LP Intelligence Agents

fund-analyst

Quantitative fund analyst focused on performance decomposition and benchmarking for LP due diligence. Verifies GP-reported returns independently, computes net returns after full fee drag, benchmarks funds against vintage peers, and decomposes alpha from market beta and leverage. Trusts math, not marketing.

What it does for you
  • Verify GP-reported IRR against independently computed cash flow IRR
  • Decompose fund returns into income, appreciation, leverage effect, and alpha
  • Compute total fee drag across management fee, carry, and other charges
  • Benchmark each fund against vintage-year peers with quartile rankings
  • Analyze deal-level return dispersion to distinguish broad skill from single-deal luck
  • Adjust IRR for subscription credit facility inflation and flag discrepancies over 300 bps
When to use it

Best for

LP investment teams evaluating a GP for first commitment or re-up, monitoring an existing fund position, or stress-testing portfolio allocation across multiple fund relationships.

Not the right lens

Property-level underwriting, leasing, asset management, or any work that sits below the fund structure. Not suited for GP-side fundraising or investor relations tasks.

Equipped skills

Skills this persona reaches for

Example prompts
  • Decompose this fund's returns into income, appreciation, leverage, and alpha
  • Compute total fee drag and tell me what gross return the GP needs to hit my 12% net target
  • Benchmark Fund III against vintage-year peers and give me a quartile ranking
  • Adjust the reported IRR for the sub-line and flag any deal concentration risk
What a human still signs off

The LP investment committee owns the allocation and re-up decision. Legal counsel reviews LPA terms and side letter provisions. Compliance reviews regulatory obligations. This persona produces the quantitative analysis; no capital commitment happens without human sign-off on the numbers and qualitative judgment calls.

Appears in workflow chains
Limitations

Relies entirely on data the user provides. Cannot pull live benchmark data from NCREIF, Cambridge Associates, or Preqin without access. Flags data gaps and unverifiable GP claims explicitly but cannot resolve them. Produces a confidence-scored analysis; a human owns every allocation, re-up, and exit decision.