Turn a one-line input into a full research report or presentation: sourced, validated, and built on current data, not a model’s memory.

Visit ZARC

The summary.

Most AI research tools answer from what a model already learned, which is exactly where stale numbers and unsourced claims come from. ZARC works the other way round: it goes and gets the data live, keeps a source and a validation check tied to each fact, and reconciles figures that disagree before they reach the page. You give it a line; it gives back a sourced report or deck you can stand behind.

1line
The input it needs
Sourced+ validated
On every fact in the report
Main technologies used
Custom scraping engineAWS
Engagement modelOur own product
We built ZARC because we needed research we could trust ourselves: sourced, current, and checkable, before we’d put our name on it.
Placeholder: own product; founder attribution to come

About ZARC

ZARC (Zapulse Automated Research Console) is Zapulse’s own automated secondary-research platform. It turns a one-line brief into a sourced research report or presentation, built on live-scraped data with a citation and validation trail behind every claim. It runs at zarc.zapulse.com.

The challenge.

The information that actually matters is live: current market figures, named sources, competing data points that need reconciling. A pre-trained model doesn’t go and check. It recalls, and recall is where fabricated stats and contradictions creep in.

ZARC had to pull research from live sources on demand, keep a trail back to where each fact came from, and flag what doesn’t agree, so an analyst gets a draft they can trust, not one they have to fact-check line by line.

Secondary research usually runs on whatever a general model already learned, which means numbers that are stale, unsourced, or quietly wrong. For a research firm, that’s the one thing you can’t ship.

The solution.

We built a custom scraping engine that reads live sources on demand, extracts the fields a report needs, and keeps a citation and validation trail attached to every claim, so each report is built on current, checkable data rather than a pre-trained model’s memory. From a one-line input, ZARC assembles a full research report or a business presentation.

All technologies used.

Custom scraping engineAWS

How we
built it.

The same four phases behind every build we ship: scoped in writing, delivered in milestones, and maintained by the team that built it.

  1. 1.Talk through the problem

    Discovery
    • Your business, not a feature list
    • Who actually uses it day to day
    • Timeline and budget, up front

  2. 2.Scope, estimate and plan

    Written scope
    • What gets built, and in what order
    • Which stack it runs on, and the cost
    • Nothing starts until you approve it

  3. 3.Build and ship in milestones

    Milestones
    • Working software at every milestone
    • No single reveal at the end
    • Feedback folded in as it goes

  4. 4.Live, and kept that way

    On AWS
    • Deployed and running on AWS
    • We stay on after go-live
    • Maintained by the team that built it

The outcome.