01 / 16
Kashi
Before anyone
says a word.
Structural conversation analytics for the workplace · B2B SaaS · Japan
Justine Acaylar · Founder · 2026
02 / 16
The problem
Trouble in a team shows up long before anyone talks about it.
Resignations, conflict, and disengagement only surface after the damage is done.
The early signs are visible months earlier, in how meetings actually run — but no tool reads them.
03 / 16
What existing tools miss
Every existing tool relies on what people say.
Engagement surveys depend on how honestly people put it into words.
1-on-1s and hotlines are records of issues that have already surfaced.
Meeting AI summarizes a single meeting; it can't connect dots across time.
04 / 16
The insight
Distress shows up in the structure of conversation — before anyone names it.
- A specific person's floor share suddenly drops.
- A specific person's questions get no response.
- The same person's proposals are repeatedly deferred.
In one meeting it looks like noise. Across many meetings, it's a pattern.
05 / 16
The solution
Kashi reads the shape of how a team talks — never the content.
- Transcripts → structural features (turn timing, response latency, floor-share asymmetry, interruption direction)
- Raw content is not retained past a short TTL
- 30 / 90 / 180-day comparison · role-based reports
- Designed to visualize team-level dynamics — not used for individual evaluation.
06 / 16
The product
Three views, bounded by role.
Executives
Cross-team view — which teams need attention now.
Managers
Manager Mirror — your own team's structure, mirrored back to you.
Members
Your own participation pattern — never anyone else's.
07 / 16
Information boundaries
Strict boundaries on who sees what — built in, not bolted on.
Shown
- Team-level structural change
- Long-term pattern shifts
Not shown
- Meeting content
- Individual utterances
- Emotion or mood inference
- Individual scores · HR decisions
08 / 16
Why now
Three forces converging — at the same moment.
Loss
- Rising turnover cost
- Rise of taishoku-daiko (resignation-agency) services — itself a signal of invisibility
Obligation
- MHLW Power Harassment Prevention Law (パワハラ防止法) expanded to SMB in 2022
- ~3.67M Japanese firms now legally obligated
Regulation
- EU AI Act Article 5 restricts workplace emotion-recognition
- APPI (Japan privacy law) retention rules tightening
09 / 16
Differentiation
Our moat isn't a feature. It's how the product is built.
- Content is not stored for evaluation.
- Patterns surface only across many meetings — single-meeting noise doesn't trigger anything.
- Each role's view is built into the data itself.
- No emotion, no sentiment, no voice-tone inference.
10 / 16
Where we are — honestly
Pre-pilot. But it's running.
- Working deterministic analyzer · production demo tenant live
- 908 tests, including a determinism-lock test · stress-tested at 50 meetings / 5,000 turns
- Wave 1: 28 discovery DMs · Phase 0 interviews underway
- No paid PoCs · no customer logos · no sales team — yet. That's what the next 6 months are for.
11 / 16
6-month validation plan
Three phases, with explicit failure signals.
Months 1–2
Locate the pain and the buyer.
50 customer interviews.
Months 3–4
Does the output drive real decisions?
Open 3–5 PoCs.
Months 5–6
Convert and prepare Seed.
3–5 paid + Seed round prep.
12 / 16
Initial customers
Start where the obligation is sharpest.
- 50-to-500-person, meeting-dependent SMBs in Japan
- Fast decisions · management feels the pain directly · under-served by heavy enterprise HR suites
- The buyer is the CEO / COO — not HR. Kashi sits on organizational-restructuring, retention, and manager-development budgets.
13 / 16
Market · initial hypothesis
Sized for ambition. Validated by PoC conversion.
TAM
¥164.2B
Total market (model)
SAM
¥57.5B
Serviceable market (model)
SOM
¥720M
3-year ARR target
14 / 16
Business model
Three-step ladder into an annual contract.
- Light Pilot · ¥100,000 — entry validation
- Standard PoC · ¥400,000 · 3 months — co-develop 90-day comparison, display granularity, legal acceptability
- Enterprise Diagnostic · ¥800,000+ — annual conversion
15 / 16
What we're asking for
Three asks of an acceleration partner.
- North-American market-fit validation — HR / Organization Development practitioner contacts to verify the structural angle with non-JP buyers
- Large-enterprise PoC introductions — within the program's enterprise network, channels we cannot reach unaided
- AI governance, privacy & legal review mentoring — refine the regulation-aligned posture for US and EU enterprises
16 / 16
Team · the closing ask
Justine Acaylar
Keio University · solo founder — product, code, sales, operations.
6 years on a multinational SMB board, watching dialogue break down up close.
Filipino-Japanese — exactly the multicultural dialogue Kashi is built for.
Initial concept collaboration: Haseru Ida (ICU, theoretical linguistics, non-executive).
"We never judge what your teams say.
We help them hear how they say it."
Validation partners · Mentoring · Seed introductions