Launch your NLG Project in under 30 days: our startup vision coming true

Starting out from a tech provider…

Four years ago, when we at AX Semantics were building NLG solutions for our customers internally, we decided on building a software where everyone could do his own content automation project.

Looking down that path, we anticipated that we not only had to build a tool that everyone could use, but in addition, to teach the required skills of how to identify and build such a project to our users (and customers).

Now, lots of iterations on our product, eLearning and onboarded users and customers later, we can finally say, that we have hit a major milestone:

Our users now require less than one month from start to finish for their own projects!

Defining the stages of a new "text robot" project

This includes

  1. buying the software (easily solved ;-))
  2. learning how to use it
  3. acquiring the necessary skills about the concept of Natural Language Generation and how to build an automated text project
  4. identifying a project that is doable for NLG and worthwhile
  5. providing data, and - most of the time - treating and cleaning that data for suitability
  6. quality assurance on the output
  7. API setup and automation
  8. text production at scale
  9. and roll-out of the content to the end-users

To make this easier, these 9 steps are called the "onboarding" of a new customer, where every single customers gets supported by a dedicated person from our seven-strong Customer Success Team member.

By the way: these steps are always the same, irrespective of the use case. This process has been used for product descriptions from PIM system data, for journalistic applications like the air pollution reporting or financial reporting cases.

Building a mass-market product by focussing on metrics

One of our core metrics is the on-boarding cycle time per customer - making the improvements visible and measurable: Our NLG software and editing tool, our various self-service tools like eLearning and documentation, and our customer support and processes.

We already knew it was possible to educate anyone to be able to build such a system, because of the academic supervision for our self-service tool and eLearning: where our last cohorts of standardised testing had already provided us with the evidence about our successfull knowledge transfer by our interactive eLearning via the so called "Seminar".

…fast forward to today: we did it!

And as of last week, we have officially reached this milestone for our customers as well: the average and median on-boarding phase cycle time for our customers now is below 30 days: most customers successfully launch their project in under one month! - Without having an prerequisite knowledge of content automation, NLG systems or other computational linguistic training.

Duration of the first text automation project for our customers: showing the onboarding cycle time of customers charted over time. The rolling average (thick blue line) is now below 31 days.

The chart proves us correct (reading: the on-boarding cycle time of customers charted over time), we made Natural Language Generation as a technology so easy, that is available to everyone.


Everybody can now create content from his data.

No skills (besides copywriting skills ;-) ) necessary!


envelopephone-handsetmap-marker linkedin facebook pinterest youtube rss twitter instagram facebook-blank rss-blank linkedin-blank pinterest youtube twitter instagram