Test the recommendation before the board does.
You have a working recommendation and four weeks. Kirat lets Claude put it to ChatGPT and Gemini as well: three model families take positions independently, list the risks you may have missed and rate your slide titles. You get every position and every objection in writing. It is a decision aid, not a stamp of approval, and the recommendation stays yours.
Client data first. Everything you give Kirat goes to OpenAI and Google, under your own API accounts and their terms. Check your NDA before you use it on a client project, and anonymize what you send, as in the recorded runs below: client name withheld.
Three recorded runs, one for each tool.
Real recordings of Kirat's tools, made on 14 October 2026 with the version you buy, on questions and tasks like yours. Pick a tool and watch the council work. This is a replay: nothing you click here calls a model, and nothing leaves this page. The chat is written for illustration and only restates the recorded result; what Claude sent to Kirat and what came back are shown unchanged under “Behind the scenes”.
Pick a tool
Let Kirat vote on this: Our client wants to enter the US market. Which entry route should we recommend to the board?
kirat_voteasked the council
The council has decided (unanimous 3:0): Exclusive distributor partnership first.
All three positions, mine included, point the same way. Each one comes with its reasoning, and ChatGPT ties its vote to conditions.
Behind the scenes
kirat_voteWhat Claude sends to Kirat
Question: Our client wants to enter the US market. Which entry route should we recommend to the board?
Context sent: We are a two-person strategy consultancy. The client (name withheld) is a German mid-sized maker of industrial packaging machines: 220 staff, EUR 48 million revenue, 70% from the EU, about 6% from the US today through occasional direct deals. Machines cost EUR 300,000 to 900,000 and need on-site service within 48 hours. Two US competitors dominate the market; one large US distributor has approached the client. The board wants a recommendation in four weeks and can invest up to EUR 4 million over three years. No one at the client has US management experience.
Options: Own US subsidiary with a sales and service team · Exclusive distributor partnership first · Postpone US entry, grow in the EU first
Result from the council
Recorded on 14 October 2026 with kirat_vote at the standard effort tier; the session cost under two cents. Question, options and context are shown exactly as sent, the reasoning in full. Gemini answered in German; its answer is translated here. Confidence values are each model's own self-report.
Brainstorm with Kirat: what could we be missing before we recommend the distributor partnership to the board?
kirat_brainstormasked the council
The council came up with 15 ideas and rated each one anonymously. The top three:
- Exclusivity as a Revocable Option (19/20 points)
- The Shelfware Test (18/20 points)
- Keep a Deliberate Crack in the Channel (18/20 points)
Behind the scenes
kirat_brainstormWhat Claude sends to Kirat
Topic: Risks and blind spots to check before recommending a US distributor partnership to the board of a German packaging machine maker
Context sent: We are a two-person strategy consultancy. The client (name withheld) is a German mid-sized maker of industrial packaging machines: 220 staff, EUR 48 million revenue, 70% from the EU, about 6% from the US today through occasional direct deals. Machines cost EUR 300,000 to 900,000 and need on-site service within 48 hours. Two US competitors dominate the market; one large US distributor has approached the client and wants exclusivity. Our working recommendation is an exclusive distributor partnership first. The board can invest up to EUR 4 million over three years; no one at the client has US management experience. HARD: the analysis must be finished within four weeks by two consultants. HARD: no paid market studies above EUR 10,000.
Result from the council
- 1
Exclusivity as a Revocable Option
Map the hidden legal and commercial asymmetry of exclusivity: territory, customer ownership, house accounts, online leads, term, renewal, change of control, underperformance, post-termination rights, and applicable state dealer-protection laws. Recommend staged exclusivity that expands by region or vertical only after leading indicators are met, with audit rights, customer-data access, and clean termination mechanics.
- 2
The Shelfware Test
Test the hostile hypothesis that the distributor wants exclusivity partly to neutralize a potential competitor rather than build the brand. Review portfolio conflicts, salesperson compensation, lead-allocation rules, historical launches, reference customers not selected by the distributor, and whether promised sales require displacing products that generate better commissions; translate findings into minimum pipeline, demo, staffing, and revenue milestones.
- 3
Keep a Deliberate Crack in the Channel
Break the soft working assumption that the first move should be fully exclusive: reserve several named strategic accounts or one vertical for direct client ownership while the distributor covers the rest. This is worth the friction because a fully exclusive model can deprive the manufacturer of US pricing knowledge, customer relationships, failure data, and bargaining power precisely when it lacks US management experience; compare this structure with full exclusivity before the board decision.
- 4
48-Hour Service Fire Drill
Run two unannounced tabletop drills using real customer locations and realistic failures: one near a service hub and one in a remote industrial region. Require named technicians, dispatch routes, qualification evidence, spare-part availability, escalation paths, and actual historical response data; make exclusivity conditional on passing and on measurable service SLAs with remedies.
- 5
US Compliance and Liability Tripwire Sprint
Use targeted legal and technical expert calls—not a broad paid market study—to identify deal-killing requirements around NRTL/UL expectations, OSHA and local authority acceptance, guarding standards, importer-of-record duties, product liability, insurance, remote-access cybersecurity, tax nexus, and distributor indemnities. Produce a red-flag register showing owner, cost range, lead time, and whether each issue must be resolved before the first order or installation.
- 6
The EUR 4 Million Leakage Model
Build a downside-first cash model by customer cohort rather than relying on headline distributor margins. Include discount stacking, commissions, demonstrations, US-specific engineering, warranty travel, technician training, spare-parts inventory, freight, duties, insurance, bad debt, foreign exchange, and delayed acceptance payments; stress-test whether a seemingly asset-light partnership can still consume the full investment envelope before repeat sales emerge.
- 7
Reverse audit of terminated European partners
We specifically contact European machine builders that have ended contracts with exactly this US distributor in the past. This lets us systematically identify hidden patterns of conflict, such as an insufficient stock of spare parts or extortionate renegotiations over exclusivity. This method delivers extremely valuable insider insights within 2 weeks, without third-party costs.
- 8
Spare-Part Hostage Simulation
This is a one-step mutation of the 48-Hour Service Fire Drill: substitute a technician-dispatch failure with a critical-part failure. Trace one high-risk component from Germany to a US machine under a weekend outage, including stocking ownership, customs delay, obsolescence, emergency freight, and cannibalization rules; the result reveals whether the promised service level depends on economically unrealistic inventory.
- 9
Mandatory US interim general manager
We break the soft constraint of keeping the existing management line by recommending that the board immediately hire an experienced US industry manager for 3 years. Without US management experience in-house, even steering an exclusive distributor is flying blind. Part of the EUR 4 million budget secures this key position from day 1.
- 10
Service-capacity stress test
Model how many service calls the distributor's technicians could answer within 48 hours at 10, 25 and 50 installed machines, and where the client's own engineers would have to fly in, so the board sees when the partnership stops carrying the service promise.
- 11
SLA stress test of the 48-hour service coverage
We check the US distributor's actual technical responsiveness through covert reference checks with its existing customers in remote US states. Because machines worth up to EUR 900,000 become extremely expensive when they stand still, an insufficient service network at the distributor leads directly to recourse claims against the German manufacturer. This blind spot can be uncovered within 4 weeks without expensive studies.
- 12
Trojan-horse analysis of competitive entanglement
A check of whether the US distributor uses the exclusivity to sideline the German product as an expensive add-on while pushing its own or rival products. We analyze the distributor's portfolio and hold anonymous sales conversations with its representatives. This protects the client from completely blocking its own access to the US market for years.
- 13
Service joint venture instead of pure exclusive distribution
Mutation of the distributor recommendation: the distributor receives only the sales rights, while the 48-hour service is handled through a jointly founded service company in which the client holds 51%. This secures direct proximity to customers and protects the service margin. It turns the risk of losing control into a manageable growth model.
- 14
Acquisition of a regional US service company
We break the soft constraint that an exclusive distributor has to be the first step. Instead, we use up to EUR 2 million of the EUR 4 million budget to buy a small existing US maintenance company with a network of technicians outright. This is guaranteed to deliver on the 48-hour service promise and gives the client full independence from distributors.
- 15
US product liability and litigation exposure
Analysis of the financial damage risks from US product liability laws for complex packaging machines. Because the board lacks any US management experience, the liability risk from operating errors or injuries is often dramatically underestimated. We bring in specialized US lawyers under a fixed short mandate below EUR 10,000.
All ideas, complete and in ranked order. Claude's own idea takes part anonymously and on equal terms. Topic and context were sent in English; ideas the members wrote in German are translated here.
Let Kirat write variants of the action title for our recommendation slide. Max. 14 words, no consulting jargon.
kirat_createasked the council
The winning variant (17/20 points): "Enter the US step by step, validating demand before committing major capital"
My own draft came in at #3; the members rated all three anonymously.
Behind the scenes
kirat_createWhat Claude sends to Kirat
Brief: Action title for the recommendation slide in a board presentation on US market entry for a German maker of industrial packaging machines. Audience: the client's board (CEO, CFO, owners). Goal: state the recommendation and its main reason so the board can decide. Format: one slide title, in English.
Criteria: max. 14 words · states the recommendation and the main reason · no consulting jargon · no claims beyond the analysis
Result from the council
- 1
"Enter the US step by step, validating demand before committing major capital"
This gives the board a clear course of action and the financial reason for it.
- 2
"Establish a US sales entity to capture growing demand for automated packaging machinery."
The title links a concrete operational recommendation directly to the central market driver and does without consultant-speak entirely.
- 3
"Enter the US through a distributor first: service coverage from day one"
The orchestrator's own draft.
Titles are shown as written, in English, as the brief asked. The reasoning in Claude's entry is Kirat's own label for it. Reasoning written in German is translated here.
A recommendation nobody argued against is untested.
After three weeks with the same numbers, you and your partner share the same blind spots.
A real second opinion costs a colleague a day, if anyone has a day.
The board's first question is often the one nobody on the team asked.
Not just a yes: conditions, and what to check.
Claude votes with your working recommendation as one of three members. ChatGPT and Gemini take positions independently; if they disagree, the result says so, with the reasoning, and if there is no majority, Kirat says so (needs_human). Brainstorm turns the same context into a ranked list of risks and checks, with your own idea rated anonymously alongside the others. The result shows that the recommendation was challenged from more than one direction. It does not prove it right; that part stays yours.
Let Kirat vote on this: Our client wants to enter the US market. Which entry route should we recommend to the board?
kirat_voteasked the council
The council has decided (unanimous, round 1, $0.017): Exclusive distributor partnership first. Confidence: ChatGPT 91, Gemini 85, mine 65.
ChatGPT ties exclusivity to conditions: minimum sales, trained service capacity, spare-parts inventory, customer-data access and termination rights.
Brainstorm with Kirat what we could be missing before the board sees it.
kirat_brainstormasked the council
15 ideas, rated anonymously. Top-rated (19/20 points): Exclusivity as a Revocable Option, with staged exclusivity, audit rights and clean termination mechanics.
My own idea, the service-capacity stress test, came 10th.
FIG. 1 — The vote and the brainstorm recorded above, continued as one chat. The chat is phrased for illustration; the figures and quotes are from the recorded results.
One purchase for Claude Desktop and Claude Code.
No subscription, no seats, no account with us. Check where Kirat runs before you buy:
Where Kirat runs
Works in
- Claude Desktop chats on macOS and Windows
- Claude Code in a terminal, in VS Code or in Claude Desktop's Code tab
- Claude Code cloud sessions, after a one-time setup per cloud environment With limits: everyone who uses the environment can read your keys, and the daily budget counts per session.
Does not work in
- Chat in the browser (claude.ai) or in the Claude mobile apps
- Cowork tasks that run in the cloud
- Some company setups: networks that require an HTTP proxy, and organizations that restrict Desktop extensions to an allowlist Ask your IT team first.
- 01Claude Desktop extension and Claude Code plugin, in one download
- 02Setup script for Claude Code cloud sessions
- 03All nine tools: vote, code review, idea ranking, copy critique, image comparison and more
- 04Cost control: a daily budget ($5 by default) that holds back calls which could exceed it. Best effort, so also set limits at the providers
- 05License for one person, on your own devices and cloud environments
- 06Compatibility updates when possible — free, but no guarantee
Runs in Claude Desktop (macOS/Windows) and Claude Code; Claude Code cloud sessions with a one-time setup — not the browser or mobile chat. Where it runs · Cloud limits
Get KiratOne-time purchase · Payment via Lemon Squeezy (merchant of record), VAT handled at checkout · Refunds: see Terms
API usage is paid to OpenAI and Google directly, under your accounts — $0.02–$0.15 per council decision in our measurements.
Kirat is not for you if…
- —you use Claude only as a chat in the browser or on your phone. Kirat runs in Claude Desktop (macOS/Windows) and Claude Code, and in Claude Code cloud sessions with a one-time setup — not in those chats.
- —you need a hosted service, phone support or guaranteed response times.
- —you don't want API accounts and billing at two providers.
- —you make a handful of decisions a month and are happy to compare answers in separate chats by hand.
- —you want an answer that sounds certain. A council's job is to show you the split, and sometimes the split is the answer.
- —your data may not go to OpenAI or Google. Everything you give Kirat goes to both, under your accounts and their terms.
Before you buy.
Can I use it on client projects under NDA?
That depends on your NDA and on the providers' terms, not on us. What we can tell you precisely: Kirat has no server, no account with us and no telemetry. What you give it goes directly to OpenAI and Google, under your own accounts — from your computer, or in Claude Code cloud sessions from your own cloud environment through Anthropic's network proxy for that environment. Its local data — the daily cost counter, the cost log, the last 20 results (they contain your questions and answers), the log of delegated decisions and generated images — sits in the .kirat folder in your home folder and can be deleted any time.
Compared with using Claude alone, Kirat adds two recipients: OpenAI and Google. If your NDA covers sending this material to them, you are within it. If it doesn't, Kirat doesn't fix that. Keep those projects out of the council.
Does a unanimous vote mean my recommendation is right?
No. Agreement among three models is agreement, not proof: they can be confidently wrong together, and they only know what you send them. The value is in the conditions, the dissent and the list of things to check.
In the recorded vote above, ChatGPT agreed but tied its vote to five conditions, and the brainstorm on the same case produced fifteen checks the recommendation has to survive.
Can I show the result to my client?
You can, but it is working material, not a deliverable. It records what you asked, what each model answered and what it cost. Keep it in the project file as a record that the recommendation was challenged from more than one direction; what the board reads is what you wrote.
Kirat keeps only the last 20 results in the .kirat folder in your home folder, so save the ones you need with the project.
One purchase, many clients?
For your own work, yes. A license covers one person, on any number of that person's own devices and own cloud environments, and you can use what you produce with it in client work.
It does not allow redistribution: you can't pass Kirat itself on to a client or a colleague. Each person on your team who runs it needs their own license, and a cloud environment shared with others needs a license for each person. The code is proprietary.
The details, for the record.
Everything else, in full: what Kirat does not do and further questions. Open what you need.
What this does not do
- —It does not make your recommendation correct. Different model families fail differently, but they can still be confidently wrong together. A 3:0 vote is agreement, not proof, and the confidence values are the models' own self-report.
- —It is not a data-protection measure. Kirat has no server, and we never see your keys or your data. But everything you give it (questions, context, code, images) goes directly to OpenAI and Google, under your accounts and their terms — from your computer, or in Claude Code cloud sessions from your own cloud environment through Anthropic's network proxy for that environment. Minimize or redact client data first, exactly as you should with the AI tools you already use.
- —It is not a deliverable. The result is working material. What the client reads is what you wrote.
- —It is not an archive. Kirat keeps the last 20 results in the
.kiratfolder in your home folder. Save the ones a client may ask about with the project. - —It does not run everywhere you use Claude. Only where Claude runs tools on a machine that has Kirat installed: Claude Desktop chats on macOS and Windows, Claude Code in a terminal or VS Code, local sessions in Claude Desktop's Code tab, SSH sessions — and Claude Code cloud sessions (Code tab cloud sessions, Claude Code on the web or in the mobile app) after a one-time setup of your cloud environment. The cloud is limited: everyone who uses the environment can read its environment variables, including your keys; Kirat's daily budget counts per session, so set spending limits at OpenAI and Google; background jobs and recent results end with the session; and the environment must allow two hosts,
api.openai.comand the Kirat download host. If the setup fails, the session still starts, just without Kirat, and the failed state can stay cached for new sessions until you edit the setup script. Not chat in the browser or the mobile apps, and not Cowork tasks that run in the cloud. WSL sessions and local Cowork tasks are not tested. Networks on your own computer that require an HTTP proxy are not supported. Organizations that restrict Desktop extensions to an allowlist block manual installs; ask your admin. Where it runs, in full.
More questions
Can I bill it to clients?
You pay OpenAI and Google directly for what you use, so the API cost is an expense on your own provider accounts like any other. Whether you bill the analysis time is between you and your client. We don't have a "billable AI hours" story to sell you.
Where do the API keys live?
You enter them in Claude's own install or settings dialog. Claude Desktop stores them in the operating system's secure storage (macOS Keychain, Windows Credential Manager). Claude Code stores them in the macOS Keychain, and on Windows and Linux in ~/.claude/.credentials.json, readable only by your user. In Claude Code cloud sessions, they are environment variables of your cloud environment, readable by everyone who uses it.
We recommend a separate provider project with a spending limit. Kirat's budget guard also blocks calls that could exceed a daily budget ($5 per UTC day by default, adjustable), but it is best-effort — set limits at the providers too.
What if the council is wrong?
Then you have three wrong opinions and a record of them. That is better than one wrong opinion and no record, but it is not a safety net.
Read the dissent first. It is the part a single model reviewing its own answer tends to smooth away. The evidence section on the main page shows a real 2:1 run, dissent included.