Housecarl
Housecarl Arbiter
Operational Open the console →
Solution

Solution

Arbiter is for people who make real decisions on conflicting information — and have to stand behind them afterward. If you've been burned by an AI that sounds sure and is wrong, or you're staring at sources that flatly disagree, this is the tool for that moment. It weighs your sources, settles what the evidence actually supports, and keeps the reasoning on the record. Run investigations from the Arbiter console for a quick scored verdict, or connect it to the AI tools you already use. It starts free, and the top plan costs less than most people's data subscriptions — this is a decision you can make yourself, today. Get started.

For people

Independent Traders

Cut Through the Noise Before You Place the Trade

Analyst upgrades collide with insider sells. Earnings beat expectations but guidance drops. Reddit is bullish, the short report says fraud. You don't have a Bloomberg terminal or a research team — but you're sizing positions anyway. Arbiter weighs every source by reliability and recency, settles the contradictions into a scored verdict, and shows you exactly where the uncertainty lives — before your money is in.

What you get:

  • Every source weighed across news, filings, and social — a filing outweighs a rumor, automatically
  • See where analysts actually disagree, and which side the evidence favors
  • Stale rumors fade out instead of polluting a live thesis
  • A confidence score on every claim — know what's solid versus speculative
  • A full record of the reasoning — retrace exactly why you made the trade

The math that matters: one avoided bad trade pays for years of Arbiter. You already spend more than this on data; this is what tells you whether to believe the data.

Entrepreneurs

Stress-Test Your Thesis Before You Bet the Company

You're building a business case from market research, competitor signals, customer interviews, and your own intuition. Some of it conflicts. Most of it is incomplete. Arbiter lets you lay out every claim, tag each source, and see which parts of your thesis actually hold up — and which parts are built on a single blog post from 2023. When you pitch investors or commit resources, you'll know exactly how strong your evidence is, and so will they.

What you get:

  • Build your thesis from every kind of source — market data, interviews, reports
  • See the gaps — which parts of your argument have no evidence behind them
  • Compare competing narratives about your market side by side
  • Weigh a customer interview against a trend piece, explicitly
  • Reasoning summaries you can drop into pitch decks and board memos

The math that matters: you're about to commit months of runway and your own credibility to this thesis. Arbiter costs less per month than your cheapest SaaS line item — and it's the only one that tells you whether the bet is sound.

Serious Researchers

When You Need More Than a Summary — You Need a Verdict

You've read five articles and they all say something different. You've asked an AI and it gave you a confident-sounding answer with no receipts. You don't need another opinion — you need a system that compares sources, surfaces the contradictions, and tells you what the evidence actually supports. Arbiter is for anyone who refuses to just take someone's word for it.

What you get:

  • Claims from every source, compared side by side
  • See where sources agree, where they disagree, and where nobody has evidence
  • Every conclusion traced back to the sources that support it
  • No hidden logic — every weight and score is visible
  • Works in any domain — politics, health, technology, finance

The math that matters: the free plan covers real use, no card required. If your conclusions matter enough that you'd pay for a single expert opinion even once, a paid plan costs less than that — every month, for every question.

For your tools

Claude

Formal Reasoning Inside Claude — One Connection, No Code

Connect Arbiter to Claude and it gains a reasoning engine it can call mid-conversation. When your question involves conflicting sources, Claude hands them to Arbiter for a scored, defensible verdict instead of synthesizing a guess. Works with Claude on the web, your phone, Claude Desktop, Claude Code, and the API. Five-minute setup guide.

  • Native MCP support — zero glue code
  • Ask in plain language; Claude runs the investigation for you
  • Works on Claude's free plan
  • The full reasoning comes back inline, ready to audit

ChatGPT

Give ChatGPT a Verdict It Can Actually Cite

ChatGPT's MCP support lets it call Arbiter directly. Instead of summarizing five sources into a vague consensus, ChatGPT submits them to the engine and gets back a scored verdict with full provenance — every claim traced, every source weighed.

  • Connects through ChatGPT's standard tool protocol
  • Source scoring and contradiction detection in-conversation
  • Structured results ChatGPT can cite directly
  • No prompt engineering required

Cursor

Defensible Answers in Your Development Workflow

Cursor's MCP support means Arbiter can run inside your editor. Evaluating conflicting documentation, comparing API design tradeoffs, or scoring competing technical approaches — get a weighed verdict without leaving your development environment.

  • Connects directly in Cursor settings
  • Score competing technical claims and design tradeoffs
  • Compare conflicting documentation or migration guides
  • Results inline with your AI-assisted coding

Windsurf

Structured Reasoning Alongside Your AI Pair Programmer

Windsurf's MCP integration brings Arbiter into your coding workflow. When you're weighing architectural decisions or evaluating conflicting library recommendations, get a scored verdict instead of vibes.

  • Native MCP connection in Windsurf
  • Evaluate competing technical approaches with source weighting
  • Defensible reasoning for architecture decision records
  • An audit trail alongside the generated code

REST API

Build Defensible Reasoning Into Anything

The Arbiter REST API gives any system — AI-powered or not — access to the full engine. Submit investigations, score sources, detect contradictions, and get structured verdicts over plain HTTP. Build it into your pipeline, your internal tools, your data platform. Integration guide.

  • Plain JSON over HTTPS — no MCP dependency required
  • Submit investigations with full source and claim detail
  • Structured responses: scored verdicts, contradiction maps, audit trails
  • API key authentication — integrate in minutes
  • Works from any language, any platform, any infrastructure

What it costs — and who can say yes

Arbiter is self-serve. Plans start free and run up to a few hundred dollars a month at the highest volume — current numbers are on the pricing page. There's no sales call, no procurement cycle, no annual contract to negotiate. If you have a card and the authority to buy yourself a software subscription, you have the authority to buy this — and the first verdict that changes a real decision will have paid for it.