Askia is very pleased to announce that v7.1 of AskiaAnalyse comes with a beta version of an AI analysis assistant. Ask IAN is a chat interface that allows a user to ask questions about their Askia project and ‘IAN’ will go off and look at the questions in the project, decide which questions will best solve the query, run some tabulations and then provide a summary of the results. It’s an absolute game changer.
Ask IAN is a smart use of AI. AskiaAnalyse is still doing all the calculations locally – ‘IAN’ is simply selecting the relevant questions to investigate and then summarising the findings from the tables. The tables can be saved in the system, so that you always have a reference point for checking the numbers. Respondent level data does not leave a client’s environment.
IAN is very good at summarising what’s inside a survey: it detects the questions which it thinks are interesting as demographics, the ones that could be used as KPIs and the wave variable if there is one. It also understands loop structures. It knows that it makes sense to cross KPIs by demographics and it will do that by default. But a user can ask it to cross anything.
Ask IAN is efficient, as it only creates the analysis that it thinks it needs to provide a sensible answer to the user’s question. So, it is not creating hundreds of tables and burning through thousands of AI tokens to understand the project. And if IAN can’t find the relevant questions, it will come back and ask for more information to help him do the right analysis.
Where IAN is especially good is when analysing changes in a tracker survey. A user can ask it what has changed in the last month, the last quarter, the last year or the difference between specific years. It uses the relevant test to measure the significant changes.
Clients will be delighted to hear that the new capability is included in the cost of the software licensing. The only additional charges will be the cost of the AI tokens from your AI provider, used in running each query.
Ask IAN can work well with a number of different AI providers – so clients can choose their favourites. In terms of a recommendation, we have found that Claude Sonnet 4 from Anthropic is good for both speed and quality of summarisation.
We are also looking to add a default local LLM, to make it easier for people to use it out of the box, but so far, the speed is not quite good enough compared to using the online providers. We will keep experimenting and will add that into a future release if we can get something working well enough.
As with all AI systems, very occasionally IAN may not quite summarise things in exactly the expected way, so it will remain important to include a checking process for any output that will be put in front of clients.
To start using the beta version of Ask IAN, download v7.1 of AskiaAnalyse here and then read through the steps in this related Askia KB article to link it to Anthropic or, through LiteLLM (https://www.litellm.ai), to your AI provider of choice. The Ask IAN user guide is also available from the Askia Help Centre.
Do let us know how you get on!
