francisco, USA
$123M
total raised
$24.6M ARR
confirmed 2023
Stanford RCT
clinical anchor · FDA pathway
Youper
B2C Consumer Wellness · Founded 2017
San Francisco, USA
$3.5M
total raised
~$725K ARR
estimated revenue
~5 employees
plateaued · cautionary case
Replika
B2C AI Companion · Founded 2014
San Francisco, USA
$10.7M
total raised
~$35M ARR
estimated revenue
Millions of users
2023 trust crisis · companion AI
Saati (target)
B2C → B2B · Physical Robot · Launching Oct 2026
Atlantic Canada
Grant
funding model (ACOA/IRAP)
Pre-revenue
Oct 2026 youth app launch
Psycho-spiritual
VAI model · robot + app
Funding & revenue comparison
Total funding raised (USD millions)
Wysa $30M
Woebot $123M
Youper $3.5M
Replika $10.7M
Estimated annual revenue (USD millions)
Replika ~$35M
Woebot $24.6M
Wysa ~$20M
Youper $0.7M
Strategic capability radar — 6 dimensions
Score 1–10 per dimension · Saati scores reflect planned target capability
Wysa
Woebot
Youper
Replika
Saati (target) — dashed
Feature & capability matrix
| Capability |
Wysa |
Woebot |
Youper |
Replika |
Saati |
| Clinical validation (published research) | Yes | Yes | No | No | Planned Y1 |
| Persistent memory across sessions | No | No | No | Yes | Yes |
| Physical presence / robot form factor | No | No | No | No | Yes |
| Voice-first interface (no typing required) | No | No | No | Partial | Yes |
| Human coaching / spiritual care layer | Yes | No | No | No | Planned |
| Enterprise / institutional B2B channel | Yes | Yes | No | No | Planned Y2 |
| 6-dimension wellness scoring engine | No | No | No | No | Yes |
| Monthly wellness report (email) | No | No | No | No | Yes |
| Adaptive content routing by wellness score | No | No | No | No | Yes |
| Crisis auto-escalation protocol | Yes | Yes | No | No | Yes |
| Offline capability (rural resilience) | No | No | No | No | Yes |
| Privacy-first (no data selling, Canada-hosted) | Yes | Yes | Yes | Partial | Yes |
| Correctional / justice system access | No | No | No | No | Yes |
| Culturally specific care (immigrant, Indigenous) | No | No | No | No | Yes |
| Psycho-spiritual therapeutic model (VAI) | No | No | No | No | Yes |
The three models in this market
Model 1: Clinical AI
Wysa · Woebot
Institutional distribution. Clinical validation. Enterprise healthcare contracts. Slow to start — durable and high-value at scale. Requires $20M+ in capital to reach meaningful size. NHS / Stanford partnerships as credibility anchors.
Model 2: Consumer wellness
Youper (and many others)
Easy to launch, hard to scale. Low monetization ceiling. No institutional moat. App store organic discovery is not a growth strategy. Youper is the cautionary case: good product, no distribution = plateau regardless of quality.
Model 3: AI companion
Replika
Highest engagement of any model. Highest ethical risk. Emotional dependency liability. Regulatory scrutiny (Italy ban 2023). Product governance is everything — the 2023 trust crisis shows what happens when it fails.
What actually decides outcomes — 4 key patterns
01
Distribution beats product
Wysa and Woebot scaled via NHS and healthcare systems. Youper had a comparable product and stagnated at $725K ARR. Same core technology — completely different outcomes. Institutional channels are the only durable growth engine in this market. App store discovery does not scale.
02
One study opens every door
Woebot's 2017 Stanford RCT is still cited 8 years later and opened every healthcare partnership the company has ever made. Clinical credibility is a moat that compounds over time and cannot be quickly replicated by competitors. Budget for research in Year 1.
03
B2C is a proof vehicle, not a business
Both Wysa and Woebot discovered consumer apps are a validation mechanism, not a monetization engine. The pivot to enterprise healthcare transformed both companies. Wysa made this move in 2019 and the NHS partnership in 2020 confirmed it. Make the pivot early — not after you need the revenue.
04
Emotional trust is non-negotiable
Replika's 2023 feature removal caused genuine grief among users who had formed real emotional bonds. In this market, product governance is an ethical responsibility. Every feature decision that affects the companion relationship requires a communication protocol. Breaking trust with vulnerable users is not a recoverable event.
Key risks by company — what to avoid
Wysa risks
LLM commoditization eroding chatbot differentiation
No memory — #1 user complaint since 2018
Repetitive patterns, scripted "therapy speak"
Woebot risks
Heavy FDA regulatory burden — slow and expensive
$123M raised vs $24.6M revenue — long path to profit
Weak consumer brand awareness
Youper risks
No institutional distribution — permanent ceiling
Underfunded relative to competitors
Undifferentiated product in a crowded market
Replika risks
Trust crisis 2023 — ongoing reputational damage
Regulatory exposure (Italy ban, data concerns)
No clinical pathway — cannot access institutional buyers
Saati's position — what none of these companies have
The combination below does not exist anywhere in the current market
Each item below is a gap that all four analyzed companies leave open — simultaneously.
Therapeutic depth
30 years of contemplative practice + 6 years of crisis chaplaincy. The therapeutic model is the product — not replicable by a technology team.
Physical presence
Voice-first robot for seniors who cannot use smartphones. A population none of the four analyzed companies serve at all.
Institutional access
Correctional Service Canada relationship + 4-province nonprofit network. Distribution Wysa spent $30M to approximate from scratch.
Persistent memory
RAG-based session continuity from day one. The #1 user complaint against Wysa since 2018 — and still not resolved in 2026.
Adaptive wellness
6-dimension scoring + adaptive content routing + monthly email reports. No competitor has implemented this system.
Population scope
Youth → Seniors → Correctional. Three severely underserved populations no single company currently addresses together.