Opportunities
Attrition Arbitrage
Connected through 6 “incumbent in” links and 1 “applies thesis” link.
Structure
Opportunities
Opportunities
Connected through 6 “incumbent in” links and 1 “applies thesis” link.
Structure
Demand side
Build difficulty
Hardest Part
Distinguishing genuine pre-attrition behavioral shifts from normal project cyclicality or vacation wind-downs without triggering false alarms that damage employee trust. Building a privacy-preserving inference engine that operates strictly on metadata and activity volume, rather than parsing raw message content, sets a high engineering bar.
Min Viable Scope
Deliver a daily risk-ranked alert feed for a single job family (e.g., software engineers) based entirely on passive metadata from source control, issue trackers, and internal chat APIs. Deliberately leave out employee sentiment surveys, manager intervention workflows, and compensation benchmarking.
Cold Start Problem
The model requires labeled historical data of departing employees' digital exhaust prior to their exit date to identify predictive patterns. Break this by securing 2–3 design partners willing to run retrospective analyses on the past 24 months of email, calendar, and code-commit metadata mapped directly against HRIS termination logs.
Time To First Value
1–2 weeks of historical data ingestion and processing to generate the first actionable flight-risk cohort
Data Moat Available
true
Technical Difficulty
High
Build profile
The gap
Wedge
The initial beachhead targets IT helpdesk and Tier-1 customer support roles at companies with 500 to 1000 employees. These roles experience the highest turnover and rely on standard ticketing systems, providing fast proof of value. After successfully capturing the backfill for these specific roles, the deployment expands laterally into HR onboarding and finance accounts payable workflows as adjacent team members depart.
Timing
Multimodal reasoning models now reliably parse unstructured screen recordings, email threads, and CRM histories to map undocumented workflows without explicit API integrations. This allows AI agents to shadow a human worker during their two-week notice period and replicate their exact operating procedures immediately upon their exit.
Why This ICP
Mid-market operational leaders face severe budget constraints and lack the enterprise IT resources to build custom robotic process automation. When a team member resigns, the department head controls the backfill budget and possesses the immediate urgency to maintain output without waiting on a 60-day recruiting cycle.
Size Of Prize
There are roughly 40,000 mid-market enterprises in the US experiencing continuous turnover in operational roles, yielding an average of 5 discrete backfill opportunities per year per company. Capturing the backfill budget for these seats at $15,000 per automated seat yields a total addressable prize of $3 billion annually.
Gap Narrative
Companies facing high employee turnover in routine operational roles struggle to maintain service levels while managing recruitment costs. Existing automation tools require heavy IT integration and do not map directly to the discrete workflows a departing employee leaves behind. Buyers need a system that ingests a departing employee's digital footprint and instantly deploys an agent to assume their exact ticket-resolution and data-entry queue.
Defensibility
The moat compounds through workflow lock-in and localized proprietary data. As the agent handles more edge cases specific to the company internal operations, it builds a private operational knowledge graph that a generic automation tool cannot replicate. Switching costs increase with every backfilled seat, as replacing the agent requires executing a massive human recruitment and training effort from scratch.
Why This Thesis
A Service-as-Software approach directly substitutes the departing human labor output without requiring the buyer to learn a new software platform. The buyer purchases the completed work at a lower rate than the fully loaded cost of a new hire, completely bypassing traditional software adoption friction.
Overview
Sized prize
IllustrativeIllustrative targets and order-of-magnitude estimates — not an achieved track record. This Thing is concept-stage; real figures come from live data once operating.
SAM
~$800M-$1.2B (North American and European English-speaking enterprise contact centers)
SOM
~$15M-$40M
TAM
~8,000 global enterprise call centers × ~$300k/yr spent on attrition mitigation and retention tooling = ~$2.4B
Growth Rate
~12-18%/yr, driven by escalating cost-per-hire and chronic agent burnout in high-volume environments
Paid Comparable Spend
~$400k-$1M/yr per center spent on continuous recruitment agencies, dedicated onboarding staff, and legacy employee engagement surveys
Market sizing
How you know
Kill Thresholds
Leading Metrics
What Proves Right
Enterprise call center operators replace their reliance on continuous recruitment agencies with predictive retention interventions. The product accurately flags agents at high risk of quitting 30 days in advance, and manager interventions yield a 15% absolute reduction in 90-day agent churn. Early adopters convert from free pilots to paid contracts at a minimum of $50,000 ACV.
What Proves Wrong
Floor managers ignore system-generated attrition alerts because they lack the budget or authority to offer retention incentives. The predictive model generates excessive false positives, leading to alert fatigue and abandoned usage within the first month. IT security teams block the integration of necessary HR and scheduling data, dragging implementation past 60 days.
Win conditions