Introducing Saga, the research agent that never clocks out

Every vendor in social and media intelligence is now shipping an AI copilot. That’s usually a chat window bolted onto a dashboard: you ask it a question, it narrates something on the screen, then forgets the moment you close the tab. When nobody's asking, it produces nothing.

But researchers, marketers and communication professionals don't need more shallow analysis of what's already on the screen. They need help with what's not charted yet.

Today we're introducing Saga: an autonomous research agent for social, media, and audience intelligence that will change how insights, marketing, comms and agency teams work. It’s like Openclaw for social and media intelligence.

Think of Saga as an analyst, working with your methodology directly on the data lake. You brief it once, the way you'd brief a researcher, and it stays on the job: watching your data, spotting patterns and anomalies, escalating what matters, and delivering finished work on its own clock. Where an AI copilot has a chat window, Saga has a job, and it never clocks out.

From brand health scans to cultural deep dives, to harmful narrative checks and category briefs, Saga carries out full jobs and ships finished work according to your brief and methodology, allowing you and your team to focus on interpretation, decision-making and strategy.

Brief-driven, not prompt-driven

A copilot waits. You open a window, ask a question, read the answer, and close the tab. The context dies with the session. Saga runs the other way. You brief it once and it works on its own clock, in the background, indefinitely: watching the data, escalating what matters, pushing finished intelligence work to you before you think to ask. It shifts social and media intelligence from pull to push. Saga is not a faster way to ask questions. It’s a way to stop asking.

Prove it: brief Saga and a copilot the same way, then walk away for thirty days. Saga comes back with a stack of analyses tied to real movements in the data. The copilot comes back with nothing, because nobody re-opened it.

On the data lake, not the dashboard

Every copilot in the category is a chat layer sitting on top of pre-aggregated analytics. It can only hand you what the reporting product already exposes. Saga on the other hand works directly on the raw corpus of data: it runs novel clustering, custom embeddings, and the kind of multi-step statistical work no dashboard can pre-define. And it runs on fifteen years of permissioned, audience-grade data across the widest set of data sources in the industry, not an open-web scrape.

Prove it: ask Saga and a copilot the same novel-clustering question. The copilot returns a standard demographic breakdown, or a summary of a chart someone already built. Saga surfaces an affinity group that exists in no pre-built taxonomy: a "slow beauty" cluster, say, up 23% and spread across more than twelve thousand audiences. 

Finished work, not summaries

A copilot narrates the dashboard it’s been handed. Saga produces the deliverable itself: the audience affinity report, the creator landscape map, the earned media report. It runs them across every market on your team’s methodology, so outputs are comparable, without anyone having to spend time on reconciliation. 

Prove it: the Monday brief that lands in the inbox before standup, carrying last week's signal and not last year's taxonomy. One prompt covers four markets at once.

Methodology that compounds

A copilot is stateless. Every session starts from zero, and your team's method lives in one senior analyst's head until the day they leave with it. Saga captures that method as prompt libraries: named, versioned, and owned by the people who built them. They survive attrition and scale across every new hire.

Prove it: hand a new joiner the captured house method and watch them produce senior-grade work in week one, with the library named and its author credited on the output.

So, what kinds of output can Saga produce? 

Saga is responsive to your brief: any task you’re carrying out on your social intelligence tool.

Some examples:  

• Weekly briefing: Brief Saga once early on a Monday. Every Monday after that, the brief lands in your inbox, before standup.
• Pitch audit: Three categories, four markets, one weekend. Saga audits the audience and the conversation while you write the deck.
• Early warning scan: Saga doesn't surface everything. It escalates what crosses your threshold: an emerging cluster, a tone shift, an affinity group that isn't in any pre-built taxonomy. It catches the cluster at 200 mentions, not 20,000.
• Competitive scan: What rivals are doing that's actually working, tracked as it happens. You see the move while it's still a move, not in the quarter-end deck.
• Cultural deep dive: "what's shifting in dad culture on TikTok this quarter?" and get the tensions, the top communities and real audience quotes back.
• Multi-market analysis: One brief, four markets, one methodology. Not four separate outputs assembled by hand. Comparable because Saga runs them the same way.
• Reputation pulse: The narrative watched 24/7 across stakeholders and markets, on one methodology, so the cross-market comparison holds up.
• Crisis mode: When something breaks, Saga changes gear: hourly escalation, the narrative arc tracked, holding-statement drafts, all into the war-room Slack with source, volume, prior history and a recommended hold-line. The context holds through the weekend, so you walk into Monday ready.
• Creator check: Vet a creator before you sign — controversy history, audience authenticity, brand-safety flags — then keep watching. If a signed one drifts into trouble, you know inside the hour, not after the screenshot is already circulating.
• Earned-media valuation: One EMV methodology, run the same way across every market and auditable by finance. No more reconciling three agencies' three different numbers.

That’s just a few ideas: Saga will adapt to your team’s needs. 

Give Saga a job. Walk away.

Brief Saga once, the way you'd brief a senior analyst, and it runs indefinitely, with no re-prompting, on your methodology.  Set the threshold and it escalates when the data crosses it, not when you remember to check a dashboard. The finished work, like brand briefs, competitive scans, cultural reads, campaign reports, arrives before you think to ask.

We’re currently running a small beta of Saga with a few dozen Pulsar users: if you’d like to try it out yourself, fill out the form below.