// Pricing
Prepaid hours.
No surprises.
We work by the hour, from prepaid hour packs — the same packs you can use for any eYssen service. Before we start, you get an hour estimate, and we never go beyond what you approved without asking.
Book a free consultation →// Hour packs
Pick a pack. Use it for anything.
Prepaid hour packs with guaranteed reaction times
Starter Pack
8 hours*
Prepaid
Max reaction time: 5 working days
Basic Pack
25 hours*
Prepaid
Max reaction time: 4 working days
Standard Pack
50 hours*
Prepaid
Max reaction time: 3 working days
Advanced Pack
100 hours*
Prepaid
Max reaction time: 2 working days
Professional Pack
200 hours*
Prepaid
Max reaction time: 1 working day
Inquiry and first consultation
max 1 hour*
Max reaction time: 5 working days
*The hour allowance is flexible: a junior consultant makes your time go further, while immediate phone or online support uses it faster because of the priority surcharge.
Ad-hoc hourly rate
1 hour
Postpaid — invoiced right after the session
Max reaction time: not specified
Junior consultant 30% off
For suitable tasks (such as onboarding or support) you can ask for a junior consultant: only 70% of the time is charged. If they can't answer right away, they research it and follow up — that extra effort is not billed.
Immediate phone or online support +50% time
Without a pre-booked appointment, the time used is charged at 150%.
Billed per half hour · packs valid for 12 months · prices net of VAT · one balance of hours for AI, Odoo support and development
// Typical tasks
How long it takes. What it saves.
Averages from our work so far. Your exact estimate comes after the free consultation.
| Task | Typical time | Typical saving | Fits in |
|---|---|---|---|
| Incoming invoice → data extraction → Odoo bookkeeping | 4–8 h | 2–3 min per invoice — ~10–15 h a month at 300 invoices | Starter |
| Webshop product descriptions and translations (EN/HU/ES) | 4–10 h | 10–15 min per product | Starter–Basic |
| Enquiry e-mail → CRM lead, enrichment and draft reply | 8–16 h | 5–10 min per lead | Basic |
| Bank statement matching suggestions | 8–20 h | 30–50% of the monthly reconciliation time | Basic |
| Weekly management report from several systems | 16–30 h | 3–5 h a week | Basic–Standard |
| First-line support answers from your knowledge base | 25–50 h | 20–40% fewer tickets handled by hand | Standard |
| E-mail order → stock check → sales order, with exception handling | 40–80 h | 5–15 min per order | Standard–Advanced |
Indicative averages, not a quote. Times include design, build, testing and handover; savings depend on your volumes.
// AI keeps changing
AI doesn't stand still — so your automation can't either.
An automation is never really finished. The models underneath it change constantly: providers ship new versions every month, retire old ones on weeks' notice, shift prices, and quietly change how they behave. Left alone, an automation drifts, gets slower or less accurate, and eventually breaks when the model it relies on is switched off. So we don't build once and walk away — we keep shaping it as AI moves.
// What keeps changing
- Models get retired. The one we built on can be switched off — we migrate you before it bites.
- Better, cheaper models appear every month. We re-test and move you to the best one — quality up, token cost down.
- Providers quietly change behaviour and limits. Outputs drift; we catch it before your customers do.
- Your business changes too — new products, cases and formats. We keep the agent in step.
// So every month we
- Monitor for silent failures and fix breaks
- Re-run the test suite across current models
- Swap to the best model when it helps
- Update prompts, skills and knowledge
- Review guardrails, access and security
- Report on what it saved and how it performs
That is why every automation we ship comes with Care — a small monthly hour allowance that keeps it in step.
Care: typically 2–4 hours a month, from your hour pack. The AI we spend on monthly testing and monitoring is included. For live use we choose the AI solution together, based on the task — eYssen AI Gate or any other provider or model that fits best.
// Our promises
No guesswork. No lost hours.
01
No surprises on the invoice
Before we start, you get an hour estimate. We never go beyond what you approved without asking — if more is needed, we tell you first and you decide.
02
No hour is ever lost
Stop whenever you like. Unused hours stay valid for 12 months for any eYssen service — AI, Odoo support or development.
03
You see every half hour
Every half hour is logged with what it was spent on, and you can follow it live on your customer portal.
04
The solution is yours — and measured
You get the code and the configuration, with no licence fee and no lock-in. On request, we record the starting point and report the real saving afterwards.
// What's included
In every build — and what's not.
So there are no misunderstandings: here is exactly what every engagement covers, and what sits outside it.
// Always included
- ✓Discovery from your questionnaire
- ✓Solution design and scoping
- ✓Building and integrating into your existing tools
- ✓Testing across multiple models and providers
- ✓Guardrails and human-in-the-loop where it matters
- ✓Documentation and handover
- ✓Monitoring and a monthly report
- ✓A named contact you can reach
// Not included / billed separately
- —Your automation's own live AI usage — on the AI solution we choose together for the task (e.g. eYssen AI Gate or your own provider); other SaaS on your own accounts
- —Your own infrastructure or hardware
- —New workflows beyond the approved estimate (estimated and approved separately)
- —24/7 SLA and on-call (enterprise add-on)
- —Formal EU AI Act governance (available as an add-on)
// Where the hours go
More than a few hours of typing.
A working automation looks simple. Getting there is not — and most of it happens before you ever see a result.
Research and pre-work
We study your domain, your data and the current model landscape before we build anything. Preparing the data alone is 20–40% of a typical AI project.
Testing across models
We don't guess which model or prompt wins. We run the same task through many models, providers and settings, scoring quality, cost and reliability. Agents burn 5–30× more tokens per task than a chatbot — most of it on evaluation you never see.
Subscriptions and token burn
Dozens of paid AI tools and API subscriptions, and a large volume of tokens spent on tests and iterations — much of it before a single deliverable exists.
Redone for every task, because AI moves
Models change every month; what was best last quarter may not be now. Each new client task means fresh research, fresh tests and fresh token spend — often outside billable hours. You get today's best answer, not last year's.
In short: the hours include research, testing, guardrails, security and monitoring — not just the final delivery. And you see every one of them.
// Go deeper
Curious why this is so involved — and why it costs what it does? These parts of Learn AI explain it in plain language:
AI Basics · Models & Market · Agents & Automation · Data, RAG & Self-Hosted AI
// The real question
Don't ask what it costs. Ask what it changes.
An automation is not an expense line — it's an asset that keeps working. Weigh it against two things, not one.
// What you save
- ✓Hours returned to your team, every week
- ✓Fewer errors, less rework
- ✓Faster turnaround and lower cost per transaction
- ✓A task that took hours a week now takes minutes
Savings show up first — often paying back the build within a quarter.
// What it unlocks
- +Capacity for work you couldn't take on before
- +New services you can offer
- +New customer segments — 24/7 and multilingual
- +Growth without adding headcount
Revenue upside compounds — it doesn't reverse like a one-off cost cut.
// Measured, not self-reported
+15%
average productivity gain, measured on 5,172 support agents (peer-reviewed, QJE 2025)
+30%
for the least experienced staff — this is where most of the gain lands
+16.3%
the best of seven AI workflows tested head to head; the others barely moved
These are not our averages — they are the strongest public measurements available, and they are held down by rollouts that automate the wrong process or leave the way of working unchanged. The spread is not random: in the seven-workflow experiment one and the same company, in the same week, measured +16.3% on the best workflow and next to nothing on the others. On a well-chosen process the result can land well above these figures. We design each build so the saving alone justifies the price — the upside is yours.
// Add-ons
Bolt-on when you need it.
AI Governance (EU AI Act & GDPR)
Inventory, risk classification, DPIA, policy and training. Typically 40–60 hours.
Team enablement & training
Workshops so your people can run and extend automations themselves. From your hour pack.
Only need AI access for your own system?
eYssen AI Gate gives you every leading AI model behind one API key, with automatic model selection — prepaid from 5 € or a monthly plan from 9 €. For the automations we build, any AI solution is on the table: we always decide together, based on the task, what fits best.
Gate pricing →60 minutes.
0 €.
1 action plan.
In one hour we'll map your top three workflows to automate, what it costs, and what it returns. Not a sales call.
Book the audit →