Research on early-stage investment decisions consistently finds that gut feel plays a larger role than formal analysis. Laura Huang's work at Harvard Business School, including a 2015 study in Administrative Science Quarterly and a 2014 paper in Academy of Management Journal, found that angel investors rely on intuition over structured evaluation, and that gut feel about the entrepreneur outweighs assessments of the business itself. There just isn't much information available at that stage, and the tools that exist haven't kept pace with the decisions they're meant to support. Teams rely on instinct and whoever has bandwidth that week.
Strong businesses get passed over when a pitch deck doesn't do the company justice. Investment decisions get made on the strength of a founder's narrative rather than what the analysis supports. Firms carry the same blind spots from fund to fund, year to year, because nothing in their process makes them visible.
Why gut feeling dominates
Venture capital is one of the few asset classes where information asymmetry is still significant. The data is thin, the timeframes are long, and most of what you're working with is hard to quantify.
But it also means decisions are hard to defend. Ask most investors why they passed on a company and you'll get a coherent story. Ask them what their hit rate looks like against companies they passed on for the same reasons, and they won't know. The data was never captured. The logic was never formalized.
There's a structural problem underneath this. VC returns follow a power law: the bulk of a fund's returns will come from one or two outlier investments. To have a reasonable probability of including one of those companies in a portfolio, a fund needs to evaluate a much larger universe of deals than most expect. The numbers are brutal. When Paul Gompers and colleagues surveyed 700 VC firms in 2020, they found the average closed deal takes 118 hours of due diligence spread over 83 days, and firms review roughly a hundred opportunities for every one they fund.
When reviewing a large number of applications, evaluation criteria start to drift. The first company gets a careful read. Later in the process, the bar has already shifted and nobody noticed. Ask what changed and the honest answer is: nothing was ever written down to begin with.
The process has no feedback loop and no memory. Vinod Khosla, the Sun Microsystems co-founder who went on to start Khosla Ventures, once said it takes seven years and $30 million to train a venture capitalist. That's what learning looks like when the only feedback loop is your own expensive mistakes.
It doesn't have to work that way. A team at the University of St. Gallen tested an investment algorithm against 255 angel investors, asking both to pick from 623 deals sourced from one of Europe's largest angel networks. The algorithm beat the average investor, and it beat experienced investors who fell prey to their own cognitive biases. The one group that beat the algorithm was experienced investors who kept theirs in check. Most of the gap between average and excellent, in other words, is bias control and process. Both can be built.
Auryn closes that gap. A consistent process lets even lean investment teams handle the volume of deal flow a power-law portfolio requires, and makes sure every deal gets a real assessment of its outlier potential. When that holds, every company in your portfolio earned its place.
What Auryn does
Auryn covers the investment workflow end to end: finding companies, screening them, running diligence, modeling exits, monitoring the portfolio, and administering the fund. Most tools own one slice of that and leave you to stitch the rest together. The same platform scales from an angel writing first checks to a fund reporting to LPs, so the process you build at one stage carries over to the next.
Sourcing
Deal flow shouldn't depend on what happens to land in your inbox. You write down your investment thesis in plain text and Auryn goes looking. It finds companies that match, checks them against your criteria, tracks down the right founder, and opens the conversation. Off-thesis companies get filtered out before they cost you attention, and anything borderline waits for your review instead of being acted on. What reaches your pipeline is a steady stream of companies you'd want to talk to.
Screening
Every inbound application gets screened against your investment thesis automatically. The AI reads the pitch deck, extracts key metrics, and produces a structured alignment score with reasoning tied to criteria your team defines. The output is an evaluation you can act on, not a summary.
Structured due diligence
Analysis runs as a staged pipeline: screen, follow up, analyze, quantify, verify. Each stage has a human gate, so Auryn prepares the work and your team makes the calls. Diligence questions are standardized across your pipeline and pre-populated from available documents, with confidence scores and source citations. Follow-up questions for founders come from what the analysis found missing. The framework stays consistent regardless of who runs the analysis.
Exit scenario modeling
We calculate probability-weighted MOIC for every investment, with exit research that pulls in comparable company data, market conditions, and your team's assessments. The model projects the most plausible exit path for the specific company: who the likely buyer is, how large that exit tends to run, and which rounds the company will raise on the way there, with the dilution that implies. Instead of a single number pretending to see the future, you get a range of scenarios with explicit assumptions and a structured view of the uncertainty.
Portfolio monitoring
KPI tracking, risk milestone management, and a live cap table reflecting every funding round and liquidity event. Portfolio companies report through their own founder portal, and every company carries a derived status, so the ones drifting toward trouble surface before the board meeting where you'd otherwise find out.
Fund administration
An LP registry with commitments, capital calls, distributions, and fund expenses, plus capital account statements and notices generated from the same records. For a small fund or family office, this is the difference between needing a full fund administration stack and not. Positions, TVPI, DPI, and IRR are always derived from recorded transactions, never typed into a spreadsheet cell.
Why not just ask ChatGPT?
The two aren't doing the same job. Paste a deck into a chat and you get an articulate reaction to a single document, different every time you ask. Run the same deal through Auryn and it goes through a research pipeline: screening against your thesis, background verification on the team, follow-up questions the founder answers directly, exit research across comparable transactions, and a valuation model whose probabilities are bounded adjustments to baselines calibrated on large-sample venture data. The arithmetic behind it (ownership, dilution, fund metrics) is deterministic code running on recorded transactions, not model output. And all of it lands in the same system that tracks the company for years after you invest.
A chat reacts to a document. This is diligence.
Auryn is open to sign up with a free tier, and we work closely with the teams on it to shape where it goes next. Whether you run a fund, a family office, or your own angel portfolio, if deal flow still lives in spreadsheets, the fastest way to judge Auryn is to put one real deal through it.
Want to continue the conversation?
Drop us a line if something here sparked an idea, or if you just want to talk.
Get in touch