Autonomous agents
Hand off the whole task.
Give Pascal a task and its agents run it end to end: they plan the work, search your IC memos, email, Teams, expert calls and broker research alongside public sources, check what they find, and deliver a finished, cited document.
Pascal Agents / Event Brief · Sea
Research in progress
- Searching IC Memos0:02
- Searching Emails0:06
- Reviewing Teams0:09
40%
Time saved
99.9%+
Accuracy
50M+
Pages indexed
25+
Workflows
How it works
The work an analyst would do, done for you.
01
Task
Pick a workflow and a company, or describe the job.
02
Plan
The agent decides what to search and in what order.
03
Search
Internal research and public sources, together.
04
Review
It cross-checks claims across sources.
05
Deliver
A structured, cited document, ready to use.
Searches where your team works
Your firm's knowledge, alongside the market's.
The most useful context rarely sits in a filing. It is in an IC memo from last quarter, a pinned Teams message, an email after an IR call, an expert call transcript.
Thinks like an analyst
It asks the questions you would.
Before an earnings event, an analyst checks ratings, targets, scenarios and what management has said. The agent does the same, and you can open any step to see exactly what it asked.
Run log
Event Brief · Sea
Question
Rating, target, scenarios, Shopee GMV and take rate, management commentary…
Decision-ready output
Scenarios and triggers, ready for the IC.
The agent doesn't stop at what happened. It frames what it means for the position: scenarios with probabilities, the triggers to add or exit, and a clear takeaway, every figure cited.
| Scenario | EBITDA | TP |
|---|---|---|
| Base | $3.1–3.3bn | $95–110 |
| Bull | $3.3bn+ | $110–140 |
Always on
Agents that keep working when you don't.
Set agents to run on a schedule or when something changes, and the work is waiting for you: catalysts for the week, a brief after every result, an alert when your thesis is tested.
Scheduled
4 on
- Watchlist alertsTomorrow 07:00
- Thesis monitorsWatching
Why it matters An assistant answers. An agent finishes.
| Chat assistant | Pascal autonomous agent | |
|---|---|---|
| What you give it | One question at a time | The whole task |
| Where it looks | Whatever you paste in | Your memos, email, Teams, calls and filings |
| What you get back | An answer to edit | A finished, cited document |
| When it works | Only when you ask | On a schedule, or when things change |
Frequently asked questions
An agent you give a whole task to. It plans, searches internal and public sources, checks what it finds, and delivers a finished, cited document.