Start the setup by clicking on Add New Entity in your dashboard:

Next, enter your URL and click Continue:

Beket AI will now analyze your URL and extract relevant information. On the next screen you will see a setup page with 4 sections:

1. Business - Who and Where

2. Competitors

3. Questions - What we ask AI

4. Launch - Engines and Run

1. Confirm the business

The setup step where you check what Beket AI pulled from your site. Every field rewrites the questions on the right — if those questions look wrong, update the relevant inputs.

Core details

1. Brand name — confirm the name AI should be looking for.

2. Primary domain — confirm the site being audited.

3. Business vertical — pick from the dropdown. This loads the question taxonomy.

4. Category (buyer's words) — write what a buyer types, not your positioning.

5. Primary segment — SMB, Mid-market, or Enterprise.

6. Market scope — Local, Multi-city, National, or Online-first.

Market and identity

7. Primary market — the main location this audit targets. Used to localize queries. Add more markets in the Add a market field.

8. Identity anchors (optional) — street address, phone, Google Business Profile URL, other profiles. Anchors resolve your brand against same-name entities so only your mentions count.

9. Languages in this market (optional) — AI cites different sources per language. Each language you add gets at least a quarter of the question set.

10. Also known as — add alternate names or acronyms, then press Enter.

Advanced entity signals

Expand Advanced entity signals — claims, pains, description.

11. Claims & differentiators — scraped from your site. Tap a claim to mark it verified. Only verified claims can appear in questions or reports, and never in discovery questions, which must stay in the buyer's voice. Add your own claims with Enter.

12. Customer pain points — each pain point becomes one authored buyer question. Add with Enter.

13. Neutral description — 50–100 words, factual, third person, no marketing language. Used for entity grounding.

14. Primary product — what you sell.

15. Offer — how you'd describe the offering.

16. Ideal customers — who you're trying to reach. Add with Enter.

2. Competitors

The next screen: your competitor set. Wrong competitors corrupt share-of-voice, so judge each one by name and domain, and remove anything that doesn't belong.

Each row has four fields:

Steps:

1. Review each row. Check the name and the domain — same-name companies are the usual mistake.

2. Fix the size and type if the dropdowns are wrong.

3. Remove anything that doesn't belong with the ✕ button.

4. Add competitors with + Add competitor.

3. Questions - What we'll ask AI

The step where you choose the exact questions Beket AI sends to AI assistants.

Unbranded questions are where new clients are won. They test whether AI finds you when the buyer doesn't know your name. This exact set is re-run after fixes to measure lift.

Question areas

Beket AI loads a set of question areas based on your business vertical. Toggle each one on or off with its checkbox.

Question set

Below the areas, you'll see the actual questions Beket AI generated, grouped by area. The header shows how many you've selected and the estimated run time.

Each question carries two labels:

• Branded — your name is in the question. Tests how accurately AI already describes you.

• Unbranded — your name is absent. Tests whether AI finds you at all.

The bar at the top shows your branded/unbranded split. A language tag (e.g. EN) marks which language each question runs in.

Steps

1. Turn areas on or off. Deselecting an area drops its questions from the set.

2. Review the generated questions in each group. Beket AI pre-selects some; the rest are unchecked.

3. Check or uncheck individual questions to shape the set.

4. Watch the branded/unbranded split. Lean unbranded if your goal is discovery by buyers who don't know you yet.

5. Click + Write your own question to add your custom question at the bottom.

The one catch: there's a selection limit

You can only run a fixed number of questions per audit. Once you hit it, the Write your own question box shows "Limit reached — deselect a question to add another."

That's not a bug — it's a budget. To add a question you care about, uncheck one you care about less.

Because the same set is re-run after your fixes, it's worth spending a minute here. A sloppy set makes your before-and-after comparison meaningless.

What to do

• Chasing new buyers? Keep the unbranded share high — those are the questions where you're competing to be found.

• Worried about misinformation? Weight toward Brand accuracy and Reviews & outcomes.

• Sizing up rivals? Keep Competitor comparison on and confirm your competitor set is correct first.

• At the limit? Deselect the questions closest to duplicates before cutting a distinct angle.

4. Launch - Engines and Run

The last setup step: pick which AI engines Beket AI queries, then launch.

Each question runs on every engine you select. Two engines with 25 questions means 50 answers analyzed.

You can change any of this later from Audit Settings, so nothing here is permanent.

Choose the AI engines

Engines are called connectors. The available ones are:

• ChatGPT

• Google AI Overview

• Google AI Mode

• Perplexity

• Claude

Your plan sets how many you can select at once. The header shows your allowance (e.g. "Up to 2 connectors on your plan") and how many you've picked. Once you hit the cap, the remaining engines grey out — deselect one to swap.

Different engines pull from different sources, so they won't agree. That's the point: an engine where you're invisible is a gap you can act on.

Before you launch

Beket AI shows a summary of everything you set up. Check it against what you expect:

If anything looks wrong, go back and fix it now. The inputs shape the questions, and the questions shape the results.

Then launch

Hit Start audit. Beket AI sends each selected question to each selected engine, reads every answer, and builds your report.

What to do

• Limited to two connectors? Start with the engines your buyers actually use, then swap in others on a later run to compare.

• Want a like-for-like comparison over time? Keep the same engines when you re-run after fixes — changing engines changes the baseline.

• Summary looks off? Back up and correct the setup rather than launching and re-running.