Beta: Leo

In August, 2026, we're launching Leo in Beta. Leo is Paramark's built-in AI chat assistant, and it can answer questions using your Paramark reports plus recent Slack and meeting context with your Growth Advisor. This article describes everything you need to know about it.

Beta access and pricing: If Paramark has enrolled your workspace in the Leo beta, you can start using Leo right away in Slack or from the Ask Leo button in the dashboard. There is no additional cost during the beta. Once Leo becomes generally available, pricing will move to usage-based consumption, and existing customers will receive a generous amount of credits for the first year so you do not need to worry about hitting limits while getting started.

What happens when Leo is enabled for my workspace?

In some cases, Paramark will proactively introduce Leo in your shared Slack channel when beta access is turned on for your team. In that rollout thread, a Paramark teammate may first announce that your workspace has been enrolled in the beta and explain where Leo can be used, and Leo may then reply in-thread to introduce itself and suggest a few good starter questions. Those starter prompts may also be tailored to the kinds of MMM reports, experiment readouts, data-health questions, and recent Slack or meeting context your team already has in Paramark. Once enabled, you can immediately start using Leo in either of these ways:

If your company has access to multiple Paramark workspaces: you may be prompted to choose a workspace when signing in. You can switch workspaces later by clicking your name in the bottom-left corner of dashboard.paramark.com.

• In Slack: @-mention Leo in the shared channel or reply in the thread where Leo was introduced.

• In the dashboard: Click Ask Leo at dashboard.paramark.com to open the chat panel.

Good first questions after enablement include things like “Explain our latest MMM findings,” “How is our data health looking?”, “What changed in our latest refresh?”, “What did we decide about a recent experiment?”, or “Which channels have the best ROAS or CPA right now?” If you’ve just received a new experiment analysis or readout, you can also ask Leo to summarize the biggest takeaways, explain what the result means in plain English, or recap any next steps already discussed with Paramark. Other useful prompts include “What were the key findings in our latest MMM refresh?”, “How’s the CTV experiment tracking so far?”, “Which channels are driving incrementality lately?”, and “Is our data up to date / are there any health issues?” If your team works closely with MMM and experiment readouts, good thread-starter questions also include “What did our latest OTT midpoint readout show?”, “How’s our ECP MMM looking — which channels are most efficient?”, and “Recap what we decided about the Meta holdout.”

If Leo introduces itself in your shared Slack channel with suggested prompts, you can simply reply in that same thread and start with one of them. Good examples include “What were the key findings in our latest MMM refresh?”, “How's the CTV experiment tracking so far?”, “Which channels are driving incrementality lately?”, and “Is our data up to date / any health issues?” For an experiment that is still in flight, Leo can help recap the setup, timeline, and recent context, but you should still wait for Paramark's formal readout before using any early signal to make a decision.

What is Leo and why we built it?

Leo is Paramark's built-in AI chat assistant. Ask it a question about your marketing mix and get a real-time answer in plain language. You can reach it in two places:

• In the dashboard: Click the "Ask Leo" button in the bottom-right corner of any page at dashboard.paramark.com and it opens a side panel. Just type your question into the chat box.

• In Slack: @-mention Leo in your shared Paramark channel, or send it a direct message. Leo replies in a thread.

Use Leo as a question-and-answer tool. Ask it anything about your reports or your data.

Here's how it works

AI chatbots are as good as the data they are provided and the model that powers them. For Leo:

Data: Leo sees everything in your Paramark environment:

• Reports — everything in your MMM and experiment reports.

• Paramark Live — all of the data you see on the Paramark live dashboard.

• Conversations — recent Slack history and call transcripts from your Paramark meetings.

• Help center — articles from help.paramark.com.

Model

Leo runs inside Paramark's secure cloud environment. This means that your data is secure and is not used to train any LLMs. Your Leo chats have access to your data only, ensuring a clean boundary between Paramark customers.

Why we built it

The most common question we hear between model refreshes is some version of "what does this mean?," and "ask your growth advisor" was never a fast enough answer when you're staring at a chart before a leadership meeting. Leo puts an always-available expert in the same place as your data, so you can get answers when you need them.

How to use Leo

What kinds of questions can Leo help with?

• Leo is especially useful for questions about your MMM reports and findings, experiment designs and results, data health and source coverage, and recent context from shared Slack threads or calls with your Growth Advisor. It can also summarize specific report refreshes or experiment analyses you already have in Paramark, and it will often point you back to the dashboard when the best next step is to inspect the underlying report or analysis directly.

Can Leo give an early read on an experiment that is still in flight?

• Sometimes Leo can summarize the experiment design, the latest context from your shared Slack channel, or what Paramark has said so far about the test. But you should not use Leo for a definitive early read on an active experiment before the test has reached its planned duration and enough post-treatment data have arrived. Early peeks are usually inconclusive and can be misleading, so the right move is generally to let the experiment run to completion and wait for the formal readout from Paramark. If you want background, Leo can still help you review the setup, timeline, and any prior decisions already documented.

Can Leo answer general knowledge or off-topic questions?

Not reliably. Leo is built to answer questions about the information available in your Paramark environment, such as your MMM reports, experiment results, data-health context, shared Slack history, meeting transcripts, and help-center content. If you ask a broad general-knowledge question or something unrelated to your Paramark work, Leo may not answer well, may redirect back to Paramark-related topics, or may simply not have the right context to help.

For best results, keep your questions grounded in your Paramark reports, recent analyses, or conversations with your Growth Advisor.

Good starter questions:

• If you're not sure how to begin, try prompts like “Explain our latest MMM findings,” “What's driving Day 7 ARR lately?”, “Is our data healthy or up to date?”, “What did we decide about folding in Newsletter or Google data?”, or “Walk me through our most recent experiment design or results.”

Good first questions to ask Leo:

• "Explain our latest MMM findings."

• "What's driving Day 7 ARR lately?"

• "Is our data healthy / up to date?"

• "What did we decide about folding in Newsletter/Google data?"

• "Walk me through our most recent experiment design or results."

Can Leo answer general-knowledge questions that are unrelated to Paramark?

Usually no. Leo is designed to answer questions grounded in your Paramark context — such as MMM findings, experiment designs and results, data health, recent Slack history, meeting transcripts, and help-center content. If you ask a general trivia or off-topic question that is not connected to your Paramark data or work with Paramark, Leo may not be able to help and may give a weak or unhelpful response.

As a rule of thumb, Leo works best when your question is about your reports, your data, your recent conversations with Paramark, or how to use Paramark. For unrelated questions, use a more general-purpose assistant instead.

For example, a question like “What did we decide about our latest MMM refresh?” is a good fit for Leo. A question like “What is the deal with the Mona Lisa?” is not.

These examples are a good starting point because Leo can pull from your MMM reports, experiment results, recent Slack history, and meeting transcripts with your Growth Advisor.

In practice, that means Leo is especially useful for quick questions about MMM reports and experiment insights. You can ask things like “What drove our latest MMM refresh?”, “What were the main takeaways from our most recent experiment?”, or “Summarize what Paramark has said recently about this result.”

Can Leo help me understand a newly delivered experiment analysis?

Yes. Once Paramark has delivered an experiment analysis, Leo is a good way to get a fast recap of what the readout says in plain English. For example, you can ask Leo to summarize the main takeaways, explain whether the result showed lift, no lift, or an inconclusive outcome, and restate any recommendations that already appear in the analysis or recent shared context.

Leo is most helpful here as an explanation layer over work that already exists. It can help you understand a delivered analysis faster, but it does not replace the formal Paramark readout or create a new analysis on its own.

Can Leo build out the analysis for me?

Not yet. Leo can help explain existing reports, summarize prior experiment results, pull together context from recent Slack threads or Growth Advisor calls, and point you to the relevant source material. But it does not independently run a new experiment analysis, produce a formal test readout before Paramark has completed one, or replace the work your Growth Advisor does to design, interpret, and finalize analyses.

If an experiment is still in flight, Leo may be able to restate the design or recap what has already been said, but you should wait for the planned test duration and the formal Paramark readout before using any interim observations to make a decision.

Leo is not a full-service agent. It cannot start a model refresh, change campaign categorizations, or take other actions. Continue contacting your growth advisor about these.

Can Leo tell me if I have any outstanding deliverables or next steps?

Sometimes. Leo can summarize action items that appear in recent shared Slack history or meeting transcripts, so it can be a helpful first check if you want a quick reminder of open follow-ups. For example, you can ask questions like “Do I have any outstanding deliverables?” or “What are the next steps from our recent conversations?”

That said, Leo is not a task manager or system of record, and it may miss items that were tracked elsewhere, agreed live but never written down, or not captured in the recent context it can access. For anything important, confirm with your Growth Advisor or in your shared channel.

Best practices & tips to get value from Leo

A few tips for getting the most out of Leo:

• Start with a concrete scope: Mention the report, experiment, channel, metric, or date range you want Leo to focus on. For example: “Summarize our latest MMM refresh” or “What changed in paid social efficiency this month?”

• Ask Leo to explain, compare, or summarize: Leo is strongest when it is helping you understand existing Paramark context. Good verbs include “explain,” “summarize,” “compare,” “recap,” and “walk me through.”

• Use follow-up questions: After Leo gives an answer, ask it to go deeper. For example: “What evidence are you using?”, “Can you break that down by channel?”, or “What changed vs. the prior refresh?”

• Ask for the source: Leo can give you links to reports and cite its sources; use this to verify its answers.

• Keep it grounded in Paramark context: Leo works best on your MMM reports, experiment analyses, data health, shared Slack threads, meeting notes, and help-center content. It is not a general-purpose assistant.

• Normal AI rules apply: Like any AI chatbot, Leo can be confidently wrong. Asking it to cite its sources and show its work can help you mitigate this risk.

Prompt ideas by use case

• MMM readouts: “Explain our latest MMM findings in plain English.” “Which channels look most efficient right now?” “What changed in our latest refresh compared with the prior one?”

• Experimentation: “Recap our most recent experiment design.” “What were the main takeaways from the latest experiment readout?” “What has Paramark said so far about our Meta holdout?”

• Data health: “Is our data up to date?” “Are there any data-quality issues I should know about?” “What sources have changed recently?”

• Recent context and decisions: “What did we decide in our recent conversations about Newsletter data?” “Do I have any outstanding deliverables from recent meetings?” “Summarize the latest discussion with our Growth Advisor.”

Recommended prompt patterns

If you want more mileage from Leo, try prompt patterns like these and then follow up from there:

• Summarize a result: “Summarize the biggest takeaways from our latest MMM refresh in plain English.”

• Explain change over time: “What changed since the last refresh, and which channels moved the most?”

• Compare channels: “Which channels look most efficient right now, and what metrics are you using to judge that?”

• Pressure-test the answer: “What sources are you using for that answer? Link me to the underlying report or context.”

• Recap decisions: “What did we decide in recent Slack threads or meetings about this experiment or channel?”

• Find next steps: “Are there any open questions, follow-ups, or deliverables mentioned in our recent Paramark conversations?”

A good rule of thumb is to start broad, then narrow: ask Leo for a summary first, then ask a second question about what changed, why it changed, and where the answer came from.

How can I try Leo?

If you'd like to try Leo, reach out to your Growth Advisor or post in your shared Paramark Slack channel. We can confirm whether Leo beta access is available for your workspace and help you get started.

In some cases, Paramark may proactively enroll your team in the Leo beta and introduce Leo directly in your shared Slack channel. If that happens, you can begin using Leo right away in Slack and in the Paramark dashboard as described above.

There is no additional cost to use Leo during the beta. Once Leo becomes generally available, pricing will move to a usage-based model. Existing customers will receive a generous amount of credits during the first year after general availability so you can get started without worrying about limits.

If you're already in a shared Slack channel with Paramark, you can simply ask about getting access to Leo there. Once enabled, you'll be able to use Leo in Slack and in the Paramark dashboard as described above.

If you're not sure whether you already have access, try looking for the Ask Leo button in the dashboard or asking in your shared channel. If access hasn't been turned on yet, our team can help coordinate next steps.

How can you provide feedback on Leo?

Leo is in Beta, which means your feedback genuinely shapes where it goes next. We're actively iterating on the data Leo can access, the questions it handles well, and the insights it surfaces, and we want to hear from you.

The fastest ways to reach us:

• Message your Growth Advisor directly, or drop a note in your shared Slack channel.

• Tell us what's working, what's confusing, and what's missing — especially the moments where Leo helped you move faster, or left you wanting more.

A few prompts that are especially useful to us right now: Is Leo's data accurate when you test it? What questions do you wish Leo could answer that it currently can't? Are there reports or data sources you expected Leo to know about that it doesn't?

Thanks for helping us build it. This is just the first version, and the best ideas for where it goes next will come from the people using it every day.