LLM Insights by Naina Labs

Know what AI says about your brand.

More customers now ask ChatGPT, Gemini and Claude before they buy. LLM Insights asks them your customers' questions every week, then shows you whether you're named, where you rank against competitors, how you're described and which sites the models trust.

ChatGPTGeminiClaude
The LLM Insights Overview: visibility, share of voice, average position, sentiment and how often the brand is named first, with a weekly trend for each AI model and a share-of-voice chart.

Sample data with fictional brands.

3 modelsChatGPT, Gemini and Claude
50 promptsthe questions your customers ask
Every weekrun for you on Sunday morning
6 monthsof history to see the trend
The dashboard

Everything AI says about you, in one place.

Set it up once. Every Sunday we ask each model your questions with web search turned on, read every answer and update your dashboard.

01

The numbers that matter, week by week

Five headline measures, each with its trend and the change since last week.

  • Visibility: how often AI names you at all
  • Share of voice: your mentions against every brand named
  • Position: where you appear when you are named
  • Sentiment: how favourably you're described, out of 100
  • Named first: how often you're the top recommendation
Headline metric tiles with sparklines, a visibility trend line for each AI model and a share-of-voice chart.
02

Read every answer, side by side

Open any question to see exactly what ChatGPT, Gemini and Claude said, with your brand highlighted, the position you were given, the sentiment and the sites each model cited. If an answer changed since last week, we show you what it said before.

Three answer cards, one each from ChatGPT, Gemini and Claude, with the brand highlighted, its position, sentiment and cited sources.
03

See where competitors win

A league table of you against up to six competitors, a topic-by-topic view of who gets named, and a list of the questions where competitors are recommended and you aren't. Those gaps are usually the quickest wins.

A competitor league table, a heatmap of which brands are named for each topic, and a list of gaps where competitors are named but the brand is not.
04

Sentiment, themes and wrong facts

Every answer gets a second read that scores how you're described and pulls out the themes, like "reliable" or "price rises". Add your brand facts, such as prices, coverage or policies, and we flag answers that get them wrong.

A sentiment score out of 100, sentiment by model, a themes chart and a table of inaccurate claims compared with the brand's own facts.
05

Know which sites shape the answers

AI models lean on review sites, forums and publishers. See which ones they cite for your questions, whether those pages mention you, and which are worth contacting.

A table of the websites the AI models cited most, showing how often ChatGPT, Gemini and Claude each cited them, their share of citations and whether answers citing them named the brand.
06

Set up in minutes

Search for your brand and competitors from a directory of about 700 well-known companies, and their logos and websites come in with them. Add the other names AI might use, one at a time. Not listed? Type any name in.

Brand settings: search for your brand to bring in its logo and website, and add the other names AI might use.
How it works

Three steps, then it runs itself.

  1. Step 1

    Add your brand

    Choose your brand and up to six competitors. Add the other names people use for you, and any facts you want checked.

  2. Step 2

    Add your prompts

    Write the questions your customers ask, like "best bank for a first-time buyer". Tag them by topic to compare areas of your business.

  3. Every Sunday

    We do the rest

    Each model answers every prompt with web search on. We score each answer and your dashboard updates, ready for Monday.

Why weekly, not daily? AI answers shift a little from one ask to the next, so day-to-day numbers are mostly noise. A steady weekly run, kept for six months, shows real trends, and you can start one extra run per brand each week if you've just made a change.

The models

We always tell you exactly which model answered.

Providers update their models often, and a new model can change the answers. The dashboard names the exact model on every result and flags it when a provider switches.

AssistantProviderModel usedHow it searches
ChatGPTOpenAIgpt-6.2-solOpenAI web search
GeminiGooglegemini-3.8-flashGrounding with Google Search
ClaudeAnthropicclaude-sonnet-5-5Claude web search
Who it's for

For anyone whose brand gets recommended.

Brand and marketing

Track whether AI recommends you, and prove it's moving after a campaign.

PR and communications

Spot wrong facts and negative themes early, and see which publications shape the story.

SEO and content

Find the questions you're missing from and the pages the models trust, then fill the gaps.

Insight teams

Add AI answers to your competitor tracking with a consistent weekly benchmark.

Pricing

One plan. Every model.

Everything in the dashboard, for two brands. Cancel any time.

Starter
£124.99a month

or £1,249.90 a year (2 months free)

  • 2 brands, sharing 50 prompts
  • ChatGPT, Gemini and Claude, with web search on
  • Runs every week, plus one extra run per brand each week
  • Up to 6 competitors per brand
  • Visibility, share of voice, position, sentiment and themes
  • Inaccurate claims checked against your brand facts
  • Every answer and source, kept for 6 months
  • 50 Naina credits every 3 months: use the Naina workspace for chat and file analysis with ChatGPT, Gemini and Claude

Confirm your email, then pay on Stripe's secure page.

More brands£29.99 a month (or £299.90 a year) each, with 25 more prompts. Add them from your dashboard whenever you need them.
Coming soonEmail alerts when an answer changes, a monthly PDF report and recommended actions.
Questions

Good to know

Are these the same answers my customers see?

They're close, but not identical. We ask through each model's official API with web search on. The consumer apps can personalise answers and change them by location, so treat the results as a consistent benchmark rather than a copy of one person's screen.

What's a prompt?

A question you want tracked, written the way a customer would ask it. Each prompt is asked to all three models every week. Your plan includes 50, shared between your two brands.

How is visibility worked out?

Visibility is the share of answers that name your brand. We also record every brand named and in what order, so we can show share of voice and position. A second AI read of each answer scores sentiment, picks out themes and checks claims against your brand facts.

Why weekly?

AI answers vary a little each time you ask. Weekly runs kept over months show the real trend. If you've just made a change, such as publishing a new page, you can start one extra run per brand each week.

Do you store our prompts and answers?

Yes. To show trends we keep your prompts and the answers about your brand for 6 months, then delete them automatically. You can delete a brand and all its history at any time. Answers are about brands, not people, so please don't put personal details in prompts or facts. Our privacy policy has the details.

Which models are included?

ChatGPT from OpenAI, Gemini from Google and Claude from Anthropic. The table above shows the exact model each one uses. When a provider releases a new model and we switch to it, the dashboard tells you.

How do I cancel?

Choose Manage billing in the pricing section of this page while you're signed in. LLM Insights stays on until the end of the period you've paid for. If you cancel within 14 days of your first payment, email hello@nainalabs.co.uk for a full refund.

Find out what AI says about you.

Set up your brand today, and your first results arrive within minutes of your first run.

Start tracking

Questions? Email hello@nainalabs.co.uk