AI email analysis
AI reads incoming emails and attachments, recognises enquiries, orders and invoices, and moves the data into your systems.
AI solutions
Artificial intelligence is useful when it does a specific job inside your systems, not when it sits apart as a chat window. We connect it to your documents, data and software, and decide what it may do on its own and what a person must approve. We log its actions, so you can check what was done and why.
Example
Example: a fictional company and made-up data.
AI assistant for staffSearches 48 company documents
What is the returns procedure for wholesale customers?
Done by AI: For wholesale customers, returns work like this:
- Undamaged goods in their original packaging can be returned within 14 days of delivery. source 1
- The return has to be agreed with the sales manager in advance. source 1
- Returns in a quarter may not exceed 10% of that quarter's purchases. source 2
- The money is refunded within 10 working days of the goods being checked. source 2
Sources:
- source 1Returns policy, p. 3
- source 2Wholesale agreement, clause 7.2
Sources
The assistant answers only from the company's documents. If the answer isn't there, it says so.
- source 1
Returns policy, p. 3Updated 12 Jan 2026
“A wholesale customer may return undamaged goods in their original packaging within 14 days of delivery. The customer informs their sales manager in advance and agrees the date of the return.”
- source 2
Wholesale agreement, clause 7.2Agreement template, 2026 edition
“The value of goods returned in a quarter may not exceed 10% of the Buyer's purchases in that quarter. The Seller refunds the money within 10 working days of checking the returned goods.”
What it covers
- AI assistants and chatbots that answer from your own documents and data (Websites with an AI assistant)
- Recognising invoices, orders and emails: the data is captured into the right fields and passed to your systems
- Sorting enquiries and routing them to the right member of staff
- Reply drafts that staff review before sending (AI email analysis)
- Connecting AI to your systems with controlled access, including through MCP servers (MCP servers for accounting)
- An honest assessment of where AI pays off and where something simpler will do
We are also building our own product – a multilingual AI voice agent for service companies.
Who it suits
- Every day your staff retype data from invoices, orders or emails into your systems.
- Customers keep asking the same questions, although the answers are already in your catalogue, manuals or policies.
- You want your staff to use an AI assistant, but can't let it see all of the company's data.
How the work goes
- 1We find out what you need
- 2We analyse and propose
- 3We design the solution
- 4We build in parts
- 5We test and launch
- 6We maintain and improve
Related solutions
MCP servers for accounting
Let an AI assistant work with your accounting data – securely, and only within each user's permissions.
Examples of our work
Online shop with an AI assistant and a customer portal
2026Launching
A new website and online shop with an AI assistant: orders, enquiries and chats go straight into the CRM.
Every old address was kept or redirected, and AI chats enter the same enquiry pipeline as completed forms.
Next.js, TypeScript, MongoDB, language model
Invoices from the mailbox – straight into the books
2026
We built invoice intake from the mailbox: AI recognises supplier invoices, and clear rules book them with no manual work and never declare VAT twice.
Three separate checks stop duplicates.
TypeScript, Hono, Next.js, IMAP
AI modules in a client's business management and accounting systems
2025–2026
We built AI modules that run in a client's business management and accounting systems: invoice recognition, a proposal generator, assistants for accountants and reports, and MCP servers.
The assistant carries out operations only after the user approves them, and every approval is logged.
language models, document recognition, MCP, OAuth 2.0
Enquiries from every channel and email conversations – in the CRM
2026
Enquiries from website, Facebook, Instagram and Google Ads forms reach the CRM once, and each one shows the full email conversation.
Spam is flagged automatically, and a real enquiry is never lost.
Node.js / TypeScript, AWS API Gateway, SQS, Terraform
AI that works in desktop software through MCP
2026Prototype
An MCP server through which an AI assistant sees the screen of a program running in a virtual machine and works its mouse and keyboard – where the program has no API or the API can't do everything.
Built for a Windows accounting program, but it works with any desktop program.
MCP, Windows virtual machine, screen capture, mouse and keyboard control
Langer – a messaging and calling app with a built-in interpreter
2026Beta
A messaging and calling app with a built-in interpreter: chats and calls are translated as you talk, so everyone can speak their own language.
The working prototype already translates chats, voice calls and face-to-face conversations, and beta sign-ups are open.
iOS and Android, speech recognition and synthesis, AI translation, real-time back end
Questions and answers
How do you keep the AI from making up answers?
We instruct it to answer only from your documents and data, which it reaches through clearly defined tools, and we test this on real questions before launch. When it can't find the answer there, it says so or hands the question to a member of staff.
Will our data be used to train AI models?
No. We use business AI services whose terms exclude training on your data, and the AI sees only the data the task needs.
Which task should AI take on first?
One that repeats every day and takes a lot of time. We test the AI on real examples from your work and switch it on in daily operations only once the results are reliable.
Got a task or an idea?
Describe it in a few sentences – we'll reply and talk through how we could help.