Frequently asked questions

SkaLean is an AI-as-a-Service (AIaaS) platform that combines three products: AI Studio (collaborative knowledge base), AI Automation (no-code workflows) and AI Assistant (agents on 10 channels). Everything is deployed turnkey by our team on sovereign infrastructure.

ChatGPT and Copilot are generic tools that send your data to US servers. SkaLean is a sovereign business platform: your data stays in your country, the AI is trained on YOUR documents, and everything is deployed and managed by our team. Documented ROI from month 2. Deployment in 5 to 20 days. Average savings: $500/month compared to fragmented tools.

AI Studio is your AI knowledge base: index your documents and get precise answers with citations. AI Automation orchestrates your business processes with visual no-code workflows. AI Assistant deploys a conversational agent on 10 channels (WhatsApp, phone, web, Slack, etc.) for your customer service.

You can start with a single module. Most of our clients start with AI Studio or AI Assistant depending on their most urgent need, then add the other modules. Each module works independently.

The number of documents depends on your plan. The Essential plan includes up to 500 documents, the Professional plan up to 5,000, and the Enterprise plan is unlimited. Supported formats include PDF, Word, Excel, PowerPoint, Markdown, HTML and 20+ other formats.

Any AI can produce inaccurate answers. That is why SkaLean systematically cites sources: every answer indicates the source document, section and page. You can verify in one click. Additionally, businesses can configure moderation rules and validated answers for critical topics.

Yes. Through the configuration panel, you define the tone, style, rules of conduct and prohibited responses. Your agent can respond under your brand name, with your editorial guidelines and your specific procedures.

SkaLean supports 20+ LLM models via 9 providers: OpenAI (GPT-4o, o1), Anthropic (Claude 3.5), Mistral AI, Meta Llama 3, Google Gemini, Cohere, AI21 and our sovereign GPU models. You choose the model suited to each use case.

Yes. SkaLean was designed for French-speaking and Arabic-speaking markets. The selected LLM models handle French, standard Arabic and regional dialects very well. The platform is available in French and Arabic.

Yes. SkaLean integrates with 200+ tools via AI Automation: Salesforce, HubSpot, QuickBooks, Google Workspace, Microsoft 365, Slack, Teams, Shopify, and many others. Webhooks and an SDK are also available for custom integrations.

Your data is hosted on servers physically located in your country. AI inference runs on local sovereign GPUs. No data transits to foreign servers.

Yes. SkaLean was designed with applicable regulations in mind: data residency in your country, AES-256 encryption, access logging, right to portability and erasure, privacy policy, and Privacy Impact Assessments (PIA) available on request.

SkaLean has been audited against all 17 HIPAA criteria and meets all of them. Access control, audit trail, encryption of data at rest and in transit, disaster recovery plan, BAA agreement available. Medical and dental clinics can use SkaLean to process patient data.

SkaLean uses multi-tenant isolation at the database level (PostgreSQL Row-Level Security), strong authentication (TOTP/WebAuthn), bank-grade encryption (AES-256) for data at rest, TLS 1.3 in transit, and 8-layer anti-abuse protection including rate limiting and anomaly detection.

No. SkaLean's architecture is designed so that no SkaLean employee has access to your documents or conversation data. Technical access is limited to metadata required for support. Exceptional access requires your explicit authorization and is logged.

You control data retention. By default, data is retained for the duration of your active subscription. Upon cancellation, your data is exportable and erased within 30 days upon request. Your data is never used to train AI models.

Yes. SkaLean implements 10/10 OWASP LLM protections against attacks: input validation, context isolation, manipulation detection, execution sandboxing and behavioral monitoring. Protection against system data extraction is particularly reinforced.

SkaLean includes native PII protection that detects 15 types of sensitive data before sending to the LLM: names, addresses, card numbers, insurance numbers, emails, phones, dates of birth, medical record numbers and others. Detection is automatic and configurable per field.

SkaLean is available from $29/month per module (Essential plan). Plans include: Essential ($29–79/month per module), Professional ($149–299/month per module), Enterprise (custom pricing).

Our clients see measurable results from month 2. The most common sources of value: saving 15–20 hours of administrative work per week, significant reduction in no-shows (healthcare), replacing $500/month in fragmented SaaS subscriptions, and notable gains on qualified leads (real estate, professional services).

Turnkey deployment is included in all Professional and Enterprise plans. For the Essential plan, onboarding fees may apply depending on complexity. No external consultant cost, no open-ended integration project.

Yes. We offer a free 30-minute demonstration tailored to your sector and challenges. For Enterprise projects, a paid 30-day pilot can be arranged with a defined scope and success metrics established in advance.

Monthly plans are available with no annual commitment. Annual plans get 2 months free. You can cancel at any time with 30 days notice, and your data is returned to you in standard formats.

Standard and Professional plans are flat-rate — no end-of-month surprises. Enterprise plans can include usage-based pricing for very high volumes. In all cases, limits and billing model are clearly defined before signing.

In practice, yes. Our clients replace an average of 3 to 5 tools: ChatGPT Teams ($30/user), Make or Zapier ($100–300/month), Intercom or Zendesk ($200–500/month), Notion or Confluence ($200/month), Twilio (variable). All for a unified, sovereign, turnkey tool.

Onboarding takes between 5 and 20 days depending on complexity. A simple AI Studio (indexing 50 documents, chat configuration): 5 days. A full deployment (3 modules, CRM and telephony integrations, team training): 15–20 days.

No. Our team handles everything: infrastructure, configuration, integrations, training. Your team learns to use the platform, not configure it. If you have an IT team, we can work with them on advanced integrations.

Onboarding includes: discovery session (needs and configuration), go-live session (setup and integrations), user training session (2h), and an intensive 30-day post-launch support period. Additional training sessions are available.

Yes. We have migration processes from ChatGPT Enterprise, Intercom, Zendesk, Make, Zapier and other platforms. Conversation history and existing documents can be imported. Migration timeline varies depending on volume.

SkaLean's AI Assistant handles AI voice natively. For existing telephony, we support integration with major PABX/VoIP systems, Twilio and modern IP telephone exchanges. A dedicated DID number can be provided.

Yes, for Enterprise plans. The fine-tuning module allows you to adjust a base model (Llama 3, Mistral) on your specific data. Particularly useful for sectors with specialized terminology (medical, legal, financial).

SkaLean targets 11 sectors: healthcare (dental and medical clinics), financial services, accounting and taxation, legal, insurance and brokerage, home services, human resources, consulting and advisory, professional training, non-profit and automotive.

Yes. Our team communicates in English, French and Arabic. Support is available by chat, email and phone during business hours. A self-service knowledge base is also available.

Yes. SkaLean complies with applicable personal data protection regulations. Data residency in your country, access and rectification rights, incident reporting — all obligations are covered.

Yes. The LLM models available in SkaLean handle French, Modern Standard Arabic (MSA) and regional dialects. Quality is best in French and standard Arabic, but dialects are supported for customer service use cases.

Several grant programs may apply depending on your region. We can help you identify applicable programs for your situation.

RAG (Retrieval-Augmented Generation) is the technology that allows AI to answer based on YOUR documents rather than its general knowledge. The SkaLean pipeline has 6 stages: document ingestion, smart chunking, vector embedding, indexing in a vector database, semantic search and synthesis with citations.

Inference is the computation performed by the LLM model to generate a response. SkaLean runs these computations on GPUs physically located in your country, rather than on OpenAI or Google cloud servers. This ensures your requests and data never cross your national border.

Fine-tuning involves re-training an LLM model on your specific data to improve its performance in your domain. Most clients don't need it — the RAG pipeline is sufficient. Fine-tuning is useful for very specialized terminology (medical, highly technical legal) or very high interaction volumes.

MCP is an open standard developed by Anthropic to connect AI agents to external data sources and tools in a secure and standardized way. SkaLean implements it to allow your AI agent to access your CRM, calendar, database and other systems in real time.

On local sovereign GPU infrastructure: 800ms–2s for the first token, 3–8s for a complete response depending on length. On external LLM API (GPT-4o, Claude): 1–4s. These latencies are acceptable for conversational use cases and comparable to generic tools.

AI Studio uses real-time collaboration to allow multiple users to simultaneously edit the same document. Each modification is synchronized in real time (< 100ms latency) with automatic conflict resolution, without data loss.

SkaLean uses 6 specialized databases: PostgreSQL (relational data), Redis (cache and sessions), Qdrant (vector database for RAG), MongoDB (unstructured documents), Elasticsearch (full-text search), and MinIO (file storage). This architecture guarantees performance and resilience.