Write and edit
An email, letter, message or text that reads more naturally.
Make this message clearer and warmer.
Simple, accessible AI
Useful everyday AI, at a price that stays reasonable.
Write, summarise, translate, understand and find ideas in a clear interface. Surikat focuses on what matters, from €4.50 a month.
- Invitation-only private beta
- From €4.50 a month
- Conversations and AI hosted in Europe
See before you choose
One question. One clear answer. Then keep going.
This is a faithful example of the experience: write as if you were talking to someone, then refine your request when you need to.
You could then ask
- Write a polite reply
- Make a checklist of the documents to send
- Explain it even more simply
Surikat · Assistant
Of course. This letter asks you to do two things: confirm your address before 12 September and attach a copy of your supporting document. If you do not reply by then, your application may be delayed.
Every day
Useful from your very first question.
You do not need special vocabulary or AI expertise. Just describe what you want to achieve.
Understand
A concept, letter or instruction explained without jargon.
Explain this contract in plain language.
Summarise
The important points from a long text, in the format you need.
Summarise this text in five key ideas.
Translate
A natural translation that keeps the original meaning and tone.
Translate this message into professional French.
Find ideas
Concrete suggestions to start, choose or unblock a situation.
Suggest five simple meals for this week.
Save time
Lists, plans and first drafts that are ready to adapt.
Prepare a checklist for my move.
Designed to feel simple
Everything you need. Nothing to configure.
A readable interface, answers that fit your request and useful features without turning every question into a technical project.
- 01
Speak naturally
Ask a short question or provide lots of context. The assistant adapts to the way you ask.
- 02
Keep the thread
Return to your conversations and ask for a shorter, clearer or more detailed version.
- 03
Stay in control
View, search or delete your conversations. Memory can be managed from your settings.
Privacy
Your conversations are not used to train AI.
The service and AI models are hosted in Europe. You remain in control of your conversations and memory.
- 01
European hosting
Your conversations and AI processing remain on European infrastructure, hosted in France. Nothing leaves Europe to be computed.
- 02
No training
Your conversations are not reused to train our models or a third party's models.
- 03
Data control
You can manage your history, export your data and delete your account from the app.
What about web search?
When a web search is needed, it goes through a European provider. If that service is unavailable, a US engine may take over: the app then tells you so explicitly in the answer. In both cases only your question is sent, stripped beforehand of any email addresses, phone numbers or identifiers it may contain, and never your conversation. So we do not promise a 100% European search in every circumstance: we promise to tell you when it is not.
Is it good enough?
Less famous models. Answers that do the job.
The right question is not “which model is the biggest” but “is the answer any good?”. We measured it instead of claiming it.
96 to 99%
of answers rated satisfactory
Internal measurement of 27 July 2026, on 100 everyday requests in French and English.
By task
- Writing and editing100%
- Explaining a topic100%
- Drafting an official letter100%
- Summarising a text91%
- Translating82 to 100%
Method: every request is actually sent to the models, then the answer is graded by a judge model different from the one that answered. An answer counts as satisfactory when it scores at least 4 out of 5. The three model levels are measured separately; where they differ, the range is shown.
What we do not claim: no assistant is infallible, and neither is this one. Always double-check an important figure, date or legal detail before relying on it.
The three model levels
We have no in-house model to defend. Surikat runs on open-weight models, executed on our own European infrastructure. We present them as three levels rather than under their technical names: that is the only part which changes anything for you, and you have nothing to choose.
- Surikat Small
- Surikat Medium
- Surikat Large
Why open weights matter here: these models do not run at their publishers' premises, they run on our European infrastructure. None of your conversations is ever sent to a model publisher, wherever it may be based.
Against the assistants you already know
As good? As fast? At what price?
Three fair questions to ask before paying. We answer all three, including when the honest answer is « we cannot measure that yet ».
Quality
Good enough for what you actually do
On everyday requests, writing, correcting, summarising, translating, explaining, our 27 July 2026 measurement gives 96 to 99 % satisfactory answers. That is what we can prove, and it shapes your real experience far more than where a model sits in a ranking.
What we do not publish: a score table against ChatGPT, Gemini, Claude or DeepSeek. We looked for common ground on 28 July 2026, including in the public LMArena ranking updated the day before, and we found none of the three open weight models we serve in it. Placing two different evaluations side by side would not be a comparison.
Speed
Three speeds, picked for you
Each question goes to the model that suits it. A simple request does not queue behind a heavy model, and complex topics keep the most capable one. There is nothing for you to configure: the choice is made again on every message.
What we do not publish: a speed figure. The speeds quoted by comparison sites measure a model on the infrastructure of whoever publishes them, not on ours. Until we have measured our own response times at our European host, quoting one number of seconds rather than another would be a promise we could not keep.
Price
This is where the gap is clearest
A subscription is compared per person. The Family plan gives 4 private accounts, each with its own history, for €3.00 per person, per month. None of the paid subscriptions listed below goes that low, and none includes more than one account.
One thing worth stating: other publishers' prices are quoted in the currency and tax basis they display, in dollars and excluding tax. Ours are in euros, tax included. We do not convert on your behalf: today's exchange rate would age faster than this page.
Cost per person, per month
| Plan | Per person, per month | Private accounts included |
|---|---|---|
| FamilyOur own price list, in euros, tax included. | €3.00 incl. tax | 4 |
| SoloOur own price list, in euros, tax included. | €4.50 incl. tax | 1 |
| ProOur own price list, in euros, tax included. | €9.99 incl. tax | 1 |
| ChatGPT Go (OpenAI)chatgpt.com/fr-FR/pricing, checked on 28 July 2026, with no mention of tax excluded or included. The page notes this plan may include ads. | €8 | 1 |
| Claude Pro (Anthropic)claude.com/pricing, checked on 28 July 2026. The publisher advertises $17 a month on an annual commitment. | $20 excl. tax | 1 |
| Mistral AI, Pro planmistral.ai/pricing, checked on 28 July 2026. | $14.99 excl. tax | 1 |
| DeepSeekdeepseek.com, checked on 28 July 2026: the assistant is advertised as free and no consumer subscription is displayed. | Free | 1 |
Other publishers' prices were checked on 28 July 2026 on their official pages. We only list the ones we could verify ourselves that day, and we leave them in the currency and tax basis the publisher displays. These prices change often: check with the publisher before deciding.
At that price you are not only paying for answers. You are paying for answers computed in Europe, on European servers, without your conversations being used to train anything.
Mini plan finder
Which plan fits me?
Answer four questions. The result updates with every answer, with no email and no account required.
Question 1 of 4
Pricing
Three clear plans, with no commitment.
Choose according to your needs. There is no public free trial. Private beta access opens gradually by invitation.
Mini
The price of a coin, for very occasional use.
€2.00€24.00/mo
€2.00 per person, per month€2.00 per person, per month, billed annually
- 150 interactions per month, no daily limit
- 20 web searches, 5 image analyses, 5 documents per month
- 10 min of voice dictation per month, free read-aloud (100% in the browser)
- Writing, summarizing, translating and explaining
- History and search across conversations
Solo
The essentials of AI for one person.
€4.50€45.00/mo
€4.50 per person, per month€3.75 per person, per month, billed annually
- 2,000 exchanges a month, with no daily cap
- 150 web searches, 40 image analyses, 50 documents a month
- 60 min of voice dictation a month, free read-aloud (100% in the browser)
- Writing, summarising, translating and explaining
- Conversation history and search
Pro
For deeper work on complex topics.
€9.99€99.90/mo
€9.99 per person, per month€8.33 per person, per month, billed annually
- 4,000 exchanges a month (fair use)
- Expert mode included, capped
- 400 web searches, 150 image analyses, 250 documents a month
- 300 min of voice dictation a month, free read-aloud
- Priority support
Family
Up to 4 personal accounts, one bill, the lowest price per person.
€11.99€119.90/mo
€3.00 per person, per month€2.50 per person, per month, billed annually
- Up to 4 personal accounts, fully isolated
- 1,200 exchanges a month and per person
- Expert mode included per person, the only plan besides Pro to offer it
- 60 min of voice dictation a month and per person, free read-aloud
- One bill, the lowest price per person in the lineup
Prices include VAT. Monthly or annual billing. Cancellation takes effect at the end of the billing period, with no refund for the current period.
Detailed comparison
Which plan is right for you?
Each row is a monthly cap. Family: the lowest price per person in the lineup, and the only plan besides Pro to include Expert mode.
| Feature | Mini | Solo | FamilyRecommended plan | Pro |
|---|---|---|---|---|
| Price | €2.00/mothat is €2.00 per person | €4.50/mothat is €4.50 per person | €11.99/mothat is €3.00 per person | €9.99/mothat is €9.99 per person |
| Interactions | 150/mo | 2,000/mo | 1,200/mo/member | 4,000/mo |
| Expert mode | — | — | 15/mo/member | 60/mo |
| Web search | 20/mo | 150/mo | 120/mo/member | 400/mo |
| Image analysis | 5/mo | 40/mo | 30/mo/member | 150/mo |
| Documents | 5/mo | 50/mo | 40/mo/member | 250/mo |
| Scanned document pages | — | 100/mo | 75/mo/member | 150/mo |
| Generated documents | — | 20/mo | 15/mo/member | 100/mo |
| Storage | 50 MiB/mo | 500 MiB/mo | 500 MiB/mo/member | 5 GiB/mo |
| Image generation | — | — | 10/mo/member | 50/mo |
| Voice dictation | 10 min/mo | 60 min/mo | 60 min/mo/member | 300 min/mo |
| Memory | Included | Included | Included | Included |
| Read aloud | Included | Included | Included | Included |
| Multiple accounts | — | — | 4 accounts | — |
Frequently asked questions
Answers before you get started
01What can I do with Surikat?
Write and edit, summarise, translate, explain a topic, find ideas, prepare a plan or simplify an everyday task. Write your request in your own words.
02How does the private beta work?
Request access with your email address. Invitations open gradually and each account is approved individually. The beta is not a time-limited free trial.
03Is there a free trial?
No. There is no public free trial and no free plan. You can see the experience on this page before requesting access.
04Are my conversations private?
Yes. They are hosted in Europe and are not used to train a model. On the Family plan, members can never see each other's conversations.
05How do quotas work?
Solo includes 2,000 exchanges a month, and each Family account has its own 1,200 exchanges a month, never a shared pool. A technical call that fails before delivering an answer is not deducted.
06What is the difference between Solo and Pro?
Solo covers everyday tasks. Pro adds Expert mode for complex topics, higher priority during busy periods and priority support.
07Are Family accounts really separate?
Yes. Each member has their own account, conversations, memory and quota. The paying member only manages the bill and invitations.
08Which AI models do you use?
Open-weight models, executed on our European infrastructure and never at their authors' premises. They are offered as three levels, Surikat Small, Medium and Large, and the most suitable one is selected automatically according to your request and your plan. You have nothing to configure.
09How do payment and cancellation work?
Payments are handled with Stripe Billing. You can cancel at any time. Access remains active until the end of the paid period, and that period is not refunded.
Private beta
Request access to Surikat
We are opening access gradually. Leave your email to join the list for upcoming invitations.