Kashi v5 deck (EN long, 14 slides) · 2026-06-11 · V3 doctrine aligned. Built from synthesis-brief.html findings (Phase D). All 4 dimensions on dedicated slides: Problem (S2-3) · Moat (S7) · Why not GPT (S8) · Success (S11). Inter + IBM Plex Mono. ⌘P → landscape → save as PDF. Companion: .pptx with speaker notes · synthesis-brief.html · V3 doctrine
01 / 14 Kashi

Before silence becomes resignation.

Structural pattern detection for workplace teams.
Enterprise B2B · Japan-first · designed for global compliance.

Justine Acaylar · Founder · Alchemist Enterprise B2B Interview · 2026
02 / 14 The visible symptom

Japan now has resignation agencies.

Employees pay third parties to quit their job on their behalf.
At first it sounds like a strange trend.

It's the visible symptom of dialogue breakdown inside organizations — by the time HR notices, the best person has already mentally left.

03 / 14 The problem

More communication. Less visibility.

Companies communicate more than ever — more meetings, more Slack, more tools. But they cannot see when communication is starting to break down.

Sign 01

People stop speaking up.

Sign 02

One person dominates.

Sign 03

Same issue recurs unresolved.

Sign 04

Friction nobody can name.

04 / 14 The solution

Kashi reads the shape of how a team talks — never the content.

Input

Meeting transcripts + metadata. Used once.

Compress

Turn-taking · response latency · floor-share asymmetry · interruption direction.

Compare across windows

30 / 90 / 180-day patterns surface where a single meeting looks like noise.

05 / 14 The product · 3 role-bounded views

What you see depends on your role. By design, not by promise.

ExecCross-team
CHANNEL · 90D · STATE Sales A OK Eng OK Sales B WATCH PJ-α WATCH
Cross-team trend · k≥5 anonymization
MgrManager Mirror
YOUR TEAM · 90D A 32% B 9% ↓ C 28%
Private mirror · own team only
MemMember view
YOUR PARTICIPATION · 90D team avg band YOU Reference info · not evaluative
Own participation only · no comparison
06 / 14 The boundary · built in at the data layer

The limits are physical, not a policy.

What we see What we don't Why
Team-level structural changeMeeting content · individual utterancesNo Payload — not stored
Longitudinal pattern shiftEmotion · mood · psychological stateEU AI Act Annex III §4 high-risk scope
Role-scoped reportsIndividual score · HR decisionSQL + RLS + contracts exclude by design
07 / 14 Moat — 3 architectural reasons LLMs cannot match

Architecture is the enabler. The moat is what compounds on top of it.

1 · Auditable · Reproducible · Content-Blind

LLMs are probabilistic, opaque, content-reading by design. For regulated workplace use, those aren't feature gaps — they're regulatory deal-breakers.

2 · Longitudinal baselines

Persistent state + role-bounded storage + k-anonymity + audit logs. A weekend wrapper can store state; it cannot reproduce validated, customer-specific baselines tied to documented manager actions.

3 · Doctrine in the data layer

SQL + RLS + contracts physically exclude individual scoring, ranking, HR decisions. A prompt can be rewritten next month. Ours can't be — without rebuilding the product.

08 / 14 Why won't GPT / Claude / Gemini eat us?

A weekend wrapper copies the interface. It cannot reproduce the renewal-proven asset.

What a GPT wrapper CAN do in a weekend

Run rule-based feature extraction on transcripts.
Preserve inputs + versions.
Generate an audit trail.
Mimic the role-bounded views.
Even add tenant isolation + RLS.

Microsoft can do all this beside its Viva product, using meeting telemetry it already owns.

What it CANNOT instantly reproduce

Validated longitudinal baselines — customer-specific, tied to documented management actions and renewal outcomes.

Governance evidence proving the architecture wins procurement.

Reseller-defended distribution through 産業医 · 社労士 · org-dev coaches.

Honest framing of the bet: employers and workers will not trust the meeting-platform owner to govern the diagnostic layer.

09 / 14 Why now — Japan first

Three forces converging at the same moment.

Loss

Rising turnover cost.
Taishoku-daiko (resignation-agency) services — itself a visibility signal.

Obligation

MHLW Power Harassment Prevention Law expanded to SMB in 2022. ~3.67M Japanese firms newly liable.

Regulation

EU AI Act Annex III §4 classifies workplace AI monitoring as high-risk. Kashi was designed to comply from day one.

10 / 14 Market — Tokyo knowledge-work SMB

Sized for ambition. Validated by PoC conversion.

TAM

¥164.2B
Total (model)

SAM

¥57.5B
Serviceable (model)

SOM

¥720M
3-yr ARR target
11 / 14 Success — 3 phases, each gated on real proof

Phase 1 succeeds on renewal, not on pilot count.

Phase 1 · 12 months

5 → 1
paid pilots → ≥1 renewal
+ Signal-to-Action SLA ≥80% (documented manager actions within 14 business days) · NPS ≥7.

Phase 2 · 24–36 months

50
paid customers
30 via 産業医 / 社労士 / coach reseller channels. 10,000 meetings/mo analyzed. Annex III §4 compliance verified by independent counsel.

Phase 3 · 5 years

Default
trusted lane
When workplace-AI regulation lands across EU and Japan, Kashi is the compliance-grade dialogue-diagnostics default — because of the day-0 architecture choice.
12 / 14 Traction — where we are

Pre-pilot. Honest about it. The engine works.

v1
Product
Auditable, reproducible analyzer · demo tenant live
908
Tests passing
Including a determinism-lock test
25%
Reply rate
Customer discovery · ~28 outreach contacts
Pilot partners
Conversations underway; 産業医 / 社労士 / coach co-authors recruiting
13 / 14 Founder fit

Justine Acaylar

"People don't leave suddenly. They go silent first."
14 / 14 Three asks of an acceleration partner

From an acceleration partner, three things.

Before silence becomes resignation.
Thank you.