What It Does
The Agentforce-powered prospecting agent enriches existing Salesforce data with signals from web and external sources — funding announcements, hiring trends, product launches, news events, executive moves, and 10-K disclosures for public companies. Output: a prioritized prospect list refreshed continuously, surfaced as a custom List View on the Lead and Account objects with a per-record Signal_Score__c field. Signal sources include the Data Cloud Web Engagement connector, the Sales Cloud LinkedIn Sales Navigator integration, and a Salesforce-managed news feed that covers ~40k publishers in 12 languages.
Why It’s Different
Previous “prospecting tools” required reps to run searches and interpret signals manually. An always-on agent does this in the background; reps see a curated queue when they open Salesforce. Minutes saved per day, attributable to the agent. The shift is from rep-as-researcher to rep-as-closer. Disclosed early customers report 30–60 minutes per rep per day reclaimed and a measurable lift in opportunity creation rate from agent-surfaced leads.
Agent loop (runs every 15 minutes per ICP):
1. Pull new signals from configured sources
2. Match signals to existing Accounts/Leads (via Identity Resolution)
3. Score using rep-tunable weights + ICP fit model
4. Apply suppression: do-not-contact, recent-touch, customer-already
5. Write top-N to the rep's Prospecting Queue with rationale
6. Generate suggested first-touch email draft per record
Signal Quality
Noise is real — funding round alerts don’t always correlate with purchase intent, and a hiring spike at a target account might mean expansion or might mean churn. Tune the signal weights per ICP. A healthcare rep doesn’t care about the same signals as a logistics rep. The agent supports a per-rep weight override on top of the team-level default, so a rep who finds executive moves more predictive than funding rounds can bias accordingly.
Example ICP weights (B2B SaaS, mid-market, sales tech):
Funding round (Series B/C) 0.35
Head-of-Sales hire (last 60d) 0.30
Competitor mentioned in news 0.15
10-K language signaling change 0.10
Web engagement on /pricing 0.10
Measuring Value
Track: leads surfaced per day, convert-to-meeting rate per agent-surfaced lead versus baseline, opportunity creation rate, and pipeline dollars attributed to agent-surfaced prospects via campaign attribution. If the convert rate beats baseline by 20%+ on a 90-day window, the agent earns its slot. If it doesn’t, the issue is usually ICP definition or signal weights, not the model.
Common Failure Modes
- Letting the agent surface every signal without suppression — reps drown and trust collapses in a week.
- Ignoring the rationale field — reps need to know why a lead surfaced to write a credible first touch.
- Not removing converted leads from the pool, leading to repeat outreach and bad CSAT signals.
- Skipping the rep-tunable weights — one-size-fits-all scoring underperforms persona-specific weighting by 25–40% in measured tests.
What to Do This Week
Pick one ICP, define five signal weights, and run the prospecting agent against it for 14 days. Compare the convert-to-meeting rate against your baseline list-pull process.