KDDI's strategy is convergence — au connectivity, au PAY and Ponta points, au Jibun Bank, and now half of Lawson — but its service estate answers each product separately; an agentic frontline co-engineered with KDDI's own AI investments (its majority stake in Japanese-LLM startup ELYZA, its AI data-center build-out) can resolve a billing, points and banking question in one keigo-grade conversation, making convergence real where customers actually feel it.
KDDI operates the au, UQ mobile and povo brands across premium, value and online-only tiers, with au PAY and Ponta forming one of Japan's largest points-and-payments ecosystems.
Public factKDDI and Mitsubishi Corporation jointly control Lawson, with KDDI holding 50% since 2024 — telco-konbini convergence is a publicly declared strategic bet.
Public factKDDI took a majority stake in ELYZA, a Japanese-LLM startup, in 2024, and is investing in AI data centers including the converted Sakai plant site — AI capability is a stated corporate direction.
Public factau Jibun Bank and au Financial products attach banking and insurance service load to the same customer base the telco already serves.
Public factMulti-brand service means povo's app-only model, UQ's value base and au's premium base each hit different channels with different cost profiles — a single orchestration layer arbitrages across them.
Reasoned inferenceLawson counter services (payments, tickets, parcel handling) likely generate service questions that today have no good remote channel at all.
Seller hypothesis — validateValidate with the account team before outreach: Where ELYZA and internal AI assets are mandated in customer-facing stacks · Actual contact volumes and cost per contact by brand (au / UQ / povo) · Lawson integration roadmap and whether counter-service load is in scope · Incumbent CCaaS/BPO vendors and contract timelines
Strong internal engineering — sell infrastructure, models, governance, or selected capabilities
Evidence: KDDI is investing directly in Japanese LLMs (majority stake in ELYZA) and AI data centers, so it will insist on its assets in the stack; but a production agentic voice frontline with telephony, governance and evaluation is not what ELYZA provides — co-build with declared roles for KDDI's models is the credible posture.
Why they won't build the full stack: KDDI's AI investments target model capability and infrastructure, not contact-center orchestration; assembling telephony, keigo-quality speech, offer governance and 100% evaluation internally would take years while service costs compound across three brands and a konbini network — the co-build delivers frontline economics now.
KDDI holds 50% of Lawson alongside Mitsubishi Corporation and has framed telco-retail convergence as core strategy.
GTM implication: Pitch the frontline as the convergence layer — the point where au, Ponta and Lawson stop being separate service experiences.
KDDI acquired a majority stake in Japanese-LLM startup ELYZA and is investing in large AI data-center capacity, positioning AI as a growth pillar.
GTM implication: Architect the proposal with explicit slots for KDDI's AI assets — this is a co-build sale to a buyer that wants its investments used and showcased.
| Workflow | Why it matters here | Value | Complexity | Speed | Channels |
|---|---|---|---|---|---|
Billing, plan and au PAY / Ponta service desk One conversation resolves bill, plan and points across brands with writes to the systems of record; cost per contact drops on the biggest queues. | Billing, plan-change and points questions dominate contact volume across three brands and tens of millions of lines. Friction today: au, UQ and povo route to different support models; Ponta and au PAY questions span ledgers a human agent swivel-chairs through; the premium base expects keigo voice. | VoiceApp chatWeb | |||
Churn-risk retention across brand tiers At-risk subscribers offered the right internal tier in-channel before external port-out; saves attributed against the dialer baseline. | The three-brand architecture exists to catch down-shifting customers — but only if the down-shift conversation happens before the port-out. Friction today: MNP signals fire faster than outbound capacity; cross-brand saves (au to UQ) require humans who know both catalogs. | VoiceSMSApp chat | |||
au Jibun Bank everyday banking line Routine banking resolved in-conversation with human gates on regulated decisions; the converged customer gets one service standard. | A branchless bank inside a telco group needs phone service that meets banking compliance at telco cost. Friction today: Routine balance, transfer and card questions queue for limited human capacity under banking-grade rules. | VoiceApp chat |
Voice & channel orchestration, telephony, conversational execution, session/state, routing, integration build. Capability coverage validated during implementation.
API access to these systems is the critical-path dependency.
Identity-bound sessions, policy-bounded actions, 100% audit logging, human approvals at defined points, in-tenant intelligence.
All figures are modelling estimates from the labeled inputs above — nothing here is customer-provided yet. The pilot's first job is replacing these assumptions with KDDI (au)'s measured baseline. Package price covers implementation only; recurring usage billed separately.
Why this package for KDDI (au): The value is cross-BU by design — mobile, payments, points and banking share customers — and KDDI's multi-brand estate means multiple workflows and integration domains from the start: Transform scope, entered through the billing desk.
Packages cover implementation and integration only. Recurring costs are billed separately: Tilicho Labs platform usage (~$0.15/call-min indicative, usage only), Google Cloud consumption, telephony/carrier charges, managed operations, and support & optimization. No package includes unlimited usage.
“The Lawson bet and the AI data-center build-out are declared strategy; an agentic frontline across au, Ponta and banking is where that strategy becomes something tens of millions of customers can feel — and a reference for selling AI capability to enterprise clients.”
“This respects your AI investments: ELYZA and KDDI models where they fit, Gemini-class voice and the communications layer where quality demands it, one governed orchestration layer over BSS, Ponta and banking APIs.”
“Three brands and a konbini network multiply service surfaces faster than any staffing plan; elastic capacity that serves povo chat and au keigo voice from one platform is the only model that scales with convergence.”
“Cross-brand questions are your hardest routing problem; an agent fluent in all three catalogs resolves them in one conversation and keeps the down-shift inside the house.”
“APPI-grade data handling across telco, points and banking domains, human gates on regulated decisions, 100% logged — governance stronger than the fragmented per-brand estate you audit today.”
“At millions of contacts monthly across brands the automation delta is structural margin; the pilot prices per conversation against your own blended cost, with no platform license bet up front.”
Written from this account's own research — the strategic signal, the capability gap, the entry workflow, and the modelled economics — not a mail-merge template. Pick the moment and the persona, edit anything, then copy or open in your mail client.
Customer-safe by construction: drafts are composed only from customer-facing fields. Account tier, build-vs-buy classification, priority score, internal routing, and partner-commercial detail are not inputs to the composer, so they cannot appear in a draft. Money figures are always framed as modelled from the customer's own volumes. Read before sending — you own what goes out.
Japan's #2 carrier runs au, UQ and povo across every price tier, owns half of Lawson, and is investing directly in Japanese LLMs and AI data centers — enormous frontline volume plus a declared AI agenda make it a high-stakes co-build.
Workshop with consumer-business CS leadership and the AI strategy team: map where ELYZA/KDDI models are mandated, baseline cross-brand contact volumes, and scope a billing-desk pilot on the au premium queue.
Entry: Billing, plan and au PAY / Ponta service desk · Transform package · 14–20 weeks phased across units. Human fallback throughout; success thresholds agreed before build.
Start the pursuitResearch-based priority-account universe assembled from public information, market scale, communication volume, and solution fit. This is NOT an authoritative list of top Google Cloud customers; existing Google Cloud relationships are noted only where publicly reported. Validate every account with the account team before outreach.