Social intelligence for customer-facing AI.
The plug-and-play communication layer for your AI and your teams.
What gets lost today
Your AI reads every word.
Outcomes turn on the rest.
Five observable signals
- Tone of voice
- Pace
- Pauses
- Body language
- Turn-taking
Never reaches the transcript
That’s where sales, renewals and customers are won or lost.
What a transcript misses
Read between the lines
of every conversation
What your LLM gets
“So when does someone actually come?”
Sentiment: neutralWhat PokaMind adds
“So when does someone actually come?”
- Pace speeding up
- Turns getting shorter
- Third time asking
Guidance: offer a human callback now.
How PokaMind plugs in
One layer next to the AI
you already run
What goes in
- Each live interaction
- Playbooks and offers
- The account’s history
- Tone of voice
- Pace
- Body language
- Words
Observable signals only
What comes out
- Satisfaction on every conversation
- Understanding, not just answers
- The right moment to step in
- Trust and loyalty over time
One engine across the loop
Before, during and after
every interaction
- Cooperative
- Hurried
- Firmly resistant
Test before launch
- Turning pointHand over now
Guide it live
- Earlier interactionsChurn risk
Learn what wins
Who we build for
The companies behind
customer-facing AI
The PokaMind layerThree ways in
Use case 1: service and sales agents
Guide every agent while the call is live
For agent platforms, contact centers and BPOs
- Before launchTest against tough callers
- During the callKnow when to offer or hand over
- After the callSpot churn patterns early
Live call: claim after a breakdown
IllustrationCustomer: So when does someone actually come?
Agent: Let me take a few more details first.
PokaMind guidance
Third time asking. Offer a human callback now.
Pre-launch scores
- Cooperative84
- Hurried72
- Firmly resistant46
Why it pays
Upsell, NPS and churn are
decided in the interaction
Turning point read, turned around in time:
- 5-25xmore to win a customer than to keep oneHBR (opens in a new tab)
- 25-95%more profit from 5% better retentionHBR (opens in a new tab)
- ~80%more lifetime revenue from promotersBain, mobile operators
Traction
From research to
real conversations
Research
Spun out of KTH
Backing and press
, 
and 
Customers
, 
, 
and 
Now
The communication layer for customer-facing AI
Career 180, on the observable signals PokaMind reads
Their analysis is deep and informative, drawing on learner responses, tone of voice, facial expressions, and more. They didn’t just add a layer of engagement, they gave us a deeper understanding of how learners are actually grasping the core concepts.
Ali SalehHead of Product, Career 180
What they saw, read together
- Tone of voice
- Facial expressions
- What people answered
- How well it landed
Observable signals only.
The team
Four people behind the engine
- AI and robotics
Dr. Youssef MohamedFounder and CEOPhD in AI and Robotics, KTH. Previously at PAL Robotics and Furhat Robotics. - Retail and operations
Sara ZetterströmFounder and COOBackground in retail management. Ran AI adoption programs at AI Sweden. - AI transformation
Kristina Thim PlumblyStrategic AdvisorAI Transformation Lead at SEB. Former COO at norbloc. - Go-to-market
Tomas YangbynGTM AdvisorCTO and CDO at Marina Group. Former VP Global Services at Pricer.
The rules it works by
Observable signals only
Never reads emotions or mood
Never used to hire or rank
Encrypted, in line with GDPR
Questions partners ask first
What is PokaMind?
PokaMind is the social intelligence layer for customer-facing AI. It reads observable signals like tone of voice, pace and turn-taking, and shows when to step in. It plugs in as one layer next to the AI you already run, and the same layer powers apps for your own teams.
Who is PokaMind for?
The companies behind customer-facing AI: AI agent platforms, contact-center vendors and BPOs, AI and speech API providers, robot makers and hospital networks, and meeting and sales platforms. For agent platforms and contact centers, it guides every service and sales agent while the call is live. For robot makers and hospitals, it keeps the patient in the conversation, through our care robot EMORI. For meeting and sales platforms, it shows where each deal turned and syncs those signals to the CRM.
Where did your last
lost customer turn?
Pick one KPI. We’ll run a sample analysis on your own past interactions and show you, with nothing touching your stack.
Or write to us
A person reads every message and replies by email.