🔥What Thousands of VC Agents Do All Day
1. Key Themes
Theme 1: AI Agents Are Industrializing VC Sourcing at Scale
The majority of scheduled AI agent usage in VC is devoted to thesis-driven sourcing, signaling a structural shift in how deal flow is generated — not by humans scanning markets, but by agents running continuously on defined criteria.
"55% of Scouts are some form of thesis-driven sourcing, where thesis is broadly defined as any theory about what to monitor for opportunities."
Theme 2: The VC Job Description Hasn't Changed — Only Who (or What) Does It
The core functions of venture — sourcing, market watching, pipeline and portfolio monitoring — remain constant. What's changed is that agents now execute these functions autonomously and on a schedule.
"Andre's hypothesis holds. The jobs are the same as they were 10 years ago. Funds source, watch markets, and monitor pipelines and portfolios, and all are impacted by agents."
Theme 3: Agents Are Expanding Beyond Sourcing Into the Full Investment Funnel
VC firms are deploying agents not just for top-of-funnel discovery, but for market intelligence, tracking, and portfolio support — representing a much broader automation of the investment workflow.
"Users are now extending Scouts far beyond the first stages of the investment funnel. We identified repeated patterns across 3 categories in addition to sourcing. Market intelligence, tracking, and portfolio support."
Theme 4: Firm-Specific Context (CRM + Past Decisions) Is the Next Frontier for Automation
The highest-leverage unlocked capability isn't raw market scanning — it's agents that can reason over a firm's own historical decisions, network, and scoring criteria.
"Access to the firm's own data and past decision context is the highest leverage missing piece for even more powerful automation. Harmonic already interleaves CRM data, networks, and arbitrary custom data, and now we're working on an even richer context graph."
2. Contrarian Perspectives
Contrarian #1: Human Company Discovery Is Already Obsolete
The conventional view is that experienced investors add value by finding companies others miss. The article argues this function should be fully surrendered to agents — human time spent finding companies is now simply waste.
"Humans should not be wasting their time finding companies anymore. With your network, documented theses, scoring criteria and access to the entire market, agents can not only surface everything: they can also show you only what matters."
The evidence: a single fund (Oyster Bay) has already passed 2,000 companies through automated screening across 37 sub-sector Scouts, with 80% clearing pre-screening filters — a volume no human analyst team could replicate.
Contrarian #2: More Coverage Doesn't Mean More Noise — Pre-Screening Raises Yield
The common concern with AI-driven sourcing is that more volume means more irrelevant results. The data here suggests the opposite: codified preferences plus screening agents actually increase signal-to-noise ratio significantly.
"Roughly 80% of what's surfaced clears the screening skill, and what survives goes to the investment team as a weekly digest in Slack."
"Users who have codified their preferences can invest their time chasing priority opportunities instead of wading through long lists of results."
3. Companies Identified
- Description: Startup discovery and intelligence platform for VCs; offers "Scout" — an AI agent product that can run searches on a scheduled, recurring basis.
- Why mentioned: Primary subject of the article; Earlybird uses it as their "primary data backbone." Data from 300 of its scheduled Scouts forms the empirical basis of the piece.
- Quote: "Scout is the AI for investors. It understands startups and the people behind them so you can: Map markets, Find founders, Evaluate companies, Draft outreach."
Oyster Bay (fund)
- Description: A VC fund with a food and agtech investment thesis.
- Why mentioned: Cited as the most aggressive example of vertical radar deployment — 37 scheduled Scouts covering individual sub-sectors (GLP-1 alternatives, sports nutrition, cultivated meat, etc.), targeting 50+. Their master list has surpassed 2,000 companies.
- Quote: "Nobody has pushed that further than Flynn Seeger at Oyster Bay, who recently walked us through his expansive setup... The master list has passed 2,000 companies, with 137 from something as niche as the sports nutrition Scout alone."
Earlybird (fund)
- Description: European VC firm; Andre Retterath is a partner.
- Why mentioned: Used as the first-person case study for consolidating an external data stack onto Harmonic as its primary data backbone.
- Quote: "Earlier this year, I wrote about the consolidation of our external data stack at Earlybird and why we moved to Harmonic as our primary data backbone."
4. People Identified
Andre Retterath
- Description: Partner at Earlybird VC; author of the Data Driven VC newsletter.
- Why mentioned: Framed the central thesis of the article and commissioned the data analysis from Harmonic.
- Quote: "My hypothesis going in was that the jobs have not changed, but who does them has."
Max Ruderman
- Description: Leader at Harmonic (title not specified, likely CEO or co-founder).
- Why mentioned: Provided the aggregate data analysis across 300 scheduled Scouts and co-authored the guest portion of the newsletter. Has unique cross-market visibility that individual firms lack.
- Quote: "Max Ruderman and his team can see across the market in a way none of us can from inside a single firm."
Flynn Seeger
- Description: Investor at Oyster Bay.
- Why mentioned: Cited as the most advanced practitioner of scheduled AI sourcing agents, running 37+ sub-sector Scouts within a food and agtech thesis.
- Quote: "Flynn has one scheduled Scout per sub-sector inside their food and agtech thesis, covering GLP-1 alternatives, sports nutrition, probiotic sodas, cultivated meat and the rest. The Scouts totaled 37 when we spoke, and he's working toward at least 50."
5. Operating Insights
Insight #1: Codify Your Thesis into Agents Before Scaling Outreach
The most effective users aren't running one general-purpose search — they're decomposing a thesis into discrete sub-sector agents, each with codified preferences and pre-screening criteria. The result is a high-signal weekly digest instead of a raw fire hose.
"A Scout surfaces scored targets on a schedule, then collaborates with the investor to reach conviction... Users who have codified their preferences can invest their time chasing priority opportunities instead of wading through long lists of results."
Insight #2: Automate the Handoff — From Scout to CRM to Outreach
The Oyster Bay workflow shows the full automation loop is achievable: Scout surfaces → screening skill filters → team reviews in Slack → thumbs-up triggers founder outreach and CRM entry. Each step is delegated, so human time is spent only at the conviction decision point.
"Each company gets evaluated, and a 'thumbs-up' sends an outreach to the founder and pushes the record into the CRM."
Insight #3: Volume at the Top of Funnel Is No Longer the Constraint
For operators building or investing in VC infrastructure, the bottleneck has shifted downstream. Getting companies into the funnel is solved. The new constraint is evaluation speed and conviction-building.
"Customers have found sharper ways to raise yield on their hours than we would ever have designed."
6. Overlooked Insights
Overlooked Insight #1: Vertical Radars Dominate — But Four Other Sourcing Scout Types Exist
The article notes that vertical radars make up "almost half" of scheduled Scouts, implying four additional sourcing Scout archetypes collectively account for the other ~27.5% of the 55% sourcing total. These other types are never named or described — a meaningful gap, since they could represent emerging best practices for sourcing beyond sector-based monitoring.
"55% of Scouts are some form of thesis-driven sourcing... Vertical radars make up almost half of scheduled Scouts."
Overlooked Insight #2: The Scheduled Agent Feature Is Brand New — And Already at Scale
The "scheduled Scouts" capability was only made broadly available "earlier this month" relative to the article's publication (August 2026), yet the dataset already covers 300 agents. This adoption velocity suggests VC firms were ready and waiting for this capability — and that the market for agentic VC tooling is compressing faster than most expect.
"Earlier this month, Harmonic made Scouts that run on a schedule available to everyone. That struck me as a window into how VCs are using agents and which tasks investors are willing to hand off at a regular cadence."