Three plans. Pilot first, then scale.
Each plan is fixed-fee for the pilot — no SaaS subscription, no per-seat tax. You own the agent.
Starter
One agent, one channel, one CRM.
Who it fits
You want to test one agent on one channel and one CRM before committing more.
On day 30
A live agent and a day 30 results review against the KPI target agreed before kickoff.
- ✓ Lead Qualifier or FAQ agent
- ✓ One channel (web, DM, or voice)
- ✓ One CRM integration
- ✓ Day 30 results review
- ✓ Pilot fee refunded if the KPI target is missed
Practice
Two agents, multi-channel, CRM hygiene on top.
Who it fits
Leads come from several channels, and you want qualification, FAQ and CRM hygiene together.
On day 30
Two live agents with CRM Hygiene, plus the audit and ROI spreadsheet, reviewed against the agreed KPI.
- ✓ Lead Qualifier + FAQ agent
- ✓ Multi-channel (DM + web + email)
- ✓ CRM Hygiene module included
- ✓ Audit + ROI spreadsheet
- ✓ A dedicated delivery team, founder on every call
Operator
Full stack: lead, FAQ, CRM, voice. Multi-location.
Who it fits
You run several locations or languages and want lead, FAQ, CRM and voice as one stack.
On day 30
The full stack live — four agent modules, voice, four languages — reviewed against the agreed KPI.
- ✓ All four agent modules
- ✓ Voice agent included
- ✓ Multi-language (UA · EN · RU · ES)
- ✓ Quarterly review with founder
Compare the plans
| Feature | Starter | Practice | Operator |
|---|---|---|---|
| Agents | 1 (Lead Qualifier or FAQ) | 2 (Lead Qualifier + FAQ) | All four modules |
| Channels | 1 (web, DM or voice) | Multi-channel (DM + web + email) | Multi-channel, voice included |
| CRM integration | One CRM | Included | Included |
| CRM Hygiene | Not included | Included | Included |
| Voice agent | Only as the one channel | Not included | Included |
| Multi-language (UA · EN · RU · ES) | Not included | Not included | Included |
| Audit + ROI spreadsheet | Free audit before the pilot | Included | Included |
| Refund if the KPI target is missed | Included | Included | Included |
| Review with the founder | Day 30 results review | Founder on every call | Quarterly, with the founder |
After the pilot you pay for an operator, not for hosting.
From €49/month plus tokens. Here is what is inside.
Supervisor on every reply
A second model reads each answer before it reaches the customer and blocks anything outside the agent's scope.
Dialog review
Dialogs are scored against a quality rubric. Findings turn into prompt and harness fixes.
Fixes through the eval set
Every change runs against the eval set before it ships, so nothing that already works gets broken.
Hosting and monitoring
Uptime, tokens and error alerts on your infrastructure. Cost per dialog visible to you.
Two ways to run it after the pilot
Run
from € 49 / mo + tokens
- ✓ Hosting and monitoring on your infrastructure
- ✓ Supervisor model on every reply
- ✓ Automated dialog scoring, weekly digest to you
- ✓ Prompt and harness fixes on request
Managed
from € 249 / mo + tokens
- ✓ Everything in Run
- ✓ An engineer reads the flagged dialogs every week
- ✓ Prompt and harness fixes included, each run through the eval set
- ✓ Monthly KPI report, quarterly review with the founder
Run: €49 / €99 / €149 per month for Starter / Practice / Operator. Managed: €249 / €390 / €590. Tokens billed at cost, per dialog.
The build is the cheap part.
Four numbers from independent research, sources linked. A cheaper build is not cheaper if nobody operates it afterwards.
of agentic AI projects will be cancelled by the end of 2027: escalating costs, unclear business value or inadequate risk controls.
Gartner, June 2025 ↗of the five-year cost of an AI deployment is maintenance, scaling and optimization, not the initial build or license.
Maven AGI ↗real first-year cost against the vendor quote once the hidden costs land. The quote covers 50–60 % of what gets spent.
Hypersense Software, 2026 ↗of total cost of ownership is AgentOps: monitoring, prompt iterations, tuning. Unlike software, it does not decline after year one.
Korvus Labs ↗That is what the run-rate pays for: the part that does not stop after launch.
Let's do the math out loud.
A worked example for a typical inbound flow — plug in your own numbers on the audit call.
Run your own leak math.
An estimate on your numbers. The exact math happens on the free audit.
An estimate, not an offer. Formula: leads × % lost × conversion × deal value. The agent recovers half of the loss; the run-rate from €49/mo is already included.
Defending the budget in front of the CEO? Here's the case.
Three arguments that hold up in a meeting. The free audit supplies the rest of the numbers.
Cheaper than hiring
A manager costs €1,500–3,000/mo plus 2–3 months of onboarding. The agent pilot is €1,800 once and from €49/mo after. Run the calculator above.
Prepaid, refundable
You pay the €1,800 pilot upfront. The KPI target is in the contract; if it isn't reached by day 30, the fee comes back. The budget risk is one line item, not a subscription.
Numbers before you start
The free audit delivers the where-leads-die spreadsheet and an ROI calculation. Walk into the meeting with data, not a slide deck.
Aelo — AI call analytics for your CRM
Not ready for a custom agent? Aelo listens to your sales calls, scores conversations and suggests next steps right in the CRM. Plans from $59/mo, with the first 100 minutes free every month.
Eight questions every client asks.
What if it doesn't work?
Results review on day 30 against the KPI target agreed before kickoff. The pilot is paid upfront — we don't start without payment. If the agent is live and the target isn't reached, we refund the pilot fee in full; the agent is switched off and your data is returned.
Do we own the agent?
Yes. Code, prompts, eval set, training data — all yours. We deploy on your infra (Cloudflare or AWS) under your account.
What about data security?
Deployment in the EU region. PII redaction on logs. DPA signed on request. Built around GDPR principles.
Can you integrate with our CRM?
If it has an API: yes. We've shipped on Bitrix24, HubSpot, Salesforce, AmoCRM, Pipedrive. If yours is custom — we'll tell you on the audit call.
How fast can we start?
Audit call within 48 hrs. Pilot kickoff within 7 days of signed scope. Days 8–10: agent in shadow mode. Days 11–14: live.
Is this just GPT-4 with a system prompt?
No. We use multi-step reasoning, retrieval over your scripts, structured outputs to your CRM, and a supervisor that audits every reply. Eval harness ships with the build. In plain terms: the agent reasons step by step, checks itself against your scripts, writes to your CRM in a strict format, and a second model reviews every answer.
What if the agent says something wrong to a customer?
Three layers of protection: a supervisor checks responses, guardrails block topics outside its competence, hard cases go to a human. For the first days the agent runs in shadow mode — your team approves every reply before it goes live.
The team is afraid of being replaced. What do we tell them?
The agent takes the routine, not the people: 'how much is it', statuses, CRM logging. Managers get time for complex deals and full context on every lead.