The AI Companies Building Japan's Sustainability Reporting Stack

Search “AI企業” in Japanese and you mostly get recruiter pages and generic startup lists. You won’t find the three companies actually shipping AI into Japan’s sustainability reporting cycle right now, with named clients and dated press releases. The mandate forcing this market open is real and close. Japan’s Financial Services Agency has set mandatory SSBJ disclosure for Prime-market companies above ¥3 trillion market cap starting with fiscal years ending March 2027, with ¥1 trillion-plus companies following in 2028 and ¥500 billion-plus in 2029 (FSA, Japanese Roadmap on Sustainability Disclosure and Assurance, April 2026). Add the EU’s CSRD, narrowed since the March 2026 Omnibus to companies over 1,000 employees and €450 million turnover, and most large Japanese subsidiaries of EU-linked groups are still caught. Finance and sustainability teams in Tokyo are shopping for tools on a deadline.

Cost is part of why the search volume exists. Most companies now under CSRD expect to spend more than €100,000 a year on reporting once the more than 1,000 ESRS data points are in scope, and 83% of them name data collection, not writing, as the actual bottleneck (Novata; Workiva survey, n>2,200). SSBJ compliance in Japan runs on the same underlying problem: getting scattered operational data into a form a reporting framework can use. That’s the job every one of the three companies below is chasing a piece of.

Here’s who’s actually selling into that moment.

Fujitsu: benchmarking, not drafting

In May 2026, Fujitsu launched Sustainability Disclosure Navigator, an AI service built with Deloitte Tohmatsu that analyzes and benchmarks non-financial disclosures against more than 1,000 Japan-listed companies (Fujitsu Global press release). Fujitsu is the vendor here. It compares a client’s disclosure against the market. The release doesn’t say Fujitsu uses the tool on its own filings. Fujitsu’s stated roadmap extends toward AI-drafted disclosure documents, but the May 2026 launch is a benchmarking product. The Deloitte Tohmatsu partnership matters here: Fujitsu is supplying the software, Deloitte Tohmatsu is supplying the advisory judgment on what “good” looks like against 1,000-plus peers, and that split is the closest thing to a verification layer in the three tools covered here, short of a formal assurance engagement.

aiESG: the one with a named client

aiESG Inc. runs a generative-AI service called aiESG for IR that evaluates integrated reports and ESG disclosures. The company’s own February 2024 press release names a real, specific adopter: Kyushu Electric Power. “本サービスは、既に九州電力株式会社にご採用いただいております” (“This service has already been adopted by Kyushu Electric Power,” aiESG’s phrasing) (aiESG / PR TIMES). That’s aiESG describing its own client. Kyushu Electric Power hasn’t said this in its own voice, as far as we found. Still, it’s a dated, sourced, named example, which is rare in “AI for ESG” coverage. By February 2026, aiESG had updated the product with a proprietary LLM and 104-language support, moving it from evaluation into a full evaluation-to-production workflow (aiESG).

HEROZ x NZAM: scoring, not preparing

HEROZ, known for shogi-engine AI before it moved into enterprise work, partnered with Norinchukin Zenkyoren Asset Management (NZAM) to build a multimodal-LLM agent that auto-evaluates listed companies’ ESG disclosures against 57 criteria. It went into business use in June 2025 (HEROZ press release). This is the investor side of the market. NZAM is scoring what issuers publish, not helping issuers prepare it. Keep it separate from Fujitsu and aiESG, which sell to the companies doing the disclosing.

None of this is happening in a vacuum. Global platforms like Workiva and Persefoni are also selling into Japan, and we’ve laid out how they stack up against each other separately. What’s different about Fujitsu, aiESG, and HEROZ x NZAM is that they were built for this market specifically, with Japanese-language disclosure and the SSBJ/TCFD-to-ISSB path as the starting assumption rather than a localization afterthought. That’s worth something on its own, independent of feature comparisons.

What to check before you sign anything

If you’re a finance or sustainability lead actually evaluating one of these three, or the global platforms we’ve compared elsewhere, the question isn’t “does it use AI.” All three do. What matters is which step of the pipeline the AI sits in, and what happens after it. Fujitsu’s output tells you where you rank against the market. aiESG’s tells you how the report itself reads and holds up. HEROZ x NZAM’s tells an outside investor what they already think of you, after the fact. None of that produces a number a third-party assurance provider will sign off on for FY2028, when SSBJ third-party assurance starts landing on the same companies now building their FY2027 disclosure. Ask a vendor directly where the audit trail lives, not just where the AI output lives. Those are two different files in most of the products above.

The pattern across all three

Line the three up and a shape appears. None of them does full end-to-end report preparation with third-party assurance built in. Fujitsu tells you where you stand against the market. aiESG helps you produce and refine the report itself. HEROZ x NZAM sits outside the whole process, scoring what’s already public for an asset manager’s own investment decisions. All three launched or updated within the last eighteen months, with named clients and dated releases behind them. But each one solves a narrower piece of the problem than “AI writes my SSBJ report and an auditor signs off on it.”

That gap gives context to a number PwC published from its 2025 global survey of roughly 500 executives across 40 countries: 28% of companies now use AI for sustainability reporting, up from 11% a year earlier (PwC, via ESG Today). It’s a global number, not a Japan-specific one. Read it as “AI is somewhere in the workflow,” not “AI is producing the filing.” The market behind that shift is sized at roughly $1.3 billion in 2025, growing toward $7.4 billion by 2034 globally, with Asia-Pacific estimated at about 26% of that and Japan alone around 6% (Fortune Business Insights).

Where the gap actually sits

Every tool above does drafting, benchmarking, or scoring. On the public record, none does independent third-party verification that the underlying numbers are right before they go out the door. That’s a separate function from writing good prose about the numbers, and FSA’s own roadmap flags it as coming due for larger issuers on its own timeline (FSA roadmap). That’s the specific space Socious Report is built for. AI drafts the CSRD, SSBJ, or ISSB-aligned report from one underlying dataset, and Socious Verify provides an independent credential on top of it, instead of the same system checking its own homework.

We’re not going to tell you a company you’ve heard of is running on Socious Report. We can’t verify that claim any more than we’d want a competitor inventing one about us. What we can offer is a free CSRD/SSBJ readiness check, so you can see where your own data stands against the FY2027 and FY2028 timelines before you’re shopping for a vendor under deadline pressure.

By the time the ¥1 trillion tier of Japanese companies hits its own SSBJ deadline in 2028, expect more entrants in this list, not fewer. The three above are the ones with a public paper trail today. Whoever your finance team is talking to next quarter, ask the same question we’ve asked here: which part of the pipeline does the AI actually touch, and who checks it after.