Summary
Saysimple's Flow Builder now supports AI intent recognition. Instead of routing every customer through the same fixed set of scripted questions, a flow can read what someone actually writes, work out whether it is a service question or a sales question, and use an AI conversation layer to ask for and store the specific detail that department always needs, such as an email address. That stored variable can then optionally trigger an API call to push the information into another system. Setting this up involves more configuration than a single toggle, but the result is a customer who reaches the right person with the right context already in hand.
TL;DR
- Flows can now detect intent (for example service vs sales) directly from a customer's own words, not from a menu choice.
- An AI conversation layer asks for and stores the specific detail a department always needs, such as an email address, rather than pulling it from a single message automatically.
- That stored variable can optionally trigger an API call to push the data into another system - this is something a team configures deliberately, not something that happens by default.
- The real goal is routing: get the customer to the right department with the information it needs already collected, so nobody has to ask again and nobody ends up with the wrong team.
The problem with flows that ask everyone the same questions
Most flow builders, including earlier versions of Saysimple's, work the same way. You lay out a decision tree, write the questions in advance, and every customer answers them in the same order, whether or not each question actually applies to them. A customer chasing an invoice gets asked the same opening questions as someone reporting a broken product, until the flow finally branches somewhere further down.
That is a fine model when a business genuinely has one or two request types. It breaks down once conversations get varied, because someone has to predict every path in advance and hand-build a branch for each one. Miss a path, and the bot either loops the customer back to a menu or hands off to a human anyway, at which point the flow has not saved anyone any time.
What changed in Flow Builder
Saysimple's Flow Builder can now build flows that determine intent from the customer's own text rather than from a fixed menu choice. A flow can be set up to recognise, for example, that a message is a service question or a sales question, and branch accordingly, without the customer ever picking an option from a list.
This sits inside Saysimple NEO, the platform version where AI has a materially larger role across the product than it did before, and it is already the default for new customers. Intent recognition in Flow Builder is one concrete, practical result of that shift: it is not a separate AI add-on bolted onto the side of the product, it is the flow logic itself getting smarter about what it is looking at.
How the flow collects what a department actually needs
Detecting intent is only half the job. The real goal is collecting whatever a specific department always needs before it can handle a conversation properly, most often something like an email address, but it could just as easily be an order number, a company name, or anything else a team routinely has to ask for. That part runs on an AI conversation layer inside the flow: it asks the customer for the missing detail in a natural way, and stores the answer as a variable as the conversation moves forward.
Once that variable is stored, it can optionally trigger an API call to push the information into another system, for example to look up an order, check an account, or hand the data to a CRM, using the same API and integration layer Saysimple already offers for messaging, contacts, and connected systems. That call is not automatic. A team sets it up deliberately, for the specific variables that are actually worth passing along.
Configuring this well, deciding exactly which detail the AI conversation layer should ask for, how it gets stored, and which of those variables are worth wiring into an API call, is more technical than the intent-branching itself. It is not a single toggle to flip. Saysimple can set this part up directly for customers, rather than leaving it as something a team has to figure out on their own.
Why this matters for how flows get built
The real shift is in what a team has to design up front. A traditional flow needs every question and every branch mapped out before it goes live, which means someone has to guess every way a conversation might unfold. An intent-driven flow does not need that same exhaustive map. It reads what the customer actually said, works out what is still missing, and asks for that, rather than working through a script written for the average case.
In practice that means:
- Fewer scripted branches to build and maintain for every possible customer path.
- Fewer redundant questions, since the bot does not ask for information the customer has already given.
- Customers reach the right department with the details that team needs already collected, instead of getting bounced between agents or having to repeat themselves.
None of this replaces a team's judgement about what the flow should actually do with a service question versus a sales one. It changes how much of that judgement has to be hard-coded in advance, versus handled by the flow recognising the situation as it happens.
Getting started
Intent recognition is configured inside Flow Builder, on top of the same flow structure teams already use today. A practical way to start is to pick one flow that currently forces every customer through the same opening questions, most commonly a general contact flow that splits into service and sales further down, and rebuild the first step so the bot reads the opening message instead of presenting a menu. From there, decide the one detail each department genuinely always needs, such as an email address or order number, and have the AI conversation layer collect it before handing the conversation off. Because the configuration involves more moving parts than a simple toggle, Saysimple's team can set this up for you directly rather than leaving it as a DIY project.
Get started with Saysimple
If your flows still ask every customer the same fixed questions before finding out what they actually need, this is worth a look. Book a demo to see how AI intent recognition in Flow Builder fits into your existing setup.