Our autonomous research agent enters public beta as Pulsar STORY
Our autonomous research agent is now open to everyone in public beta, and it has a new name: Pulsar STORY.
STORY is not another AI copilot: it's an analyst that never clocks out, operating directly on the data lake, pushing finished intelligence to you before you ask.
It has a job, not a chat window. You describe what you want to understand and confirm the plan. STORY runs the research on demand or on a schedule, and sends the finished report to the people you choose.
What five months of private beta changed
During a five-month private beta, clients used STORY as a daily briefing service, an incident monitor, a source of reports ready to share with senior stakeholders, and much more.
“Pulsar’s new AI tool, STORY, has significantly reduced the time spent analysing large volumes of online conversation, helping our team identify emerging narratives faster and turning data into actionable insight,” said Luke Dolan, Insight and Evaluation Manager at the UK’s Department for Transport, who was part of the private beta.
Those five months, and the Pulsar customers who pushed STORY hardest, strengthened its harness: the code wrapped around the model that keeps the books. Most agent failures aren’t reasoning failures, they’re bookkeeping: date math, recipient lists, filters, attribution, permissions. In STORY, none of that is left to the model, so its judgment goes into what matters, and why.
Improvements live today include:
- Cloud management: Scheduled briefings and tasks run on Pulsar’s servers, not in the user’s session, so a 7am brief lands at 7am even with the laptop closed. A failed step retries without restarting the run.
- Deep context: STORY doesn’t lose the plot. Hours and thousands of posts into an analysis, it still holds the brief, its sources and its early findings, so page twenty still matches page two.
- Accuracy: Quotes are copied, not generated by the model. STORY works on the real raw documents, not embeddings or summaries. Every quote is the exact text of a post, attributed to a real author. No source data, no reference or quote.
- Precision: Dates are computed, not interpreted. Each schedule stores its reporting window as a rule, such as “the 24 hours ending 07:30, London time”, and works out the exact dates at run time, across time zones and clock changes. The model never does calendar math.
- Integrity: the plan is approved once. Goal, searches and timeframe are confirmed before anything runs, then carried into every turn and every scheduled run. The fortieth turn still answers the question from the first.
- Governance: STORY sees only what the user sees. It runs with each user’s Pulsar permissions, including private searches and shared folders, and nothing more.
- Thoroughness: Instructions aren’t suggestions. Recipients, country, source and hour-by-hour filters apply exactly as set on every run. The agent never decides who receives a briefing.
What to use it for
Public beta is open now to Pulsar customers and new teams alike. Use it for brand health scans, crisis monitoring briefings, campaign measurement reports and more:
- Brand health scans. A scheduled read on how your brand is discussed, with the change against the previous period.
- Crisis monitoring briefings. A daily watch on an emerging issue: who started it, why it's spreading and what it means for you.
- Campaign measurement reports. How a campaign landed, with the evidence attached, ready to share with senior stakeholders.
Learn more and try it out
If you’re a Pulsar customer and you’d like to learn how STORY can help you scale your intelligence function, just reach out to your Account Manager. If you’re not a Pulsar client, yet, just fill in the form below to learn more and get a tour of how STORY works.
