BoldAgent by Syncfusion has launched

Book a demo Free trial

Competitive Intelligence Agent

Your market, summarised before your first meeting of the week.

Competitors change their prices on a Tuesday, ship a feature on a Thursday and announce a partnership on a Friday. Most teams find out weeks later, mid-deal. A BoldAgent competitive intelligence agent researches the public web on a schedule you choose, decides which changes matter, checks the maths, and leaves a finished brief waiting for your team.

Start building Read the documentation

What Monday morning looks like

A finished brief, waiting before the week starts

  1. 8:00 AM
    A new conversation appears in your workspace.

    Nobody started it; the schedule did. Inside is this week's brief, ready to read or download.

  2. Coffee
    Product marketing reads it over coffee.
  3. First call
    Sales gets the talking points before their first call.
  4. Friday
    Nobody spent their Friday afternoon assembling it.

Weekly Competitor Brief — 2 November 2026

3 high-impact changes 4 worth knowing 11 changes filtered out

  1. 01

    Competitor A raised its Team plan from $49 to $59 per seat (+20.4%) and cut its free tier from five users to three.

    For us our Team plan is now $8.50 cheaper per seat. For a 50-seat team, that's $5,100 a year. Lead with total cost in mid-market deals.
  2. 02

    Competitor B launched a native Salesforce integration.

    For us this closes a gap we've used in our battle card. Update the "Integrations" objection response.
  3. 03

    Competitor C announced a partnership with a major cloud marketplace.

    For us expect them in more procurement-led enterprise deals.

Anatomy of the brief

Same structure, every week

Every brief follows the same structure, so readers learn where to look and week-on-week comparisons are easy.

  1. 01

    Headline count

    High-impact, worth-knowing and filtered-out totals

    From The scoring step
  2. 02

    High impact

    Changes likely to affect deals or roadmap

    From Research, scored and verified
  3. 03

    Worth knowing

    Smaller changes to keep an eye on

    From Research, scored below the top tier
  4. 04

    "For us" notes

    What each change means for your product, pricing or pitch

    From Your own positioning library
  5. 05

    Price table

    Current per-seat prices and change since last week

    From Calculated in the sandbox
  6. 06

    Sources

    The pages each finding came from

    From Web Fetch and Web Search results

The format lives in a skill, so changing it once changes every future brief.

A research team, not a single prompt

Each specialist gets a narrow brief

One agent trying to read twenty pages, search the news, calculate price changes and write a brief tends to do all of it shallowly. Instead, an Intelligence Lead agent delegates through Agent Orchestration, giving each specialist a narrow brief:

Intelligence Lead Delegates through Agent Orchestration
  • To the

    Pricing Watcher

    Web Fetch
    Brief Read each competitor's pricing page listed in the watchlist. Return every plan name, price, billing period, seat limit and headline feature exactly as shown.
  • To the

    News Researcher

    Web Search
    Brief Find announcements, launches, partnerships and funding news for each competitor from the past seven days. Return the claim, the date and the source page.
  • To the

    Analyst

    Sandbox code executionFile Generationpositioning library
    Brief Compare this week's prices with last week's, score every finding, write a "for us" note for each high-impact change, and generate the brief as a downloadable report.

Because each worker is a normal agent, you can also run the Pricing Watcher on its own when someone asks, "What does Competitor D charge now?"

Signal versus noise

Only what matters reaches the brief

Most of what changes on a competitor's website doesn't matter. A Finding Significance decision model judges every finding against the same criteria, returning a category, an impact score from 1 to 4, and whether it's relevant to your product. Only findings scoring 2 or higher reach the brief. Here's how a typical week's raw findings were judged:

  • Team plan price rose from $49 to $59 Kept: impact 4 Directly affects deals in your core segment
  • New native Salesforce integration Kept: impact 4 Removes a differentiator you rely on
  • Free tier reduced from five users to three Kept: impact 3 Changes the self-serve comparison
  • New case study with a retail customer Kept: impact 2 Signals a push into a vertical you also target
  • Homepage hero image and headline changed Filtered: impact 1 Cosmetic, no change in positioning
  • Blog post about remote work Filtered: impact 1 Not related to product or market
  • Footer copyright year updated Filtered: impact 1 Noise

The filtered findings aren't lost. The brief reports how many were dropped, and the full list stays in the run's conversation if anyone wants to check.

Numbers you can trust

Calculated, never estimated

Language models are good at reading pricing pages and bad at arithmetic under pressure. So the Analyst doesn't estimate: it writes and runs code in an isolated sandbox.

analyst_prices.py Sandbox 
plans = {
    "Competitor A": {"last_week": 49.00, "this_week": 59.00},
    "Competitor B": {"last_week": 45.00, "this_week": 45.00},
    "Our Team plan": {"last_week": 50.50, "this_week": 50.50},
}
for name, p in plans.items():
    change = (p["this_week"] - p["last_week"]) / p["last_week"] * 100
    print(f"{name}: ${p['this_week']:.2f}/seat ({change:+.1f}%)")
gap = plans["Competitor A"]["this_week"] - plans["Our Team plan"]["this_week"]
print(f"Annual saving vs Competitor A, 50 seats: ${gap * 50 * 12:,.0f}")
Output Exit 0 
Competitor A: $59.00/seat (+20.4%)
Competitor B: $45.00/seat (+0.0%)
Our Team plan: $50.50/seat (+0.0%)
Annual saving vs Competitor A, 50 seats: $5,100

Short-lived and isolated

Each sandbox is short-lived and discarded when the run ends, so nothing from one week's calculation leaks into the next.

Choosing a cadence

Weekly, daily, or once before planning season

Different questions need different rhythms. Scheduled Tasks support one-time, daily, weekly, monthly and yearly runs, with a time and timezone, start and end dates, and a maximum run count. You see the next three runs before saving.

  • Weekly

    The weekly brief.

    Every Monday at 8:00 AM, no end date. The backbone of the program.

  • Daily

    The launch watch.

    A competitor has an event on the 18th. Run the Pricing Watcher daily from the 15th to the 25th, then stop automatically.

  • One-time

    The quarterly review.

    A one-time run before planning season, with a longer brief that covers the full quarter for leadership.

  • Daily

    Ad-hoc watches from your team.

    A rep can ask in a conversation, "Check Competitor B's pricing every day this week," and the agent creates the schedule itself. You set the guardrails in advance:

Limit Suggested setting for this use case
Allowed schedule types One-time and daily only
Schedules per conversation Two
Minimum interval between runs 24 hours
Consecutive failures before pausing Three

Your watchlist, in one place

Add a competitor here, publish, and the next run includes them

Competitors, URLs and brief format all live in a single versioned skill linked to every agent in the team. Add a competitor here, publish, and the next run includes them.

  • Competitors and URLs to watch
  • Last recorded prices to compare against
  • Brief rules every run follows
  • Versioned and linked to every agent in the team
Competitor Watchlist · skill v3 · Published 
# Competitor Watchlist
## Competitors
| Name | Pricing page | Release notes | Segment overlap |
|---|---|---|---|
| Competitor A | https://competitor-a.example/pricing | https://competitor-a.example/changelog | SMB, mid-market |
| Competitor B | https://competitor-b.example/plans | https://competitor-b.example/whats-new | Mid-market |
| Competitor C | https://competitor-c.example/pricing | https://competitor-c.example/updates | Enterprise |
## Last recorded prices (per seat, monthly)
Competitor A Team: $49.00 · Competitor B Business: $45.00
## Brief rules
- Include only findings with impact 2 or higher.
- Every finding needs a source page and a "For us" note.
- Never speculate about unannounced plans or internal decisions.

When the web fights back

Built for messy public websites

Public websites are messy. Here's how the agent copes with the problems you'd otherwise discover the hard way.

  • A pricing page is redesigned.

    The Pricing Watcher reports what it can find and flags fields it couldn't locate, rather than guessing. The brief notes the gap so someone can update the watchlist URL.

  • A page hides instructions for AI.

    Some pages contain text aimed at manipulating AI readers. A Responsible AI policy with prompt injection detection screens content before it can redirect the agent.

  • A source is unreachable.

    The run continues with the other sources and reports which one failed. For agent-created schedules, repeated failures pause the schedule instead of retrying forever.

  • A finding looks wrong.

    Open Trace Logging for that run to see exactly which page was fetched, what it returned and how the Analyst used it. You'll know whether the problem was the source, the scoring or the write-up.

  • You change the brief format.

    Edit the skill as a draft, run the evaluation dataset of past findings, and publish only when the results hold up. The schedule keeps using the published version until then.

Beyond the weekly brief

More value from the same research team

  • An archive you didn't have to build

    Every run is a separate, dated conversation, giving you a searchable history of how each competitor has moved over months.

  • Answers on demand

    Publish the Intelligence Lead to the Agent Hub, and anyone in sales can ask, "What changed at Competitor C this quarter?" before a call.

  • Fresher battle cards

    Point the "For us" notes at your battle-card content in the positioning library, and the brief tells you which card sections are now out of date.

  • Cost under control

    Analytics shows credits and latency per agent, so you can see whether the News Researcher is worth searching daily or only weekly.

Get started

Four steps to your first Monday brief

  1. Write your watchlist skill with competitors and URLs.
  2. Create a positioning library with your own pricing and messaging.
  3. Set up the research team with Web Fetch, Web Search and the sandbox.
  4. Schedule the brief for Monday mornings.
Start building Talk to us about market intelligence