The AI platform from TechNative

SME AI Platform

One secure foundation in your own AWS account, onto which we connect your processes and systems module by module. Start, for example, with processing incoming email, searching documents, preparing quotes or writing up conversations.

The platform in short

One foundation, modular to extend

1 platform

All AI applications on one foundation, not six separate tools with six contracts.

Modular

You start with one use case and switch on the next module whenever it suits you.

Your own account

Runs in your AWS account in a European region, with your data under your control.

Per module

A monthly platform fee based on the number of modules, plus usage costs. No seats.

The idea

From scattered subscriptions to one foundation

What is the SME AI Platform?

Most SMEs start with AI through scattered subscriptions: a chatbot here, a summariser there. Handy to try, hard to build on. Your data sits in five places, nobody knows what it costs and nothing talks to your own systems.

In your AWS account we set up one secure foundation: identity, network, logging and cost monitoring. On top of that we put the AI platform, with models from Anthropic, Meta, Mistral and Amazon through a single entry point. Switching model is a setting, not a project.

After that we connect module by module to what you already use: document management, ERP, mailbox, ticketing system. The AI comes to the work, not the other way around.

The layers of the platform

The foundation: your AWS account in a European region, set up to AWS standards, with logging and budget alerts on. Above it the models through a single entry point, and above that your own documents and files, searchable with source references.

We build the bottom two layers once. After that every new module goes faster than the last.

You are not renting a service, you get a foundation

The environment sits in your AWS account in your name. The data, the environment and the configuration are yours. We build and manage, and the generic building blocks and our own software stay ours; the underlying services remain AWS services.

If the partnership ends, the platform stays standing and running. No proprietary data models locked in with a vendor and no export problem. There is, and this is only fair to say, a dependency on AWS: that does not move at the press of a button.

Building blocks

What can run on your own AI platform

Four building blocks you see back in every module. They are there from day one, even if you start with a single use case.

Answers from your own documents

Manuals, contracts and files go in. The platform answers on that basis and adds the source, so you can check every answer. That strongly limits made-up answers; verifiability therefore stays part of the design.

Connected to your systems

Look up an order, fetch a status, prepare a draft in your own application. You decide per system what is allowed and what is not.

Boundaries that are fixed

Topics that stay out, personal data that is filtered, answers that are blocked when the source is missing.

Seeing what it delivers

Quality, usage and cost per module in one overview. You choose on results instead of on gut feeling or the latest hype.

How it works

Module by module, with a way out at every step

1

Choose one job

We look for the recurring task with the clearest time saving, no broad inventory taking months.

2

Lay the foundation

The secure environment in your own AWS account, with logging, access management and budget alerts in place.

3

First module live

Working on your own data, for a small group of users. If it turns out not to work, it stops here.

4

Expand

More users, next module. The foundation is already there, so every step costs less than the first.

FAQ

Frequently asked questions

Want to know what suits your situation best? Ask Pim. Describe which work takes your team the most time. You get a straight answer about what can be automated, what is better left by hand, and when doing nothing is the cheaper option. Book a moment with Pim or one of the team.

Pim Snel

Pim Snel

Migration & modernisation

hello@technative.eu

No. Start with the work: which recurring task costs you or your team time every week? Together we pick one concrete job and decide whether AI can genuinely deliver something there.

No. With regular processing through Amazon Bedrock your input and answers are not shared with the model provider and not used to train the foundation models. For some models and configurations separate retention rules apply, so per use case we record which models are allowed. Read how Amazon Bedrock works

You decide that per module and per integration. Access rights, logging and the limits of what a module may retrieve are fixed in the foundation, not in the individual application.

No. The whole point is that the platform connects to what you already use: Microsoft 365, your documents, ERP, CRM, mailbox or ticketing system. Per integration we decide which access is needed.

The first module usually runs within a few weeks on your own data for a small group of users. If you want to know faster whether an idea holds up, the five-working-day AI Prototyping Sprint is the shorter route. See the AI Prototyping Sprint

The price has two parts: a fixed monthly platform fee for managing the environment and the modules that are switched on, and the AWS costs by usage. The fee depends on the number of modules, not on the number of employees, and the usage part is billed per piece of text processed. No licences, no minimum spend, no seats.

Further reading

Where to read on

Get started

You build the foundation once. After that every module goes faster.

Choose one recurring job, and we build a working module on your own data. For suitable engagements AWS funding programmes can cover part of the preparation; we check up front whether your organisation qualifies.