Your users want to ask data questions directly in your product. Let them! camelAI is a drop‑in <iframe> that gives your users a complete analytics chat experience.
The Problem
- Users constantly request custom insights and charts that aren’t worth hard‑coding into your dashboard.
- Engineers waste cycles on one‑off data pulls.
- Building a chat‑with‑data feature from scratch means months of work plus ongoing maintenance.
The 5‑Minute Fix
With camelAI’s API & iframe, you can embed a fully featured, customizable Chat With Your Data experience in ~5 minutes.
👉 Get started self‑serve: https://camelai.com/api
YC Demo: Chat with the YC Company Directory
We scraped public YC data (5,294 companies) and wired it into camelAI to show what’s possible.
Try it yourself — no login required, completely free: https://camelai.com/demos/y-combinator-companies/
Ask questions like
- “Five years after Demo Day, what % of companies are still active?”
- “Plot founding team size against survival rates.”
- “Which batches have the highest acquisition rates?”
Note: Funding and revenue data isn’t publicly available, so it isn’t included in this dataset.
What We Discovered About YC Companies
I started by asking camelAI to plot YC companies per year, broken down by status:

Key Findings
- YC’s footprint has grown every year since 2005, with explosive growth post‑2018
- Peak intake: 2021 (727 companies), followed by 2022 (633) and 2024 (599)
- 2025 already shows 365 companies (dataset still incomplete)
Survival Rates (2005‑2020 Cohorts)
Then I analyzed survival rates for 2005- 2020 cohorts to remove any recency bias.
YC Survival Rate Rate Line Chart (2005-2020)
YC Survival Rate Rate Table (2005-2020)
| batch_year |
Total |
Inactive |
Survived |
Survival Rate % |
Period |
| 2005 |
8 |
3 |
5 |
62.5 |
2005-2010 |
| 2006 |
18 |
11 |
7 |
38.9 |
2005-2010 |
| 2007 |
32 |
16 |
16 |
50 |
2005-2010 |
| 2008 |
43 |
31 |
12 |
27.9 |
2005-2010 |
| 2009 |
42 |
21 |
21 |
50 |
2005-2010 |
| 2010 |
63 |
24 |
39 |
61.9 |
2005-2010 |
| 2011 |
105 |
38 |
67 |
63.8 |
2011-2015 |
| 2012 |
149 |
64 |
85 |
57 |
2011-2015 |
| 2013 |
98 |
36 |
62 |
63.3 |
2011-2015 |
| 2014 |
152 |
44 |
108 |
71.1 |
2011-2015 |
| 2015 |
216 |
62 |
154 |
71.3 |
2011-2015 |
| 2016 |
224 |
73 |
151 |
67.4 |
2016-2020 |
| 2017 |
241 |
68 |
173 |
71.8 |
2016-2020 |
| 2018 |
277 |
64 |
213 |
76.9 |
2016-2020 |
| 2019 |
371 |
81 |
290 |
78.2 |
2016-2020 |
| 2020 |
436 |
92 |
344 |
78.9 |
2016-2020 |
camelAI definition
Survived = Active + Acquired + Public
Inactive = Dead / Defunct
-
Average survival sits just under two‑thirds
- Across the full window the mean survival rate (2005‑2020) is 61.9%
-
Clear upward trend over time
| Period |
Avg. Survival |
| 2005‑2010 |
~50% |
| 2011‑2015 |
~65% |
| 2016‑2020 |
~75% |
Survival has improved by roughly 25 percentage points in fifteen years.
-
2008 is the outlier low point
- 2008 cohort survival: 27.9%
- Aligns with the global financial crisis suggesting macro shocks hit young startups hard.
-
Post‑2014 cohorts consistently clear the 70 % bar
- 2014 & 2015: just above 71 %
- 2016‑2020 climb steadily, reaching ~79 % for the 2020 class after 4+ years in the market.
-
Why the improvement?
- YC dramatically scaled batch sizes after 2014 but managed to hold number of failures fairly flat (≈ 60‑90/yr).
- With a larger denominator, the inactive share naturally shrinks.
- A friendlier venture environment (cheap capital) in the 2013‑2021 cycle extended runway for many companies.
Acquisition Dynamics
- 2014‑2017 mark the peak years for acquisition counts.
- Later cohorts: more of the “survived” bucket remains Active (fewer exits so far) indicating longer paths to liquidity.
Take‑Aways (Removing Recency Bias)
- A YC startup launched 2005‑2020 has ~2‑in‑3 odds of being alive or having exited successfully today.
- Survival odds have improved materially, but downturns (e.g., 2008) can halve those odds for a given cohort.
These insights come straight from camelAI’s filtered Survival‑Rate Table and Line Chart.
Bring Chat With Your Data to Your Product
Stop hand‑coding SQL queries. Let your users discover insights themselves.