By clicking “Accept All Cookies”, you agree to the storing of cookies on your device to enhance site navigation, analyze site usage, and assist in our marketing efforts. View our Privacy Policy for more information.
From flexible cloud models to sovereign AI hosted in France - choose the right source for your data, your constraints and your budget.
Until now, TimeTonic offered a single way to integrate artificial intelligence into your automations: via major public cloud models. It was already powerful. But for many organisations - regulated industries, demanding IT departments, teams concerned about data sovereignty - one option is simply not enough.
With this new release, TimeTonic moves from a single offering to three distinct AI modes, designed to cover every profile and every requirement: from the startup that wants access to the best models on the market, to the IT department of a hospital that cannot transfer its data outside France.
This complete guide explains how these three sources work, how to configure them, how to manage your credits, and how to get started quickly with ready-to-use templates.
In TimeTonic, artificial intelligence integrates directly into your automation scenarios, via a dedicated action: "Ask an AI". This action can be triggered automatically on any field change, document upload, or any condition you define.
The basic flow is simple and applies regardless of the AI source chosen:
The configuration is identical regardless of the AI source: AI role, instruction (prompt), optional attachments, maximum token count, and response format (free text or structured JSON mapped to your fields).
Public cloud AI is the most accessible entry point. It gives direct access to the most powerful models available on the market from within TimeTonic, with no provider subscription required.
| Model | Level | Best for |
|---|---|---|
| Gemini 2.0 Flash Lite | Economy | High volumes, simple repetitive tasks |
| GPT-OSS 20B | Economy | Classification, short reformulation |
| DeepSeek V3.1 | Economy | Data extraction, strong French language support |
| Mistral Small | Standard | Native French, public sector, long documents |
| GPT-4o Mini 📎 OCR | Standard | Versatile, reliable, image analysis |
| GPT-4.1 📎 OCR | Premium | Complex documents, long deeds, legal analysis |
| Claude 3.5 Sonnet 📎 OCR | Premium | Nuanced analysis, long-form writing, complex instructions |
| Mistral Large 📎 OCR | Premium | Advanced French, contracts, reports |
Public cloud AI offers two billing modes, switchable at any time from your action configuration:
Deducted from your monthly quota. Centralised tracking in Administration. Recommended to get started.
0 TimeTonic credits debited. Billed directly by your provider. Access to the latest models.
Public cloud AI is recommended for teams that need flexibility, want access to the latest model versions, or whose data does not require localised hosting. It is the go-to solution for SMEs, marketing teams, customer service departments, and all business units without strong regulatory constraints.
Available from the FREE plan (10,000 credits/month included).
This is the major new feature of this release for organisations with strict data protection requirements. Private and secure AI allows you to process your data using artificial intelligence models hosted in France, on TimeTonic's sovereign infrastructure.
The models used (Mistral, LLaMA, Qwen and other open source models) are deployed directly on TimeTonic's infrastructure in France. Your data never leaves this controlled environment. No transfer to an external provider, no exposure to third-party cloud API terms of use.
Unlike public cloud AI which routes your data to OpenAI, Mistral or Google servers, private AI stays entirely within the TimeTonic environment. This architecture is designed to meet the strictest GDPR requirements and sector-specific obligations for non-transfer of data outside the European Union.
| Model | Level | Best for |
|---|---|---|
| Mistral 7B Instruct | Economy | Repetitive text, high volume, sensitive data |
| Llama 3.1 8B Instruct | Economy | Simple extraction, sovereign data triage |
| GPT-OSS 20B | Economy | Light reasoning, short analysis on internal data |
| Mistral Small 3.2 24B 📎 OCR | Economy | Native French, image analysis, sensitive data |
| Llama 3.3 70B Instruct | Premium | Complex documents, deeds, contracts |
| Qwen 2.5 VL 72B 📎 OCR | Premium | Advanced OCR: ID cards, invoices, bank details |
Patient records, medical reports, sensitive forms
Contract analysis, bank details/company extract extraction, risk qualification
Notarial deeds, contractual clause analysis
Organisations subject to inter-ministerial security framework (ANSSI)
Confidential technical data, maintenance sheets
Available from the PRO plan (500,000 private AI credits/month included).
The third mode is aimed at organisations that already have an AI infrastructure or wish to deploy their own models. It allows TimeTonic to connect to an AI server you fully control.
Dedicated infrastructure managed by TimeTonic. Your models deployed in an isolated environment, operated on your behalf.
On-premise or private cloud. You connect TimeTonic to your own AI server via URL and authentication.
In both cases, no TimeTonic credits are consumed. Billing is entirely separate from your other AI sources. TimeTonic automatically detects the models available on your server during configuration.
This mode is designed for CIOs and systems architects with specific requirements: proprietary or fine-tuned models, air-gap architectures, existing infrastructure to leverage, or performance and throughput constraints that shared solutions cannot meet. This is the maximum level of control.
Available exclusively on the BUSINESS plan. Contact TimeTonic for setup. Server configuration is carried out by a technical administrator.
One of the most powerful AI features in TimeTonic is optical character recognition (OCR), available across all three sources. It enables automatic extraction of structured data from scanned documents.
The flow varies depending on the file type:
The AI analyses the attached file directly and generates structured JSON. Models: Mistral Small 3.2 24B, Qwen 2.5 VL 72B.
OCR action (Mistral AI OCR) converts the PDF to JSON text, then a second AI action structures and extracts the information.
| Document | What the AI extracts automatically |
|---|---|
| 🧾 Expense report | Amount, date, category from a receipt photo |
| 🪪 ID card / Passport | Identity data injected into a CRM form |
| 🏢 Company extract / Legal documents | Company information in your database |
| 🏦 Bank details (RIB) | IBAN and BIC extracted without manual re-entry |
| 👤 Business card | Contact automatically added to your CRM |
| 🔧 Equipment nameplate | Brand, model and references for your maintenance sheets |
| 🚗 Vehicle registration document | Vehicle data extracted from photo or PDF |
For a robust scenario, use a condition that automatically routes processing based on file type: if the filename contains ".pdf" - PDF branch, otherwise - image branch.
The BYOK (Bring Your Own Key) feature lets you use your own OpenAI or Mistral API key in your automations, replacing TimeTonic credits. It is available on all plans.
In your "Ask an AI" action configuration (public cloud), enable the "Use my own API key" button, select your provider and enter your key. You then have access to your provider's full catalogue, including the most recent models (GPT-5, o3, o4, Pixtral 12B, etc.).
| TimeTonic Credits | Personal API Key (BYOK) | |
|---|---|---|
| Billing | Via TimeTonic | Directly with the AI provider |
| Usage tracking | TimeTonic Administration space | To be checked with the provider |
| Limit | According to your TimeTonic plan | According to your provider account |
| Auto suspension | At 100% of monthly quota | No TimeTonic suspension |
| Plans | All plans | All plans |
Yes, securely. Your API key is stored in the TimeTonic database in encrypted form. It can only be decrypted by the application itself and is never accessible in plain text, not even by the TimeTonic team. It is used solely to execute your AI actions within your automations, and is never shared or transmitted to third parties.
If you believe your key has been compromised, revoke it directly from your provider's dashboard (OpenAI, Mistral AI) and enter a new key in TimeTonic.
Basic rule: 1 TimeTonic credit = 1,000 tokens. Private AI and public cloud AI each have their own independent monthly quota.
| Plan | Public cloud AI credits/month | Private AI credits/month | Dedicated server |
|---|---|---|---|
| FREE | 10 000 | - | No |
| START | 100 000 | - | No |
| PRO | 1 000 000 | 500 000 | No |
| BUSINESS | 10 000 000 | 5 000 000 | ✅ Yes |
Unused credits do not roll over to the following month. With your own API key (BYOK), no TimeTonic credits are debited. Additional credit packs are available from the Business plan.
Go to your Administration - Plan & Billing notebook to view the status of your two counters (private AI and public cloud AI). The gauges update in real time:
Automation logs allow you to analyse usage action by action: model used, input tokens, output tokens and total credits consumed. This is the ideal tool for comparing two models on the same task and choosing the most cost-effective one.
When the quota reaches 100%, AI automations are automatically suspended. Other automations (without AI calls) continue to function normally. Three options are available:
Your quota resets to zero at the start of each month, automatically.
Additional credit pack available from the Business plan.
Your automations resume immediately with your personal API key.
The model is the main lever on your credit consumption. Here is a quick selection guide:
| Need | Public Cloud AI | Private & Secure AI |
|---|---|---|
| High volume | Gemini 2.0 Flash Lite, GPT-OSS 20B | Mistral 7B, Llama 3.1 8B |
| Native French | Mistral Small | Mistral Small 3.2 24B |
| Image analysis / OCR | GPT-4o Mini, GPT-4.1 | Mistral Small 3.2 24B, Qwen 2.5 VL 72B |
| Complex docs | GPT-4.1, Claude 3.5 Sonnet, Mistral Large | Llama 3.3 70B Instruct |
| Advanced OCR | GPT-4.1, Claude 3.5 Sonnet | Qwen 2.5 VL 72B |
| Sensitive data | Switch to Private & Secure AI (France hosting) | |
Before deploying an automation on a large volume of data, ask yourself three questions:
Always test on a sample of 5 to 10 representative cases before deploying at volume. Compare the logs to validate your estimate and adjust your model choice if needed.
No need to start from scratch. TimeTonic offers a library of 10 ready-to-use AI templates, with pre-configured tables, structured fields and AI automations already set up.
| Template | Type | What the AI does automatically |
|---|---|---|
| 🪪 ID card / Passport | OCR - Document analysis | Extraction of identity data into your fields |
| 👤 Business card | OCR - Document analysis | Automatic contact addition to your CRM |
| 🏦 Bank Details (RIB) | OCR - Structured extraction | Extraction of IBAN and BIC |
| 🔧 Maintenance Analytical Sheet | OCR - Document analysis | Extraction from an equipment nameplate |
| 📞 International Phone Format | Summary & classification | Automatic conversion to correct international format (+33, +49...) |
| 🌍 Multilingual Generator | Summary & classification | Content in multiple languages from a single source |
| 🏢 Company Identity (KBIS) | OCR - Structured extraction | Extraction of legal information |
| 🚗 Vehicle Registration | OCR - Document analysis | Extraction of vehicle data |
| ✉️ Personalised Message | Summary & classification | Generation of a professional introduction message |
| 🧾 Expense report | OCR - Extraction structurée | Montant, date et catégorie depuis un reçu photo |
After installation, open the automation and check four key elements:
A template is a starting point, not a final configuration. Always test with realistic documents before deploying to production.
With this triple offering, TimeTonic covers the full spectrum of enterprise AI needs: from the maximum flexibility of the public cloud to the total sovereignty of the dedicated server, via the security-accessibility balance of private AI hosted in France.
Whatever your context - agile startup, growing SME, mid-sized company subject to sector obligations, or large group with a structured IT department - you will find in TimeTonic the AI configuration that matches your technical, regulatory and budgetary requirements.
Ready to get started? Explore the AI templates available in your TimeTonic space, test on a real use case and check the logs to fine-tune your configuration.
