Skip to content
English
Go to merchantspring.io
  • There are no suggestions because the search field is empty.

AI Conversational Agent

Learn how the MerchantSpring AI Conversational Agent works, what questions you can ask, how AI credits are used, how to download and share reports, and what features are currently supported.

1. Overview

What is the AI Conversational Agent?

The AI Conversational Agent is a natural language chatbot within MerchantSpring that allows users to ask questions about their data and receive:

  • A written summary and analysis

  • A visualisation (artifact)

  • A downloadable PDF

  • A shareable public board link

It sits within the new AI Toolbar (the grey strip in the UI), which is designed to expand over time with additional AI tools.

2. UI & Navigation

What happened to the yellow Commentary widget?

The yellow Commentary box has not been removed — it has been moved into the AI Toolbar and restyled to align with the Conversational Agent.

Clients may notice the change in placement and appearance.

Can users go full screen?

Yes.
The Conversational Agent can be expanded to full screen, which is useful when reviewing dense outputs.

Can users rename conversations?

Yes.
Users can rename chats and access their personal chat history.

3. What the Agent Can Do

What kind of questions can clients ask?

Clients can ask natural language questions about virtually any data available within MerchantSpring, including:

  • Top performing campaigns

  • Channel-specific analysis

  • Portfolio or tag-based analysis

  • Account-wide performance queries

If no constraints are provided, the agent will analyze across all managed channels.

Does the agent always create a visual?

Yes.

Each output includes:

  • A written summary

  • A visualisation artifact (chart, table, etc.)

Because AI includes inherent randomness, outputs may vary slightly between runs unless tightly controlled through prompting.

Is there a limitation on the number of columns or width (like Report Builder)?

There is no explicit column limitation like the traditional Report Builder.

However:

  • The visualisation engine has formatting parameters.

  • Very complex tables (e.g., large ASIN-level datasets with many columns) may require prompt refinement.

  • Some complex outputs may not render perfectly on the first attempt.

4. Downloads & Sharing

How does PDF download work?

Reports download directly to the user’s device. They are not emailed.

What is a Public Link?

Users can toggle a report as “shareable” and generate a public link.

The public board:

  • Applies client logo

  • Applies client theme colors

  • Matches chart colors to configured branding

  • Respects custom domain setup (if configured)

This makes it presentation-ready for client sharing.

5. AI Credits & Commercial Model

Is the AI Conversational Agent available on all plans?

Yes.

The tool is available across all plans. The difference between plans is the number of AI credits included.

What are AI credits?

AI credits are MerchantSpring’s internal AI currency.

  • They do not directly map to AI tokens.

  • A typical query uses approximately 300–400 credits.

  • If purchased, this equates to roughly $0.80–$0.90 per report.

When do credits reset?

Credits reset at the start of each new billing cycle (not calendar month).

Each billing cycle:

  • Restores the full monthly credit allocation

  • Includes base plan credits + any additional purchased credits

Can clients view their credit usage?

Yes.

Users can view:

  • Current AI balance

  • Transaction history

  • Monthly reset line item

Does more back-and-forth use more credits?

Yes.

  • Each interaction consumes credits.
    More iterations = more credits used.
  • Well-structured prompts reduce usage.

Can clients purchase more credits?

Yes.

Clients can:

  • Add additional monthly credit tiers

  • Stack credits on top of their existing plan allocation

Example:

  • Base plan: 5,000 credits/month

  • Add-on: 20,000 credits/month

  • New total: 25,000 credits/month

6. Prompting & Output Consistency

Is there a prompt library?

A prompt library has been designed but is not yet built.

Currently:

  • Users can access their personal chat history.

  • There is no formal reusable prompt library.

Will the same prompt always generate identical output?

Not necessarily.

AI includes some randomness in responses.

However:

  • System configuration can reduce randomness (temperature control).

  • More precise prompts increase consistency.

100% identical replication cannot be guaranteed.

7. Privacy & Access Controls

Who can see AI conversations?

Chat visibility is restricted to the specific user only.

  • Users cannot see other users’ chats.

  • Chats are not visible at the organization level.

  • There is no cross-user visibility.

Do external seats have access to the AI Conversational Agent?

No.

External seats do not have access to the AI Conversational Agent.

Access is restricted to internal users only.

8. Limitations

Can the agent read links (URLs)?

No.

The agent:

  • Cannot fetch or read external URLs.

  • Cannot ingest content directly from links.

Can users upload Excel files, Google Sheets, images, or documents?

Not currently.

Can the agent reference an existing MerchantSpring report by link?

Not currently.

9. Recurring Reporting

Can users schedule recurring AI-generated reports?

No.

Recurring AI-generated conversational reports are not currently on the roadmap.

We are actively gathering customer feedback to determine how AI reporting should evolve before committing to automation.

10. Future Direction (Conceptual / Exploratory)

Are there plans for specialized AI agents (e.g., profitability-only agent)?

Currently, there is one general-purpose agent.

Future possibilities discussed include:

  • Domain-focused agents (profitability, operations, etc.)

  • Orchestrator model delegating tasks to specialized sub-agents

No confirmed timeline yet.