Knows your data and rules
It answers from your documents and follows your rules - not general knowledge scraped off the internet.
AI agents · integrated with your systems
An AI agent that knows your companyand works inside your tools.
Off-the-shelf chatbots know the internet but not your data or your rules. We build an agent wired into your email, CRM, calendar and document base - one that actually gets work done, not just talks.
- Integrated with your systems
- Runs on your data
- GDPR compliant
You start by describing the task - the rest, from integrations to rollout, is on us.
What an AI agent for business is
An AI agent is a program that runs a multi-step case on its own inside your tools: it reads an email, pulls the data out of it, checks it against your CRM and knowledge base, drafts the reply or the document, and hands it to a person for approval. A chatbot answers a question and stops there; an agent takes the case to the end. AppWave builds agents in Python, wired into the tools a company already uses, and the client sets the limits of their autonomy before the build starts - the final click always belongs to a human.
AI agent, chatbot and off-the-shelf tool: the differences
| AI agent | Chatbot | Off-the-shelf SaaS | |
|---|---|---|---|
| Access to your data | Reads the CRM, the mailbox and the document base, on permissions you grant. | Only what someone pastes into the chat window. | Only the data it collects inside its own module. |
| Acting in your systems | Writes and updates records in the tools you already use. | Writes nothing. It stops at the answer. | Works in its own interface; integrations within the price plan. |
| Scope of decisions | You agree before the start what it does alone and what goes for approval. | No decisions, only the wording of a reply. | Rules from the vendor's list, no exceptions. |
| Entry cost | From 3,875 PLN net a month, care and fixes included. | The lowest of the three, because the scope is the smallest. | Typically 200-2,000 PLN a month per seat. |
| Time to deploy | Counted from the signed scope. The first process goes live before the next ones start. | Short, because there are no integrations to build. | An account straight away; the time goes into bending the process to the tool. |
Access to your data
- AI agent
- Reads the CRM, the mailbox and the document base, on permissions you grant.
- Chatbot
- Only what someone pastes into the chat window.
- Off-the-shelf SaaS
- Only the data it collects inside its own module.
Acting in your systems
- AI agent
- Writes and updates records in the tools you already use.
- Chatbot
- Writes nothing. It stops at the answer.
- Off-the-shelf SaaS
- Works in its own interface; integrations within the price plan.
Scope of decisions
- AI agent
- You agree before the start what it does alone and what goes for approval.
- Chatbot
- No decisions, only the wording of a reply.
- Off-the-shelf SaaS
- Rules from the vendor's list, no exceptions.
Entry cost
- AI agent
- From 3,875 PLN net a month, care and fixes included.
- Chatbot
- The lowest of the three, because the scope is the smallest.
- Off-the-shelf SaaS
- Typically 200-2,000 PLN a month per seat.
Time to deploy
- AI agent
- Counted from the signed scope. The first process goes live before the next ones start.
- Chatbot
- Short, because there are no integrations to build.
- Off-the-shelf SaaS
- An account straight away; the time goes into bending the process to the tool.
Why
Off-the-shelf AI doesn't know your company.An agent built for you does.
A generic chatbot guesses. An agent built for you answers from your documents, sticks to your rules and works in the systems you use every day.
Works inside your systems
It reads and writes in email, CRM, the document base and calendar - through APIs and webhooks, not copy-paste.
Takes over repetitive work
It picks up the tedious tasks nobody wants and does them without fatigue or breaks.
How we work
From the task scopeto an agent that grows with your needs.
Four steps that turn a task description into a working agent.
- 01Step 1
Scope and the agent's task
We define exactly what the agent should do, which process it works in and which systems it must connect to.
- 02Step 2
Integrations and data
We connect email, CRM, calendar and the document base. The agent starts working on your data, not general knowledge.
- 03Step 3
Rules and security
We set the rules the agent follows and secure the processing of sensitive data in line with GDPR.
- 04Step 4
Rollout and growth
We launch the agent on production, train the team and keep developing it as needs grow.
What you get
An agent embedded in your process,not a demo on a slide.
The four parts that make up every agent we build for you.
Wired into your tools
Integrations with the systems you already use - no ripping out the whole stack.
- Email, CRM, calendar, document base
- Connections via API and webhooks
Runs on your data
Answers grounded in your documents, not the model's general knowledge.
- Search across your knowledge base
- Answers based on company sources
GDPR
Sensitive data kept safe
Personal-data protection where it's needed - with restricted access.
- Sensitive data under control
- GDPR and EU AI Act compliant
Rollout and care
Launch on production, team training and ongoing development of the agent.
- Live on production, not a demo
- The agent grows as needs grow
Numbers
Numbers you can checkbefore you ever contact us.
No savings percentages pulled out of thin air. What stands below is only what you can verify in the calculator and in the contract.
- Approves every decision that carries weight
- A human
- Sending a quote, replying to a client, releasing a payment - the agent prepares, you click. We write down the limit of its autonomy before we start.
- Failure fixed, included
- within 16 h
- Response within 8 hours, fix within 16 hours on business days. Included in the package, with no separate service contract.
Prices and response times come from the calculator and the standing service agreement, so you can verify them before we ever speak. We publish no savings percentages here because they depend on the process; we calculate them for you during the audit.
Evidence
14 days from start to production
Szkoła Maturzystów
Every lesson ends with homework, and the marking grows with the school. There are more than five thousand students.
We built the grading engine as an API. It scores a solution against criteria, points out flaws in the reasoning and remembers the ones that repeat.
Marking stopped growing with the number of students, and a student learns which mistakes they make systematically.
See the case studyWhy us
Agents that work,not ones that look good in a deck.
Three reasons our agents make it to production instead of a drawer.
Production stack
Agents are built on real infrastructure, not a demo. They're meant to work in real operation, not just on a slide.
We connect real systems
We know how to wire an agent into company tools, databases and processes so it actually does the work, not just chats.
GDPR from day one
Where customer data is involved, we design the agent so data processing is secure and lawful from the very first day.
Guarantee
The risk sits with us,not with you.
The Working Implementation Guarantee
If the agent does not handle the process the way we agreed in the scope, we fix it at no extra cost for 12 months after handover - no new quote, no cap on revisions.
We can promise this because we start with a process audit and turn down projects where we cannot see a return. Saying no before we start costs both sides less than fixing things afterwards.
- The code and the repository are yours from day one; property rights transfer with each stage payment.
- Scope and price are agreed before we start. We do not add line items along the way.
- After the rollout you keep the same person you talked to at the beginning.
What it costs
| Package | Hours per month | Net price | Incident response |
|---|---|---|---|
| Basic | 25 h / month | 3875 zł | within 8 h |
| StandardMost often chosen | 50 h / month | 7000 zł | within 8 h |
| Growth | 90 h / month | 11 750 zł | within 8 h |
Basic
- Hours per month
- 25 h / month
- Net price
- 3875 zł
- Incident response
- within 8 h
StandardMost often chosen
- Hours per month
- 50 h / month
- Net price
- 7000 zł
- Incident response
- within 8 h
Growth
- Hours per month
- 90 h / month
- Net price
- 11 750 zł
- Incident response
- within 8 h
Tell us what your agent should do.We'll quote it and show how we'll build it.
Just describe the task. The rest - integrations, data, security, rollout - is on us.
Let's build your agent45 minutes · free · no obligation
- 5.0 / 5 - 16 reviews
- Uptime guarantee - if a system breaks, the fix is included, no extra charge.
- You talk to the owner, not a salesperson
office@appwave.dev+48 538 441 413appwave.dev
FAQ
Common questionsabout AI agents with integrations.
A chatbot waits for a question and answers with text. An agent acts on its own: it picks up a request, checks data across several systems, works through the steps in order, and either closes the case or hands it to a person with the context already assembled. The practical difference is that after a chatbot conversation someone still has to go and do the thing. After an agent has worked, what is usually left is an approval.
Most often email, CRM, calendar and a document base, but we integrate with the systems you already use, through APIs and webhooks. We define the scope up front, based on the process the agent will handle.
Yes, when it's well designed. We secure the processing of sensitive data, restrict access to what's necessary, and ensure GDPR and EU AI Act compliance - especially where customers' personal data is involved.
It depends on the scope and the number of integrations. After a short conversation about what the agent should do, we prepare a quote and show how we'll build it. Start by describing the task and we'll propose a concrete plan.
We assume it sometimes will, and design for that. Decisions with outside consequences - sending a quote, replying to a client, releasing a payment - need a human click. Every step the agent takes is logged, so you can see what it based a conclusion on. Where a mistake would be expensive, the agent does not act alone.
Almost every client asks this, and rightly so - in EY’s 2026 study, 39 percent of companies named security as the biggest barrier to adopting AI. We agree with you which data reaches a model at all and which never leaves your infrastructure. The processing agreement and the record of activities are part of the rollout, not an afterthought.
A chatbot stops at the answer; an agent stops at the finished case. A chatbot knows what someone pastes into the chat window; an agent reads your CRM, mailbox and document base on the permissions you grant, then writes the result back into the same tools. You agree the scope of its independence before the start: what it does alone, and what it brings to a person for approval.
When your process fits a ready-made module without being bent to suit it, and when you do not need anything written back into your systems. At that point 200-2,000 PLN a month per seat is cheaper than anything we would build, and we say so plainly. An agent starts to pay off when a process crosses several tools at once, or when the vendor's rules do not cover your case.
Other servicesThis might be your problem too.
MVP for startups
A system that runs in production, not a demo. Fixed scope and a fixed date, agreed before anyone signs.
Custom web software
Off-the-shelf tools force you to compromise. We build a system around your process, with AI where it genuinely gives an edge.
AI agents and process automation
Repetitive work a machine does faster and error-free. AI only where it pays off.
Your IT team on a monthly plan
Five subscriptions, zero order? We take over licences, access and security. One partner, one invoice.
AI implementation review
Before you spend a single euro - we show, in numbers, where AI pays off and whether it's legal.
See all services
Seven ways to put AI to work. See where to start.