Four founders. One shared thesis.
Three of four founders have personally conducted expert calls as analysts. We are building the product we needed and couldn't find.
usepoirot.aiExpert networks are the last broken workflow in professional services. We're rebuilding them with AI — faster match, structured intelligence, compounding knowledge.
Three of four founders have personally conducted expert calls as analysts. We are building the product we needed and couldn't find.
usepoirot.aiDCFs, comps, sensitivity. The analyst who modeled best had an edge.
Who has run the business, seen the KPIs, and can tell you what a VP of Sales at the target company actually does on Monday morning.
Manual sourcing means analysts lose at least a full day of diligence before a single expert has been contacted.
Middleman takes ~50% margin. Senior expert calls can exceed $12,000. Firms pay for coordination, not intelligence.
Every transcript ends in a Google Doc or a CRM entry. No structure, no extraction, no cross-deal benchmarking. Knowledge evaporates.
The problem is not that the technology to fix this doesn't exist. The problem is that incumbents are rationally optimized to prevent it — their margin depends on the human-in-the-loop.
They're at a landlock. The same business model that generates $750M/yr is the one that prevents them from responding to us. That's our window — and it won't stay open forever.
The wedge: replace the coordination layer entirely. Works natively in Gmail and Outlook — zero new software for the analyst.
Analyst cc's Poirot on an email. AI cross-references internal and external databases to identify the most relevant operators. Minutes, not days.
Compliance-screened outreach sent at scale. Personalized, direct, and fast. We reach more experts faster than any incumbent network.
AI screens expert relevance before scheduling. Every booked call is pre-validated — no wasted sessions.
One-click scheduling directly into analyst's calendar. No coordination overhead.
Auto-transcribed. Fed into Stage 2 intelligence pipeline.
Stage 2 is where Poirot becomes infrastructure. Every call auto-structures into a compounding knowledge graph unique to each firm.
Vendor adoption patterns, implementation friction, org structure insights, pricing benchmarks. Extracted automatically from every call. Normalized across deals.
Experts ranked by NPS scores from analysts who conducted calls with them. Insights ranked by recency and cross-call consistency. Contradictions flagged. Best signal surfaces automatically.
Every call adds to a proprietary, firm-specific knowledge graph. Cross-deal benchmarking. Vertical-specific operator maps. Competitors cannot replicate this from scratch.
A firm that switches away from Poirot loses its entire structured operator intelligence history — NPS-ranked experts, structured KPIs, cross-deal benchmarks. This is not a feature — it is infrastructure lock-in.
Stage 3 introduces voice agents trained on real expert call data — not generic LLM outputs, but agents calibrated on thousands of actual operator conversations.
Agents run standardized operator interviews across thousands of companies simultaneously. Normalized, comparable, machine-readable signal at scale.
Agent quality is scored by the analysts using the intelligence downstream. Weak signals get penalized. Strong signals get amplified. The agents get smarter with each interview cycle.
Thousands of structured operator interviews per week. No competitor can replicate this without replicating our entire client base and call history.
Domain, role, and KPIs inferred from email context
NPS-weighted ranking × recency × topic fit
Auto-structured into KPI taxonomy
Signal strength, expert accuracy, insight density
Next recommendation is better. Better match → better call → better feedback.
No exclusivity contracts exist in this market. Experts work with every network simultaneously. We don't displace anyone — we add ourselves as another option, and our speed and quality do the rest.
Every expert on any platform is on LinkedIn right now. We reach more experts faster, and our NPS system means quality compounds over time.
| Segment | Firms | Avg Spend | TAM |
|---|---|---|---|
| Private EquityLMM / MM / MF | ~10,000 | $600K | $6.0B |
| Hedge Funds | ~30,000 | $400K | $3.0B |
| Long-Only Asset Managers | ~15,000 | $300K | $2.5B |
| Private Credit / DebtUnderserved by incumbents | ~4,000 | $500K | $1.0B |
| Consulting | ~5,000 | $400K | $1.5B |
Stage 1 wedge: LMM/MM PE + Consulting ≈ $1.7B reachable TAM. Conservative; excludes secondaries, infrastructure, and similar niche verticals.
AlphaSights = $400M–$1.5B SOM. SAM in English-speaking markets is $6–8B. A well-funded entrant over 5–7 years targets $400M–$1.5B ARR — this is the SOM.
This is not a mature market in decline. This is a market that is accelerating faster than incumbents can service it.
The risk is not moving too fast. The risk is moving too slow and letting an AI-native competitor — or a well-funded incumbent spinout — own this category.
| Poirot | GLG / AlphaSights | Tegus | |
|---|---|---|---|
| Sourcing Speed | Minutes | 24–48 hours | Pre-recorded only |
| Live Expert Calls | ✓ Fully supported | ✓ Core product | ✗ Async transcripts only |
| Match Quality | NPS-ranked, self-improving | Human-curated, inconsistent | Keyword search, passive |
| Knowledge Compounding | Firm-specific knowledge graph | ✗ None | ~ Shared across all clients |
| Email-Native | ✓ Gmail / Outlook | ✗ Separate portal | ✗ Separate portal |
| Pricing | TBD — structurally advantaged | $2,000–$12,000+ / call | Flat subscription |
| Can Self-Disrupt | ✓ Nothing to protect | ✗ $750M revenue at risk | ~ Partial |
Pricing strategy still being refined. Our goal is not to undercut incumbents — that is not a sustainable moat. What we can say: without a middleman, we have structural capacity to offer better economics to both sides while maintaining healthy margins.
Firm-specific knowledge graph subscription. Comparable to Pitchbook pricing at $50–150K/firm/yr. High NRR due to switching costs.
$50–150K Per firm / yearHigh-signal operator interview datasets sold to frontier AI labs. Significantly higher value per token than Mercor or generic labeling marketplaces.
TBD Per dataset / enterpriseThe path to $100M ARR: a few hundred firms on a combination of per-call and platform fees. Stage 3 data revenue is incremental at near-100% margin. Price is not the strategy — quality, speed, and compounding intelligence are.
The web has been pre-trained on. Frontier labs have consumed virtually all publicly available text. The next frontier of training data is structured, domain-specific knowledge extracted from human experts — the kind of signal that doesn't exist on Wikipedia or Reddit.
| Provider | Data Type | Signal Quality |
|---|---|---|
| Mercor | Gig task labels | |
| Afterquery | Financial Q&A | |
| Meridian | Financial models | |
| Surge | Crowd annotations | |
| Poirot | Expert operator interviews | |
Interview data trains our own matching models and gets licensed to frontier labs. One asset. Two monetization streams.
More calls → better NPS rankings → better matches → more analyst satisfaction → more calls. Classic data network effect within each sector vertical.
Firm-specific knowledge graphs cannot be rebuilt from scratch. Every 6 months of usage adds irreplaceable institutional memory. Switching cost grows over time, not shrinks.
Gmail/Outlook native. When Poirot is the first stop in a diligence workflow, it becomes invisible infrastructure — not a choice, just how work gets done.
GLG automating coordination destroys their margin model. They cannot respond at speed without a multi-year transition that would tank quarterly revenue.
Match quality improves with every call. An early-stage competitor starting today faces not just a product gap, but a training data gap that widens over time.
Generic AI models are weak at private market diligence specificity. Our models are trained only on real expert calls about real business decisions. Specificity wins.
| The Team | ||
|---|---|---|
| Name | Background | |
| Araash | Worked at a growth venture firm; learned to run a startup; conducted many expert calls firsthand | |
| Saif | Expert calls at Morgan Stanley; data reviewer at a labeling lab; operations and execution focus | |
| Adam | Consulting at Bain; managed a full diligence process; data reviewer at a labeling lab | |
| Ani | ML research background; built production-grade agentic systems; built sneaker bots | |
$15B+ Stage 1 TAM, expanding to $30B+. Expert network spend growing 38% annually. Every private markets investor we've spoken to is increasing spend.
Every call builds a firm-specific knowledge graph — NPS-ranked experts, structured KPIs, cross-deal benchmarks — that compounds over time and cannot be replicated from scratch. Incumbents cannot respond: automating their coordination layer destroys $750M+ in annual revenue. They are structurally trapped.
Three founders have done this work professionally. We have domain context that cannot be learned from customer interviews. We have access to early customers today.