Client assistant (RAG)
Answers your clients from your own documents. 24/7 support and FAQ, every answer sourced.
ARTIFICIAL INTELLIGENCE
AI that actually holds up — not another wrapper.
Client assistants, qualification agents, business automation. Hosted in Europe, sourced, measured, no lock-in. You own it.
WHAT SETS US APART
Everyone plugs in a chatbot. Few deliver an AI that actually holds up.
Data sovereignty
Sent to US servers, reused for training.
Processed in Europe, zero retention, DPA provided. Your data stays yours.
Zero lock-in
Locked into one vendor, impossible to leave.
Interchangeable models, prompts and evals in your name, open export.
Guardrails & evals
A demo that hallucinates once it's in production.
Tested, measured, bounded: PII/injection guardrails, error rate measured before launch.
Grounded in your business
A generic chatbot disconnected from your business.
RAG on your documents, wired into your tools (invoicing, calendar, CRM).
Sourced answers
Confidently makes up an answer.
Cites its sources and says "I don't know" instead of making things up.
Price & outcome
Billed by the hour, fuzzy scope.
Fixed quote, audit first, measured outcome (hours saved).
Evals, not magic. We measure the error rate before shipping.
WHAT WE BUILD
Assistants and agents grounded in your business.
Qualification agent
Qualifies your leads, replies at night, suggests a time slot. You get requests ready to work.
Internal copilot
Searches your documents, summarizes, drafts. Your teams save hours.
Business automation
Email sorting, PDF and invoice extraction, data entry avoided. The AI does the chore, not you.
E-invoicing 2026
Factur-X / Peppol extraction and validation, ready for the mandate arriving in 2026–2027.
AI search (GEO)
Be found and cited by generative search engines, not just ranked by Google.
THE METHOD
Audit first. Fixed-price build. Care after.
- 01
AI audit
We map what's automatable, cost out the ROI, and write a roadmap. Nothing gets built blind.
- 02
Build
Fixed price, written scope. RAG on your data, guardrails, evals. Delivered in your name.
- 03
AI care
Monthly: ongoing evals, model updates, token-cost monitoring. The AI doesn't drift.
AI PRICING
An audit first, then a fixed price. Never by the hour.
AI audit
the entry point
from €900
- Mapping of automatable tasks
- Costed ROI, roadmap
- Deducted from the project if you continue
Single-task agent
1 to 2 weeks
from €2,500
- One task automated end-to-end
- Guardrails + eval suite
- Fixed price, code delivered
RAG assistant
3 to 6 weeks
from €6,500
- Assistant on your own data, sourced
- Guardrails, evals, anti-hallucination
- Multi-model, no lock-in
AI care
no commitment
€90+/month
- Ongoing evals, model updates
- Token-cost monitoring
- Guardrails maintained
Indicative amounts, excluding VAT. Token costs at actuals or capped — shown, never hidden.
COMING SOON — LIVE ON THIS SITE
The Solive assistant will answer right here.
An assistant that answers from this content, cites its sources, says "I don't know" when it doesn't, and puts you in touch. The demo is the product.
FREQUENTLY ASKED QUESTIONS
What people ask about AI.
Does my data train the AI?
No. Processed in Europe, zero retention on the provider's side, DPA provided. Your data stays yours.
What if the AI makes something up?
The assistant cites its sources and answers "I don't know" instead of making things up. We measure the error rate with evals before launch.
Am I locked into a vendor?
No. Interchangeable models via a gateway, prompts and evals in your name, open export. You can leave.
How much do tokens cost?
Varies by usage. We show them at actuals or cap them — never a hidden cost.
Got an AI use case in mind?
We start with a short audit: what's automatable in your business, the ROI, the roadmap. Then a fixed quote.