Skip to content
SSyntalith
PLEN
Book a scan

Custom AI apps for business

A custom AI app is built around one process when a ready-made tool does not fit or company data must stay in an approved environment. It can be a copilot, document search tool, data extractor or case classifier. Prices start at €6 000 net and launch takes 4–10 weeks. We test quality on client examples, route uncertain answers to a person, and hand over the code and test set.

Book a free process scan (30 min)
See Company Brain AI

In short

Price and time
from €6 000 net · 4–10 weeksprice follows integrations, data sources, and quality requirements; fixed quote before the contract
How we guard quality
Accuracy measured on your dataaccuracy checked on your documents before launch; uncertain cases go to a reviewer
What stays yours
The code, system instructions, and test examplesno lock-in to a single provider
For whom
When a ready-made tool does not fitfor internal tools when data must remain in an approved environment
Book a free process scan (30 min)

Problem

Your team spends hours hunting for knowledge scattered across documents and pulls data from PDFs and scans by hand, because ready-made tools do not fit the process or cannot use your data. A quick API call bolted onto an app can look good on sample data and break in production. One made-up answer delivered with confidence is enough for the team to stop trusting any of them. Those hours have an hourly rate; the status quo is not free.

Outcome

In 4–10 weeks you get an application where a language model performs a specific, repeatable task: answering from cited documents, classifying cases or extracting invoice data. We measure accuracy on your data and route uncertain answers to a reviewer. You commit to the full build after the pilot meets the accuracy target.

How we control risk

Each risk has its own control in a running system.

The control sits where an error could harm the user, a document, or the company.

  • Legal research · answer integrity

    Case-law research

    The system creates an answer after finding an exact source passage. Questions that require legal advice go to a lawyer.

    Open the deployment

  • Internal knowledge · access control

    Permission-aware memory

    Across 750 trials on 600 test documents, five fictional accounts received only content allowed by their permissions. We repeat the control in the company's environment before acceptance.

    View the access-control resultsSee the Company Brain AI scope

  • Confidential documents · on premises

    A document flow with no internet egress

    Documents and answers remain on company infrastructure. The system blocks external connections, while an administrator controls data sources and updates.

    Open the deployment

For each project, the controls and quality threshold are set on data from the client's process.

  • Artem Lisovtsov

    Founder, Full-Stack & AI/ML Architect

  • Serhii Ivanchatenko

    Head of Backend & AI Integrations

This is how one sourced answer is built.

From a question or a document to a checkable answer. The confidence gate marks where a person takes over.

  1. 01

    A query arrives

    a user question or a document enters the app

  2. 02

    Source search

    the app retrieves and reranks the right passages in your data

  3. 03

    Answer with a source

    the model answers, pointing to the passage it uses

Confidence threshold

confident answers go to the user, doubtful ones wait

Answer with the source cited

a confident case reaches the user with its source and decision history

Uncertain answer to a human

a doubtful case the app does not serve on its own

We set the confidence threshold where a human takes over on your data during scoping. Operations with production impact can require approval.

The difference shows when the model is wrong.

Bolted-on API

  • Quality

    “Works on my examples”

  • Hallucinations

    Reach the user

  • Boundaries

    The model does as it pleases

  • Trace

    No idea where the answer came from

  • Ownership

    Locked in someone else's ready-made tool

  • Cost and time

    Lower entry, a subscription you switch on today

An app built around the process

  • Quality

    A fixed example set and measured accuracy on your data

  • Hallucinations

    Safety rules, source checks, and a handoff to a person

  • Boundaries

    Content from documents is data rather than commands; actions under conditions and minimum permissions

  • Trace

    A log of input, sources and decision

  • Ownership

    Code, system instructions, and test examples are yours

  • Cost and time

    Higher upfront cost, with hosting, models and maintenance accounted for separately instead of a permanent product licence

A build costs more upfront and takes longer than switching on a ready-made solution. With growing seat counts or volume it can have a lower multi-year cost, so we compare 36 months of licences, integration, migration, hosting, models and maintenance. When the ready-made tool wins that calculation, we recommend buying it.

An app built around the process

Answer with a source

a confident case reaches the user with its source and decision history

Sources

  1. Passage from your knowledge baseRAG
  2. Record from your systemRAG

Answer and rule checks

  • Safety rules, source checks, and a handoff to a person
  • Content from documents is data rather than commands; actions under conditions and minimum permissions
  • A log of input, sources and decision

Code, system instructions, and test examples belong to you

A named package in this line

Company Brain AI

An assistant that answers from your own documents and shows where the answer came from.

A packaged build inside the AI Apps line. The assistant reads your procedures, contracts, offers and system notes, answers in full sentences and names the document it relied on. With no basis in the corpus it says "I don't know" and points to the person who owns the topic.

How it behaves at work

  • 01

    Answers from company documents

    The corpus is your procedures, contracts, offers and system knowledge. Each answer is composed from retrieved passages, with no filling of gaps from the model's training memory.

  • 02

    Cites the source

    Every answer carries the document name and the passage it rests on, so an employee can verify it in seconds.

  • 03

    Says "I don't know"

    When retrieval finds no basis, the assistant declines to answer and routes the question to the person who owns the topic.

  • 04

    Inherits permissions

    Each person sees only the documents they can already open in the source system. Permissions are resolved in the retrieval layer, before a document reaches the model.

When a separate search layer is worth it

No

fixed page threshold

Simpler route
Working over one file set can be enough when the corpus is small, changes rarely, and has no complex access rules.
Separate search layer
It becomes useful with several sources, frequent changes, document versions or different permissions. We then combine semantic and keyword search and resolve access before retrieval.

Page count alone does not decide the architecture. We review document volume, update rate, file quality, user questions and permission requirements.

Price inside the AI Apps line

Entry
from €6 000 net
Typical implementations
€6 000–35 000 net
Maintenance
priced individually

The range follows the number of sources, required permissions, and scope of testing. You get a fixed quote in the proposal after the free scan, before any contract.

Book a free process scan (30 min)

In 30 minutes we determine which questions your documents can answer and whether ordinary search is enough or a dedicated app is justified.

Read further

What the build includes.

  • Internal assistants and panels working on your data
  • Answers from your documents with the source cited
  • Extracting data from invoices, contracts and scans into an agreed format
  • Classification of cases and documents according to your rules
  • A fixed set of client examples with measured accuracy
  • Safety rules and a human handoff for uncertain answers
  • Integration with your systems over APIs, on minimum permissions and with a log
  • Code, system instructions, and test examples belong to you

Data & compliance

Production cloud (AWS, Google Cloud, Azure) or your own infrastructure, minimum permissions and a decision log, uncertain answers through a human. We provide technical input to EU AI Act documentation (accuracy, transparency, oversight), but not legal advice. If the app talks to users or generates content, the Art. 50 transparency duty (disclosing it is AI) applies from 2 August 2026.

Not for

  • Marketing websites and landing pages
  • Web development with no AI component
  • A chat widget pasted onto a site
  • A process a ready-made tool handles cheaper and faster: we'll recommend buying it

Entry price

from €6 000

Delivery window · 4–10 weeks

How much does a custom AI app cost, and what affects the price?

You get a fixed quote in the proposal after the free scan. You know the full cost before signing, with no hidden items.

Book a free process scan (30 min)

What drives the price

  • The number of integrations and data sources
  • Quality requirements and the scope of testing
  • Security, consent, and action history

Cost over time

  1. Implementation buildfrom €6 0004–10 weeks
  2. Maintenance and oversightpriced individuallymonthly

Free process scan

Start with a free process scan.

  • A 30-minute call with the engineer who would lead the work.
  • A review of the processes that cost you the most time and money.
  • A written summary of what to automate first and the likely cost range.

The scan chooses one process to assess, and within 2 business days you receive a recommendation, including when a simpler route is the better fit.

€0

30 minutes · written takeaway within 2 business days

Book a free process scan (30 min)

Times are shown in your own time zone. We work with clients across time zones.

Describe the process in the form

AI app questions

  • What if a ready-made tool is enough?

    The scan recommendation will identify it when it handles the process at lower cost. A custom build makes sense when the process or data is too specific for available solutions.

  • How do you keep the model from making things up?

    We check the app on a fixed set of examples from your process, verify each answer against its source, and define the handoff to a person. Accuracy is measured before launch, and answers below the agreed threshold go to a reviewer.

  • What if the model gets it wrong in production?

    For each answer, we record the question, sources, and decision, so a mistake can be traced and fixed at its source. Safety rules hold doubtful cases back from the user, while maintenance improves the rules and test examples after launch.

  • Where does our data live, and does it leave the company?

    We deploy to an agreed cloud or your infrastructure, with minimum permissions and encryption. The app accesses the required data, while the code, system instructions, and test examples stay on your side. For personal data, we sign a GDPR data processing agreement and limit engineer access. The AI provider works under terms that exclude training on your data, or the system runs locally.

  • What about security and the EU AI Act?

    We work on minimum permissions and with a decision log, and uncertain answers go through a human. We also provide technical input to the operator documentation the EU AI Act expects (accuracy, transparency, oversight), but we don't give legal advice or classify the system.

  • What if the app doesn't work in our process?

    We set the quality threshold on your data before the build. If the scan or test shows that accuracy is too low for your process, we tell you before paid development begins.

  • Who maintains the app after launch?

    The same engineer who built it. After launch, maintenance covers monitoring, fixes, and updates to the safety rules and test examples as data and the process change. Scope and response time are set in the contract.

  • How much does a custom AI app cost?

    Implementations start at €6 000 net. Price follows the integrations, data sources, and quality requirements. The proposal gives you a fixed quote before signing.

  • Is the pilot included in the price?

    Yes. A pilot on your data is the first implementation stage inside the fixed price. If it does not reach the written accuracy threshold, we improve it within the same price, and the decision on full deployment stays with you.

  • When does a company knowledge base not need RAG?

    For fewer than roughly 500 document pages, a simpler assistant over one file set is often enough. For larger collections, we combine semantic and keyword search and check permissions before releasing a document. The free scan identifies the suitable approach.

The scan ends with a written recommendation.

  • A 30-minute call with the engineer who would lead the work.
  • A review of the processes that cost you the most time and money.
  • A written summary of what to automate first and the likely cost range.
€030 minutes · written takeaway within 2 business days
Book a free process scan (30 min)

Times are shown in your own time zone. We work with clients across time zones.

The scan chooses one process to assess, and within 2 business days you receive a recommendation, including when a simpler route is the better fit.

Full pricing