FlowRunner
PricingContact
Theme
Start Free

Maybe Your AI

AI

Drive the Your AI assistant inside the Maybe social marketing platform, running conversations through assistants. Agents handle social marketing conversations grounded in the platform's own data.

2 triggers 22 actions API key available
On New Conversation fires when a thread appears on the Your AI account
Agent reads the opening message and isolates the product claim being asked about
Agent calls Get Assistant Memory and finds no entry that covers the claim
Agent gathers the answer from the system of record and drafts a memory entry in the wording the assistants would use
Agent posts the draft entry next to the conversation that prompted it
The brand owner approves before Add Assistant Memory changes what every assistant says from then on

What This Integration Enables

Your AI is not a raw model behind a key. It is the assistant layer inside the Maybe social marketing platform, and the reason to reach it from a workflow is that it answers from that platform's own context rather than from general training data. Conversations can be pinned to one named assistant, or left unpinned so Your AI routes them itself. Assistants are configured with a stated purpose and a creativity setting. Uploaded file IDs can be attached to a message so an answer comes from those documents. Generate Task turns a plain-language instruction into a drafted title and body, and Create Task saves it, which makes the platform a place where AI-drafted briefs and reusable prompts accumulate rather than scatter.

The surface worth understanding is the shared memory. Get Assistant Memory returns the free-form context every assistant on the account draws on, Add Assistant Memory appends to it, and Delete Assistant Memory retires an entry. That pool is not per assistant and it is not versioned. Anything written to it becomes part of how the whole workspace answers, immediately. Used well, that is the mechanism that keeps brand voice and product facts current without anyone maintaining a document. Used carelessly by an automated flow, it is how one stale price becomes every assistant's answer in every conversation. Those are the automation exceptions worth catching, and it is why human-in-the-loop sits on the write rather than on the read.

Without FlowRunner

Assistants improvising A product question with no stored answer gets a plausible one instead of a correct one
Brand facts in people's heads The current positioning lives in a deck, and the assistant has never seen it
Corrections after the fact A wrong answer is noticed in a conversation that already happened

With FlowRunner

Answers from stored context The assistants draw on shared memory that someone put there deliberately
Memory kept current by workflow New product facts reach the assistants as part of the process that created them
Additions reviewed first A fact enters shared memory only after the person who owns it has read the wording

Use Case Scenarios

Asking the assistant a question from inside a workflow

A campaign flow needs the approved way to describe a product tier. Rather than embedding the wording, the agent calls Create Conversation, sends the question with Send Message, and uses the reply downstream. Leaving the assistant unspecified lets Your AI route the question to whichever assistant fits; naming one pins the conversation for its whole life, which is what a flow wants when consistency matters more than coverage. The answer feeds the post that goes out through LinkedIn.

Turning a brief into a saved, filed task

A launch is approved and someone writes one sentence describing it. The agent calls Generate Task with that instruction and gets back a drafted title and body, then Create Task to save it under the right category, marked shared so the workspace can see it. Over a quarter this builds a filed library of briefs and reusable prompts rather than a folder of documents nobody opens. The task list mirrors into Asana for the people who work outside the platform.

Answering from documents rather than from memory

A partner sends an updated rate card. Rather than teaching the assistants the numbers, the flow attaches the uploaded file IDs to a Send Message call, so the assistant answers that specific question from that specific document. Nothing enters shared memory, the answer is scoped to the conversation, and when the rate card changes next quarter there is nothing to retire. This is the right pattern whenever a fact has an expiry date.

Human-in-Loop Highlight

Add Assistant Memory is the operation that deserves a person, and the reason is its blast radius rather than its size. Shared memory is a single pool that every assistant on the account draws on, with no version history and no per-assistant scoping. A one-line append made by an automated flow changes how the workspace answers in every future conversation, including the customer-facing ones, and the only correction available is Delete Assistant Memory after the wrong answers have already been given to people who believed them. So the agent prepares the write and stops: "Proposed addition to shared assistant memory: 'The Growth tier includes unlimited seats.' Triggered by a conversation this morning where no stored answer existed. Every assistant will use this wording in customer conversations from the moment it lands. Source is the pricing page as of today. Approve, reword, or answer this one conversation without teaching it?" The brand owner decides whether a fact is durable enough to live in memory or belongs in one reply. Two other operations sit behind the same gate for a different reason: Delete Chat User and Delete Conversation are permanent, and both are typically run to honour an erasure request, which is precisely the kind of request that should be executed deliberately and recorded.

Agent processes routinely
Detects exception requiring judgment
Clear match Continues automatically
Ambiguous Routes to human via preferred channel
Human decides
Agent resumes with decision

Agent Capabilities

22 actions

Chat

5
  • Create Conversation Opens a conversation and returns the conversation ID that every subsequent message needs. Leaving the assistant empty lets Your AI route the conversation itself, while naming an assistant pins the conversation to that one for its whole life. Used at the start of any workflow that will exchange more than a single message.
  • Send Message Sends a prompt into an existing conversation and returns the assistant's reply. Uploaded file IDs can be attached so the assistant answers from those documents, and naming an assistant forces that one to answer instead of routing. Used both to ask a question from a flow and to ground an answer in a specific document.
  • List Conversations Returns the conversations on the account with their name, owning username and timestamps, paginated, or narrowed to a single thread by ID. Used to reconcile threads a workflow created.
  • Delete Conversation Permanently removes a conversation and its history. Used to clean up threads created by automated workflows or to honour a deletion request, and it cannot be undone.
  • Add Favourite Message Marks a message inside a conversation as a favourite for a given user, so a useful answer can be recalled later from the Your AI interface. Requires the chat, user and task identifiers together.

Knowledge

4
  • Get Assistant Memory Returns the shared memory the assistants draw on, which is the free-form context Your AI holds about your brand and business. Used to check what is already known before proposing anything new, and as the read step in front of every write.
  • Add Assistant Memory Appends content to the shared assistant memory so every assistant answers with that context in future conversations. Used to teach brand voice, product facts or campaign details from a workflow, and gated because the change reaches every conversation immediately.
  • Delete Assistant Memory Removes a piece of content from the shared assistant memory so the assistants stop drawing on it. Used to retire an outdated brand or product fact, which is the only correction available once a wrong one has landed.
  • List Assistants Returns the assistants configured on the account with their ID, name, stated purpose and creativity setting. Used to pick the assistant a conversation should be pinned to instead of relying on routing.

Tasks

8
  • Generate Task Asks an assistant to draft a task from a plain-language brief and returns the generated title and content. Used to turn a one-line instruction into structured content that Create Task can then save.
  • Create Task Saves a task with its title and body, optionally filed under a category, and marked personal or shared with the workspace. Used to store AI-drafted content, or to save a prompt template for reuse.
  • List Tasks Returns saved tasks with their title, body and timestamps, paginated, narrowable to one user or one task. Used to pull the workspace's task library into a flow.
  • Update Task Changes the title, body or category of a saved task, modifying only the fields supplied. Used to refresh a task's content without changing how it is filed.
  • Delete Task Permanently removes a saved task from the workspace. This cannot be undone.
  • Create Task Category Creates a category that tasks can be filed under, so a growing library of saved briefs and prompts stays organised. Returns the new category with its identifier.
  • List Task Categories Returns the task categories on the workspace with their identifiers and timestamps, paginated. Used to discover the filing structure before creating or moving tasks.
  • Delete Task Category Permanently removes a task category from the workspace. Tasks previously filed under it are not deleted, but the category itself cannot be recovered.

Account

5
  • Get Account Returns the account behind the API key, including the holder's name and email, business name, country, phone number and remaining balance. Used to check the balance before a workflow starts a batch of calls.
  • Update Account Updates the name and email address on the account. All fields are required by the API, so the current values must be read first and resent for anything not being changed.
  • List Chat Users Returns the chat end users registered on the account with their username, email and timestamps, paginated. Used to reconcile who has been talking to Your AI.
  • Get Chat User Returns a single chat end user by ID with their username, email and timestamps. Used to resolve the identity behind a conversation before syncing it elsewhere.
  • Delete Chat User Permanently removes a chat end user from the account. Used to honour an erasure request, and it cannot be undone.

Triggers

2 triggers

Event Triggers

2
  • On New Conversation Fires when a new conversation appears on the Your AI account, emitting the conversation record with its name, owning username and timestamps. Your AI exposes no webhook mechanism, so this trigger polls the conversation listing endpoint on a configurable interval with a minimum of 30 seconds. The first poll establishes a baseline without firing, so existing conversations are not replayed, and new ones are emitted oldest first. Used to start a workflow the moment someone opens a thread.
  • On New Task Fires when a task is saved on the Your AI workspace, emitting the task record with its title, body and timestamps. This trigger also polls, on the same configurable interval, and the first poll establishes a baseline without firing. Used to pick up an AI-drafted brief the moment it is saved and carry it into the systems the rest of the team works in.

Frequently Asked Questions

What can FlowRunner do with Maybe Your AI?

FlowRunner agents can run Create Conversation, Send Message, and List Conversations in Maybe Your AI, plus 19 more actions.

Does connecting Maybe Your AI to FlowRunner require OAuth?

No. Maybe Your AI connects to FlowRunner with an API key, no OAuth flow required.

Can Maybe Your AI trigger a FlowRunner workflow automatically?

Yes. Maybe Your AI supports 2 triggers that can start a FlowRunner workflow automatically.

Start building with Maybe Your AI

$100 in credits. No card required. Connect in minutes.