01 · Teach
Your Brand Brain
Save approved facts, preferred tone, claims to avoid, and supporting source notes in each signed-in company profile. All plans can maintain these fields.
AI visibility intelligence · with human control
See what AI says.Know what to do next.
Buyers are asking AI about your market. Arbor shows the answers, the competitors mentioned, and the gaps in how your brand is understood. Keep your facts in Brand Brain and prepare evidence-grounded fixes with Claude. Approved drafts can become unpublished WordPress posts or GitHub draft pull requests in the self-host pilot.
Free report without signup · 14-day cloud trial, no card · Claude drafts and connected WordPress/GitHub actions are a human-approved self-host pilot.Start a trial
01
Monitor AI answers
02
Verify brand facts
03
Approve a proposal
04
Create a draft or PR
A look inside Arbor
From question to proposed action
Buyer-intent prompt
01 · MonitorWhich services help publishers translate books with independent editorial review?
Observation
Brand absent from this answer
Review the raw model response
Brand Brain
Approved facts and brand voice
Keeps fix drafts on-message
Claude action workspace
Awaiting human reviewProposed: clarify product scope, strengthen FAQ and verify sources
Compare with existing content before approving. No automatic publishing.
Illustrative workflow, not a live customer result. Claude drafting is available in the self-host pilot.
New · Brand Brain + connected implementation pilot
From AI visibility to changes you control.
Your company knowledge stays separate from the independent visibility checks. Arbor uses it when preparing draft fixes, so teams can assess proposals against verified facts instead of trusting a generic AI response.
02 · Draft
Claude-assisted fix drafts
Self-host pilot users can request an AI draft for a specific stored result, using its evidence and Brand Brain context. Maximum five requests per day per account; every request is deliberate.
03 · Review
Human sign-off, not autopublishing
Examine the proposed copy and technical notes, compare them with existing content, and approve or reject the exact asset. Approval alone makes no external change.
04 · Implement
Create a draft or pull request
Send approved content to WordPress as a new unpublished post, or put one approved file into a GitHub draft PR. WordPress publication requires a separate confirmation; Arbor never merges a PR.
The action workflow is a limited self-host pilot. Arbor does not promise that a particular edit will change any AI model's recommendations.
Sample run · OrbaOS · 28 buyer-intent prompts · Claude Sonnet
AI Visibility Score
38%
% of buyer prompts that mention you
Website Citation Rate
12%
% of answers that link your site
Competitor Presence
71%
% of answers that name a rival
Accuracy Risk
24%
% of answers misdescribing you
Below the dashboard: every prompt, every raw answer, every competitor mention, and a recommendation per result.
Live research
Explore what AI actually says in your market.
We publish visibility indexes and head-to-head comparisons for SaaS categories — every ranking grounded in buyer-intent prompts run against ChatGPT and Claude, with the raw quotes shown. Browse the open research; if your category is missing, request a free report on your company.
Your competitors are getting recommended in conversations you will never see.
The buyer journey moved.
A growing share of category research now happens inside an AI chat, not a Google search. Most companies have no idea what AI says about them.
Brand > backlinks.
AI assistants weight authoritative descriptions, not domain rating. The companies cited the most are the ones whose category language is consistent across the web.
Drift is invisible.
Models update. Competitors publish. A position that was safe last quarter can be gone before anyone in marketing notices.
How Arbor works
STEP 01
Describe your company (or let AI draft it)
Type your name and website and AI auto-fill drafts sector, target audience, products, named competitors, and key terms from your homepage. Edit anything before you save — the draft is a starting point, not a black box.
STEP 02
Generate prompts by objective
Pick who you are testing for — buyers, investors, job seekers, journalists, partners, or the wider public. Arbor builds questions across the same eight intent types (discovery through authority) so scores stay comparable across lenses.
STEP 03
Run the visibility check
Each prompt is sent to Claude or GPT. Raw answers are stored, then analysed by a second LLM pass for mentions, citations, sentiment, and accuracy risk. On cloud plans, each check counts toward your monthly included allowance; runs that need more draw prepaid credits — you confirm before any credits are spent.
STEP 04
Review the evidence and approve a fix
Inspect every original AI answer. Keep verified facts in Brand Brain, then request a Claude fix draft for review. Self-host customers can send an explicitly approved asset to WordPress as an unpublished post or to GitHub as a draft pull request.
Built differently from the closed-cloud incumbents.
The visibility-monitoring category is dominated by black-box SaaS at £200–2,000 a month. Arbor is the open, auditable alternative.
60-second profile setup with AI auto-fill
Type your company name and website. GPT-4o-mini reads your homepage and drafts the seven-field profile (sector, audience, products, competitors, key terms) for you to review and edit before you save. Most AI-visibility tools make you fill the form by hand on every brand. Available on every plan, including the trial.
Source-available & auditable
The full scoring code is 30 lines of arithmetic, not a black-box index. Read it, run it on your own servers, evaluate before you commit. No incumbent in this category will let you do that.
Predictable cloud usage
Subscriptions include a monthly allowance of visibility checks. Need more in a busy month? Top up prepaid credits in Billing — larger packs lower the cost per check. Before any run uses credits, Arbor asks you to confirm or cancel so there are no surprise charges for that run.
Bring your own API keys (self-host)
Self-host runs on your infrastructure with your Anthropic and/or OpenAI keys and unlimited internal checks. Cloud plans use plan-default models for reproducibility; you choose the subscription tier that matches your quality and volume needs.
Auditable analysis
Every score traces back to a stored raw answer. Re-analyse old runs after editing your profile without re-paying for completions.
Brand Brain, controlled by you
Keep checked product facts, brand voice, forbidden claims, and supporting source notes with each company. This context is used for proposed improvements, not to bias the independent visibility tests.
Approval-gated implementation (self-host pilot)
Claude prepares a draft against the evidence and Brand Brain. After review, self-host users can send approved content to WordPress as a new unpublished post or to GitHub as a one-file draft pull request. WordPress publishing is a separate decision; GitHub merging is always manual.
Pricing
Choose a 14-day free trial or subscribe now and start your paid plan immediately. The trial includes magic-link sign-in, no credit card, 1 profile, 8 prompts per run, and 24 visibility checks total across the 14 days (lifetime cap, not monthly) on OpenAI GPT-4o-mini. Pick a plan or stop at day 14 — nothing auto-charges. After you subscribe, each tier has a monthly included check cap (UTC calendar month; defaults on this site: Starter 400, Pro 2,000, Agency 3,500). Beyond the cap, buy prepaid credit packs in Billing (1 credit = 1 extra check); Arbor always asks before any run spends credits. Self-host is an annual licence with unlimited checks on your own keys.
Cloud Starter
£39
per month
400 visibility checks included per calendar month (UTC). Beyond that, each extra check uses 1 prepaid credit — buy packs in Billing after sign-in; Arbor asks you to accept or cancel before credits spend on a run.
- 3 company profiles
- Manual runs (Anthropic, OpenAI)
- Run history & per-prompt evidence (in-app)
- Weekly automated runs
- Public share linkRoadmap
Cloud Pro
£109
per month
2,000 visibility checks included per calendar month (UTC). Beyond that, each extra check uses 1 prepaid credit — buy packs in Billing after sign-in; Arbor asks you to accept or cancel before credits spend on a run.
- 15 company profiles
- Manual runs across all providers
- Run history + competitor breakdowns
- Daily automated runs
- REST API access (read-only v1)
- Slack & email drift alerts
- Workflow assistant + CSV/PDF export pack → Agency or Self-host
Agency
£309
per month
3,500 visibility checks included per calendar month (UTC). Beyond that, each extra check uses 1 prepaid credit — buy packs in Billing after sign-in; Arbor asks you to accept or cancel before credits spend on a run.
- 50 client profiles
- Everything in Cloud Pro
- Workflow assistant — per-prompt paste-ready prompts for the tested AI (Claude or ChatGPT)
- Report export pack — print/PDF + CSV with full raw answers
- Priority email support
- Multi-workspace + role permissionsRoadmap
- White-label PDF reportsRoadmap
Self-host
£3,000
per year · annual licence
Unlimited visibility checks on your infrastructure and your own API keys — no monthly cloud cap or prepaid credit wallet.
- On your infrastructure, behind your firewall
- Core free to self-host (ELv2) — licence key unlocks the features below
- Unlimited internal profiles & runs — free core, no key needed
- Workflow assistant + report export pack (licence key)
- Brand Brain + manual Claude fix-draft pilot (5 requests/day)
- Approved new WordPress posts + single-file GitHub draft PRs (15 actions/day)
- Scheduled runs, drift alerts & REST API (licence key)
- AI content saturation index — BYO GPTZero key (licence key)
- For agencies under client NDA & enterprise compliance
- Annual updates + security patches
Have a subscription promotion code? Enter it during secure Stripe Checkout. If the code offers an ongoing discount, it continues on eligible renewals while that subscription stays active. Codes do not discount one-off credit packs or the Self-host licence.
What each plan includes
Cloud tiers share the same monthly included checks + prepaid credit model (caps differ by plan; top-ups and balance live under signed-in Billing). The workflow assistant and report export pack (per-prompt prompts for the same AI product you tested, plus browser PDF + CSV with full raw answers) are reserved for Agency and Self-host so lower tiers stay simple while teams that hand work to clients get the hand-off tooling.
| Capability | Starter | Pro | Agency | Self-host |
|---|---|---|---|---|
| Company profiles 3 / 15 / 50 / unlimited (self-host) | Tiered | Tiered | Tiered | Yes |
| AI auto-fill profile from your website (GPT-4o-mini) Included on every plan, including the 14-day trial. You review and edit before saving. | Yes | Yes | Yes | Yes |
| Prompt generator — 6 objectives × 8 intent types (buyer, investor, talent, media, partner, public) | Yes | Yes | Yes | Yes |
| Visibility runs (Claude / GPT) | Yes | Yes | Yes | Yes |
| Brand Brain — approved facts, voice and claims to avoid Editable for signed-in company profiles. Your facts guide future remediation drafts; monitoring remains independent. | Yes | Yes | Yes | Yes |
| Claude-generated fix drafts (self-host pilot) On-demand Claude drafts, five per owner per UTC day. Approval is required before using a connected destination. | — | — | — | Tiered |
| WordPress drafts and GitHub draft pull requests (self-host) Self-host pilot: new WordPress post drafts (optional separately confirmed publication) and single-file GitHub draft PRs. No automatic merge. | — | — | — | Tiered |
| Scheduled automated runs Weekly on Starter; daily on Pro / Agency / Self-host. Trial is manual-only. | Tiered | Yes | Yes | Yes |
| Slack + email drift alerts Cloud Pro and above; licence key on self-host. Per-profile threshold (default 5 points). Slack webhook URL and / or email destination, configured on the profile page. | — | Yes | Yes | Yes |
| REST API access (read-only v1) Bearer-token auth; list / get profiles, list / get runs. | — | Yes | Yes | Yes |
| Monthly included checks + prepaid credit top-ups Tiered monthly caps (UTC); beyond that, 1 credit per check — packs in Billing — unlimited on self-host | Yes | Yes | Yes | Yes |
| Stored raw answers + re-analyse without re-query | Yes | Yes | Yes | Yes |
| Open, auditable score math (source-available ELv2) | Yes | Yes | Yes | Yes |
| Answer + analysis model quality (plan-default) GPT-4o-mini / Haiku 4.5 / Sonnet 4.6 / your env | Tiered | Tiered | Tiered | Tiered |
| Workflow assistant — paste-ready prompts per result for the same AI product | — | — | Yes | Yes |
| Report export pack — print/PDF + CSV (full raw text) | — | — | Yes | Yes |
| On-prem / air-gapped deployment | — | — | — | Yes |
| AI content saturation index (GPTZero-backed) Aggregate, directional signal over the URLs AI cites about your category. Self-host only (licence key) because per-URL GPTZero spend isn't cost-viable for cloud — bring your own GPTZero key. | — | — | — | Yes |
How Arbor compares to typical AI-visibility SaaS
The category is crowded with black-box dashboards. Below is an honest, high-level contrast using patterns we see in publicly listed plans — always verify a vendor's current site before you buy. Representative published price bands for comparable tools often start around $49–$99/mo for lighter SMB usage and climb into the hundreds per month for more prompts, brands, or refresh frequency (examples: GrackerAI, PromptEden, ChatObserver, CiteMetrix).
| Dimension | Arbor | Typical AI-visibility SaaS |
|---|---|---|
| Pricing entry signal (published SMB tier) | £39/mo Starter (this page) | Roughly $49–$99/mo is common for AI-visibility tools with modest prompt volume; multi-brand tiers often $200+/mo. |
| How scores are defined | Documented in-repo (visibility scoring); you can read the exact arithmetic. | Usually proprietary indices; customers rarely see reproducible definitions. |
| Evidence trail | Per-prompt raw model answer stored; optional CSV export on top tiers. | Often dashboard summaries; full raw traces vary by vendor and tier. |
| Engines monitored (today) | Anthropic Claude and OpenAI GPT today; more engines on roadmap. | Many vendors market 5–10+ AI surfaces (ChatGPT, Perplexity, Gemini, etc.). |
| Deployment | Cloud SaaS or annual self-host licence (data stays on your infra). | Almost always vendor cloud; self-host / VPC is uncommon in this niche. |
| Profile setup speed | AI auto-fill from your website (GPT-4o-mini) drafts the 7-field profile in seconds; you review and edit before saving. Available on every plan including the trial. | Setup wizards typically ask you to type sector, audience, products, competitors and key terms by hand for each brand — onerous on multi-brand Agency tiers. |
| Workflow hand-off | Top tiers: generated prompts to execute fixes in the same LLM product you tested. | Varies; often recommendations without copy-paste prompt artefacts. |
| Usage beyond the plan bundle | Tiered monthly included checks (UTC), then prepaid credit packs in Billing; you explicitly accept each run that would spend credits. | Often opaque prompt or query limits, seat-based upsells, or sales-negotiated overages with little upfront clarity. |
Roadmap
Items marked Roadmap above are not yet shipped. Here's the order they're landing in. Trial users get every feature for free as it ships during their 14-day window.
Next
Multi-workspace + roles
Agency-tier workspace switcher with role permissions so multiple operators share one account without sharing one inbox.
Soon
PDF reports + share links
Branded PDF reports for every run, plus signed share links so you can hand a snapshot to a non-customer.
Later
REST API write endpoints
v1 of the API ships read-only (list profiles, list runs). Write endpoints — create profiles, kick off runs, manage prompt sets — land next.
Later
More connected marketing channels
WordPress and GitHub are available for explicit, approval-gated actions in the self-host pilot. Google Search Console and direct social/email integrations remain on the roadmap.
Later
More AI systems
Perplexity, Google Gemini, Microsoft Copilot adapters added to the run queue.
FAQ
Is this just AI SEO with a different name?
No. Traditional SEO ranks pages on Google. Arbor measures how AI assistants describe your company, recommend your competitors, and cite your website when buyers ask category questions in natural language. The optimisation surface is the answer, not the SERP.
Which AI systems do you check?
Today: Anthropic Claude and OpenAI GPT. Coming next: Perplexity, Google Gemini, and Microsoft Copilot.
What's in the 14-day trial?
A lean version of the product so you can validate the workflow end-to-end without a card. You get: 1 company profile, manual runs with up to 8 prompts each, and 24 visibility checks total over the 14 days (no monthly reset — it's a fixed lifetime allowance). Runs use OpenAI GPT-4o-mini. Stored raw answers, per-prompt evidence, competitor breakdowns and the auditable scoring are all included. The free-report tool on this page is unlimited and separate from the trial allowance. At day 14 you pick a plan (Starter / Pro / Agency) or stop — nothing auto-charges, and Cloud Pro upgrades unlock GPT-4o + Claude Haiku 4.5, 15 profiles, 20 prompts/run and the Pro monthly check cap.
How do cloud checks and prepaid credits work?
Cloud plans include a monthly allowance of visibility checks (one check = one buyer prompt in a run: answer plus analysis). If a run would go past that allowance, Arbor shows what will happen and asks you to accept or cancel before any prepaid credits are used. You buy credit packs in Billing (Stripe Checkout); self-host licences have unlimited internal checks on your own keys.
Is it open source?
Arbor is source-available under the Elastic License 2.0. You can read, fork, and self-host the core product free of charge — profiles, prompt generation, runs, analysis, and the published scoring math all work without payment. What ELv2 doesn't permit is reselling Arbor itself as a hosted service or circumventing the licence-key functionality; premium features (export pack, workflow assistant, scheduled runs, alerts, REST API, saturation index) unlock with a paid licence key. The auditable scoring stays readable because that's the trust signal.
How is the score calculated?
AI Visibility Score = % of prompts in a run that mention your company. Website Citation Rate = % of answers linking your domain. Competitor Presence Rate = % of answers naming a rival. Accuracy Risk = % of answers flagged medium or high risk by the analysis pass. The full code is in scoring.ts.
Why is self-host annual rather than monthly?
The core of Arbor is free to self-host under ELv2; the annual licence covers keyed premium features, updates, and support. The application and its database run on your infrastructure. Visibility checks and Claude drafts still send the necessary prompts or brand context to the AI providers you configure, and connected actions communicate with your approved WordPress or GitHub destination. The annual licence supports deployments with greater infrastructure control and a defined maintenance cadence.
What is Brand Brain, and does it change the visibility score?
Brand Brain holds your approved company facts, writing preferences, claims to avoid and source notes. It helps Claude prepare better proposed fixes in the self-host pilot. Brand Brain does not get inserted into the independent AI visibility prompts or change their scores.
Does Arbor automatically publish AI-written fixes?
No. In the self-host pilot, a customer approves a particular Claude-generated asset and separately chooses a destination. Arbor can create a new, unpublished WordPress post, or put the complete approved content of one file into a GitHub draft pull request. Publishing the WordPress post needs another explicit confirmation; merging a GitHub PR is always manual. It does not yet modify existing WordPress pages, automatically patch arbitrary code, send email or post to social networks. Rollback can trash Arbor-created WordPress posts or close unchanged, unmerged Arbor PRs.
How do the WordPress and GitHub connections work?
Connect WordPress with a dedicated, revocable Application Password for your company domain, or GitHub with a fine-grained token restricted to one repository. Arbor encrypts stored credentials. Only approved assets can be sent, each request is explicitly confirmed, and actions are recorded. This pilot is available to self-host accounts, not the free trial or cloud plans.
What happens to my data if I cancel?
Cloud: we export your run history and competitor data as JSON within 30 days, then delete it. Self-host: it never leaves your servers — cancellation just stops new releases and support, and licence-keyed premium features lock when the key expires; the free core (profiles, runs, scores, history) keeps running with no data loss.
Start your 14-day trial
Self-serve, no credit card. Magic-link sign-in, then you create 1 company profile, run up to 8 prompts per check, and have 24 visibility checks to spend across the 14 days on OpenAI GPT-4o-mini. Pick a paid plan any time to unlock more profiles, more prompts per run, and Claude / GPT-4o-grade analysis.