Frequently asked questions

It is a tool that puts your business strategy through an AI council before it reaches real decision-makers. You submit a brief, three AI models draft competing strategies from different roles, they blind-review each other, and a chairman model synthesizes a final version. Selected stakeholders then role-play objections in a stakeholder review, producing an objection table alongside the final document. Consultants, shindanshi (certified SME management consultants, 中小企業診断士), and business owners in Japan can use it to pressure-test strategy before a board or client meeting.

A single ChatGPT answer is one model's first guess; this runs that guess through an adversarial process before you see it. Three models with different roles draft independently, blind-review each other's work anonymized as Response A/B/C, and a chairman model synthesizes the strongest ideas. Role-played stakeholders then attack the result with their sharpest objections, and the chairman revises again. You get a document that has already survived a round of scrutiny, not just a first draft.

The Japan edition is built for foreign founders and executives doing business in Japan, Japan entities of foreign companies, international staff in Japanese companies, and their advisors, including shindanshi (certified SME management consultants, 中小企業診断士). Use it to prepare for Japanese subsidy reviews, bank and JFC (Japan Finance Corporation) loans, and internal approvals. It is not designed for programs outside Japan.

Yes, you choose Japanese or English as the output language for each brief. The Japan edition supports Japanese approval practices. The interface and document languages are independent: work in English and choose Japanese output for Japanese approvers. You can write the brief in either language regardless of the chosen output. If you need the other language later, you can rerun the council with the language switched.

It means your strategy gets debated by a council of AIs before it ever reaches a human boardroom. Multiple models draft, critique each other, and stakeholder roles attack the plan with objections, all before you present anywhere. This reduces the chance of being blindsided by a weakness in the actual meeting. The final call is still yours; the council just stakeholder reviews the material first.

They are draft, blind review, synthesis, and stakeholder review. In Stage 1, three models independently write drafts from different roles such as a skeptical investor or an operator. In Stage 2, each model critiques and ranks the drafts anonymized as Response A/B/C so no model favors its own work. Stage 3 has a chairman model combine the strongest elements into one final strategy. Stage 4 role-plays each selected stakeholder raising their toughest objection, and the chairman revises the document and adds an objection table.

Stakeholders are roles you ask the AI to role-play so it can raise realistic objections, like a CFO or a sales director. You pick the roles most likely to push back in real life, and the AI generates each one's strongest objection from that perspective. The chairman model then responds to each one and updates the strategy. Note that a run with more than six stakeholders costs two runs instead of one.

It lists every objection raised by a stakeholder, whether it was valid, and how the strategy responded. Validity is marked Yes, Partly, or No, paired with either a change made to the plan or a prepared response. It effectively becomes a ready-made Q&A sheet for the real meeting. The table is included as part of the final strategy document.

Yes, the brief has a known-objections field, and the council factors those in while drafting. Surfacing concerns you already know about up front sharpens the later stakeholder review. The field is optional, but it's worth filling in if you already know what's coming. You can also raise a new one later from Refine, e.g. typing "CFO objection: ..." to rerun just that stakeholder review.

Yes, it uses Anthropic Claude, OpenAI GPT, and Google Gemini via OpenRouter rather than one model wearing different hats. Each has different training and tendencies, so even with similar roles they produce genuinely different drafts. That diversity is the core mechanism for reducing single-model blind spots. Details on which models are used live on the AI policy page.

The chairman role is filled by one of the models as part of each run's synthesis step, informed by the blind-review rankings from Stage 2. It is not always the same single model locked in forever; the exact composition can vary by run. What matters is that every model involved, including the chairman, is a commercial AI accessed via OpenRouter. The AI policy page lists which providers are in the mix.

No. Presets model how a role typically thinks; they are a starting point, not a substitute for asking the real people. They help you surface likely objections before the meeting so you are not caught off guard. They cannot know the specific person's situation or internal politics, so always confirm important points directly. The objection table makes a good Q&A sheet for that conversation.

No, your input is not used to train any AI model. Briefs and related materials are sent to US-based AI providers for processing only, and not used for training. The subprocessors page details how each provider handles data after processing. For sensitive material, we still recommend referring to people by role rather than by name.

Your brief content is sent to US-based AI providers such as Anthropic, OpenAI, and Google for processing. Processing is not confined to Japan, so this is worth knowing before you submit anything sensitive. The subprocessors page lists each provider and how data is handled. Avoid entering anything you wouldn't want to cross borders.

No, please don't enter personal data at all. Refer to people by role, such as CFO or sales director, instead of by name. A personal-data detector scans your brief and warns you before the run starts if it spots something that looks personal. If you get a warning, replace the flagged text with a role description before continuing.

It scans your brief before submission and flags things that look like personal names or identifying details, but it's a safety net, not a guarantee. It won't catch every possible case, so a final manual check on your side is still worthwhile. Even if no warning appears, it's worth re-reading for names or contact details. Replace anything flagged with a role description before running.

In Settings, choose auto-deletion N days after each brief's creation. This deletes briefs and related results, not the workspace after inactivity. Workspace deletion is separate and has a 30-day grace period. Backups expire under their retention schedule; billing records may be kept as legally required.

Request workspace deletion from Settings, and it will be permanently removed after a 30-day grace period. You can export a data ZIP beforehand if you want to keep a copy. During the grace period, you can cancel the deletion and restore the workspace. After the grace period ends, the data cannot be recovered.

The file is read in your browser. Extracted, masked text is used for the review and removed within 24 hours. The review result itself is private and expires 14 days after it's generated. The free review can be used once every 30 days. Download the result before it expires if you want to keep it.

There are five tiers from a free trial up to Enterprise. The free trial gives 3 runs with a watermark on exports; Solo is ¥14,800/mo for 15 runs, 1 seat, and your own logo; Pro is ¥39,800/mo for 50 runs, 3 seats, white labeling, and client share links. Firm is ¥128,000/mo for 200 runs, 10 seats, and batch CSV upload, and Enterprise starts at ¥300,000/mo with a DPA and annual invoicing on request. All prices are tax-excluded, with 10% consumption tax added.

These don't use a run: (1) fixes made by selecting text and commenting (light edits, with the same free allowance and hourly limit as chat refinement); (2) the revised v2 after a rejection or hold, and the final version that reflects only the conditions after a conditional approval (once per decision); (3) runs that stop partway, fail or time out are not charged — the run is returned automatically and you can re-run it once for free. A full finishing pass on the whole document uses a run as usual.

Yes, you can buy extra runs at ¥800 each (excl. tax), ¥880 incl. 10% consumption tax. If you hit your monthly limit, buying more lets you keep going without interruption. Keep in mind a run with more than six stakeholders consumes two runs instead of one. Upgrading to a higher plan is another way to raise your limit going forward.

No, read-only Viewer seats are free and don't count against your paid seat limit. Only owner, admin, and member roles, which can edit or create content, count as paid seats. If you just want a client or manager to see results, invite them as a Viewer at no extra cost. Roles can be changed anytime from team settings.

Invoices and receipts are issued in English by our payment provider, Dodo Payments, and are not qualified invoices with a Japanese registration number. Please consult your tax advisor about input tax credits.

Payments are processed by Dodo Payments as the merchant of record, so your statement may show a Dodo Payments related name rather than Market Leap Partners Inc. directly. Receipts are also issued by Dodo Payments. Invoices and receipts issued by Dodo Payments are in English. Please also review the Dodo Payments Buyer Terms (https://dodopayments.com/buyer-terms) and Dodo Payments Privacy Policy (https://dodopayments.com/privacy-policy). This is a standard setup for SaaS billing and doesn't change what you're being charged for. Contact info@marketleappartners.com with any billing questions.

You can cancel anytime from Plans & billing, and there are no cancellation fees. Access continues through the end of the current billing period after you cancel. You can resubscribe to the same plan anytime later. Enterprise annual contracts have their own terms, so reach out directly for those.

Sign-in is Google-only; there's no separate password to set up. Team invites are also sent to a Google account email address. Either a company Google Workspace account or a personal Gmail account works. After Google sign-in, enter your company or business name and review the required agreements and business-use confirmation. Marketing is optional.

Writing the brief itself typically takes about 10 minutes; the council run on top of that takes a bit longer since it involves multiple AI calls across four stages. It's slower than a single chatbot reply, but you get a reviewed, stakeholder review document in return. You can watch progress on screen while it runs. Once it's done, Refine lets you make further revisions without starting over.

The core fields are goal, market, constraints, resources, timeframe, and the decision needed, plus optional extra context, key numbers, known objections, and stakeholders. You don't have to fill every optional field, but being specific improves the output. The guide on writing a strong brief walks through a concrete example. Drafts can be saved and returned to later.

Yes, you can upload a CSV with multiple briefs and run them as a batch. Up to 3 run at the same time; the rest queue up and process in order. This batch upload feature is available on the Firm plan and above. Finished results can be downloaded together as a ZIP.

Refine gives feedback to the chairman model, which rewrites just the relevant parts, a lighter-weight revision. Rerunning the full council restarts all four stages from scratch. Use Refine for wording or number tweaks, and a full rerun when something fundamental about the brief has changed. Refine keeps version history (v1, v2, ...) with a diff against the previous version.

You can export as Markdown, PDF, Word, or HTML, and batch runs bundle everything into a ZIP. Paid plans support your own logo; the trial exports a watermarked PDF only. The output is formatted to be usable directly as an internal ringi (formal written approval request) or a client proposal. If you want to tweak wording after exporting, revise it with Refine first, then export again.

Yes, it can suggest relevant charts, but only using numbers you've already entered in the brief. Any chart involving a projection is labeled "Illustrative estimate," and it will not invent numbers that weren't in the brief. Suggested charts are editable so you can adjust how they're presented. If you haven't entered any numbers, chart suggestions won't be available.

They are ready-made profiles for 81 roles across 10 panels — CFO, main bank, subsidy reviewer and more — with their KPIs, frameworks, typical objections and decision power. In the stakeholder review each role argues from that lens, so the objections are more realistic. You can add one role or a whole panel such as Internal approval or Bank loan. Presets contain roles only, never personal names.

Yes — “Customize” creates a workspace copy you can edit freely. Add context such as “CFO, but our company is debt-averse” and the role will argue with that assumption. The master preset stays unchanged and other companies are not affected. You can also add your own roles from scratch.

Yes, you can draft subsidy plans and review stakeholder objections for Japanese programs here, but it does not guarantee approval and isn't affiliated with any subsidy office. Setting up stakeholders that mirror likely reviewer concerns helps surface weak spots ahead of submission. Treat it as drafting support and verify all figures and claims yourself before filing. Follow the specific subsidy program's required format and content rules.

Yes, it's useful for drafting and stakeholder review a plan ahead of a bank loan review, but it does not guarantee approval and isn't affiliated with any bank. Setting a CFO or loan officer role as a stakeholder helps catch weak repayment assumptions or soft numbers in advance. Always verify the accuracy of your figures yourself. Adjust the final submission to match the format required by your bank in Japan.

Yes, Pro and above support white-label export with your own logo and brand color. You can also generate a client-facing share link so recipients can view the result without logging in. The deliverable doesn't need to mention MLP Strategy Council at all. Consultants commonly use this to hand finished material straight to clients under their own name.

Yes, you can keep multiple briefs and strategies running in parallel within one workspace. Each client's strategy can be created separately with its own share link. You can also invite team members and split work by client. Since plans have limits on strategy count and seats, heavier caseloads are a good fit for the Firm plan.

Yes, you can choose a ringi (formal written approval request) or project proposal layout at export time. The objection table is included as-is, which makes it easy to drop into an internal approval package. Export as PDF or Word and you can fit it into your company's standard format with minor adjustments. A cover page with your logo and brand color can be added too.

Yes, you can upload a PDF, DOCX, or TXT file up to 10 MB, or paste text, and get a review. The result includes the top 5 weaknesses, objections broken down by stakeholder, and a 0–100 readiness score. Keep in mind the free review can only be used once every 30 days, and results are private and expire after 14 days. For multiple client cases, consider using a paid plan for ongoing work.

You can use the generated strategy for your own business without restriction. That said, accuracy is not guaranteed, so review the content before relying on it. Full terms are in the Terms of Service. The same applies when you pass the output on to a client as a deliverable.

No, accuracy is not guaranteed. Treat AI output as reference material and always have a human review it before acting on it. This is especially true for numbers and factual claims, which you should verify independently. The same caution applies before submitting anything to a subsidy office or bank.

A combination of Anthropic Claude, OpenAI GPT, and Google Gemini is used via OpenRouter. Which model plays which role at each stage can vary somewhat between runs. Under our agreements, none of these providers uses your data for training. Full details are on the AI policy page.

You may publish it for your own business purposes, but since accuracy isn't guaranteed, fact-check and review it carefully first. Extra caution is warranted for figures or claims that could affect third parties. For external documents like white papers or press releases, running it through your normal internal review process is a good idea. See the Terms for the full conditions.

The output is reference material with no accuracy guarantee, so responsibility for the final decision rests with you. Always review the content before anything important, and consult a professional where appropriate. The same applies to subsidy or loan outcomes, which are never guaranteed by using this tool. See the disclaimer section of the Terms for full details.