Page Comparison

What is Phocas AI

Phocas AI gives you quick answers and insights by simply asking a question in your natural language. This form of analysis is particularly helpful for new Phocas users and occasional users, as they are not yet familiar with the functionality of Phocas.

Currently, Phocas AI can only return results that can be found using the Analytics user interface. In other words, it can’t do things you can’t do in Phocas. However, it can make use of Custom mode on your behalf - a feature mainly used by more advanced users.

Period types are important. Uses first default period type (Day, Month or Year) in list when returning results, e.g. Year or Month. Sometimes one of these is not enabled in database. Supports Week and Quarter custom period types. Handles custom periods.

User database permissions and restrictions are respected, so only results that you have permission to view will be returned.

Can filter by selected entities

Use history as a baseline for similar questions. History is database-specific.

Benefit - You can ask a question in your terms and see the steps that were taken to get the results. Learn as you go.

Get started with Phocas AI

If you’ve been invited to use Phocas AI, you’ll see a welcome message on the right of your screen when you open a database in Analytics.

1. Click Let’s go to activate the AI feature.

2. Type a question into the box and click Ask. Tip: You don't have to use a question word or question mark.

3. Rate the response and ask more questions as required.

   

Get the most out of Phocas AI starts by

asking good questions

Phocas AI can’t read your mind, it takes the question you ask and uses probability to infer what you mean, the less inference required the greater chance of a correct answer. Here are some tips to think about when writing your question:

• Remember to include a date range - Compare 'What are my sales?” vs “What were my sales last quarter?”. Beside Besides the usual date types (3 days ago, last month, this year etc), we also understand any custom periods you may have defined.

• Use terms from the database where possible - we understand synonyms but they are less acurrate accurate that explicit naming the data element of interest, particularly if the term is ambiguous. e.g. If the dimension is called ‘Customer’ try not to use ‘Account’

• Think about specifying the stream - if you have 3 budget streams then it would probably be helpful to specify the one you are interested in

• You can use wildcards, either ‘*' or '%’ - eg “code like PR3986* for august August 2023 sorted by cost price descending”

• The question doesn’t need to be in English, we’ve tested a bunch of languages with great success

• If you want a pie chart or graph then ask for one. Phocas AI supports all of the standard charting options

• Don’t worry to much about spelling unless vital for the question - we know what you mean by ‘cutsomer’‘customer’, but ‘Amce’ probably wont won't resolve to ‘Acme’

Sample Questions

• Show me the best selling product over the last 12 months.

• Sales reps whose sales value has increased by less than 5% over last 2 quarters

• show me customers who brought more in 2022 than 2023

• What are the top performing sales reps this year?

• What regions are ahead of budget this quarter?

• Who sold the most products in UK last month?

• Cities with the highest sales between last Black Friday and the end of 2023. Use reporting value.

• Sales in [region] this year.

• Which branch had the most sales in August?

• top selling product for the supplier Acme by quantity last week

• Plot sales for Engines onto a column chart by month for the last 5 months

Limitations of Phocas AI

• Financial databases are not yet supported