Forecast Outcomes for a List of Entities
Take any list of entities, define a scenario or outcome, and forecast it for every row. Numeric forecasting produces a full probability distribution per entity, so you can compare across the list and see where the uncertainty lies.
Here, we forecast what 80 prominent AI researchers would command in startup valuations if they left their jobs to found AGI companies, using the market value of their skills as a proxy for their influence on the field.
Add FutureSearch to Claude Code if you haven't already:
claude mcp add futuresearch --scope project --transport http https://mcp.futuresearch.ai/mcp
Start by building a list of entities:
Find a list of prominent AI researchers who haven't yet started AGI startups.
Claude uses FutureSearch's agent tool to research and compile the list. Then forecast an outcome for each:
Forecast the seed round valuation each of these researchers would command
if they announced an AGI-related startup today. Give valuations in billions USD.
Claude calls FutureSearch's forecast MCP tool in numeric mode:
Tool: futuresearch_forecast
├─ data: [{"question": "What would Noam Brown's AGI startup be valued at?", ...}, ...]
├─ forecast_type: "numeric"
├─ output_field: "valuation"
├─ units: "billions USD"
└─ context: "Assume an AGI-related startup announced today with a completed seed round."
→ Submitted: 80 rows for numeric percentile forecasting.
Session: https://futuresearch.ai/sessions/...
Tool: futuresearch_progress
→ Running: 12/80 complete, 68 running (120s elapsed)
...
Tool: futuresearch_progress
→ Completed: 80/80 (0 failed) in 540s.
Tool: futuresearch_results
→ Saved 80 rows with valuation_p10 through valuation_p90 columns.
The same pattern works for any entity-outcome pair. Replace the list and the question:
Here's a CSV of 50 consumer products. Forecast the year-over-year
revenue growth for each one, in percent.
Add the FutureSearch connector if you haven't already. Then ask Claude:
Find a list of prominent AI researchers who haven't yet started AGI startups. Then forecast the seed round valuation each would command if they started an AGI company today. Give me the full probability distribution in billions USD.
The same approach works for any entity-outcome pair:
Here are 30 countries. Forecast the GDP growth rate for each one in 2027, in percent.
Go to futuresearch.ai/app and enter:
Find a list of prominent AI researchers who haven't yet started AGI startups. Then forecast the seed round valuation each would command if they started an AGI company today. Give me the full probability distribution in billions USD.
pip install futuresearch
export FUTURESEARCH_API_KEY=your_key_here # Get one at futuresearch.ai/app/api-key
The pattern is always the same: build a DataFrame of entities with a question column, then call forecast with forecast_type="numeric". Here, the entities are AI researchers and the outcome is a hypothetical startup valuation:
import asyncio
import pandas as pd
from futuresearch.ops import forecast
researchers = pd.DataFrame([
{
"question": "What would Noam Brown's AGI startup seed round be valued at?",
"background": "OpenAI research scientist. Known for poker AIs (Libratus, Pluribus) "
"and OpenAI's o1 reasoning model.",
},
{
"question": "What would Geoffrey Hinton's AGI startup seed round be valued at?",
"background": "Turing Award winner, 'godfather of deep learning'. Currently focused "
"on AI safety advocacy. Would attract top researchers.",
},
{
"question": "What would Alec Radford's AGI startup seed round be valued at?",
"background": "Lead author on GPT-1, GPT-2, CLIP, Whisper, and DALL-E at OpenAI.",
},
{
"question": "What would Tri Dao's AGI startup seed round be valued at?",
"background": "Created FlashAttention and co-created Mamba. Focused on compute "
"efficiency — a rare AI subfield with immediately obvious commercial value.",
},
# ... more researchers
])
async def main():
result = await forecast(
input=researchers,
forecast_type="numeric",
output_field="valuation",
units="billions USD",
context="Assume an AGI-related startup announced today with a completed "
"seed round. The founder recruits prominent researchers but is "
"the most prominent person.",
)
print(result.data[[
"question", "valuation_p10", "valuation_p25",
"valuation_p50", "valuation_p75", "valuation_p90",
]].sort_values("valuation_p50", ascending=False))
asyncio.run(main())
Replace the DataFrame and the output_field/units for any entity-outcome pair:
# GDP growth for a list of countries
countries = pd.DataFrame([
{"question": "What will Germany's GDP growth rate be in 2027?"},
{"question": "What will Japan's GDP growth rate be in 2027?"},
# ...
])
result = await forecast(
input=countries,
forecast_type="numeric",
output_field="gdp_growth",
units="percent year-over-year",
)
Since Ilya Sutskever started Safe Superintelligence (valued at $5B at seed, reportedly $32B now), AI researcher talent has been a defining factor of startup valuations. Following Sutskever came Mira Murati ($12B seed) and Yann LeCun ($4.5B seed), each raising primarily on their names alone. The top results from forecasting 80 researchers:
| Researcher | Median Valuation | Range (p10–p90) | Key Factor |
|---|---|---|---|
| Noam Brown | $6.7B | $1.4B–$13.8B | o1 reasoning model; investors would bet on another "reasoning" breakthrough |
| Geoffrey Hinton | $5.8B | — | "Godfather of deep learning"; value is in the researchers he'd attract |
| Alec Radford | $4.3B | $1B–$11B | Lead author on GPT-1/2, CLIP, Whisper, DALL-E |
| John Jumper | $2.7B | — | AlphaFold2, 2024 Nobel Prize in Chemistry |
| Yoshua Bengio | $2.3B | — | Co-godfather of deep learning, AI governance focus |
| Tri Dao | $1.8B | — | FlashAttention, Mamba; compute efficiency has obvious commercial value |
| Andrew Yao | $1.4B | $280M–$5.3B | Range driven entirely by geopolitics and US-China capital flows |
| Been Kim | $930M | — | Creator of TCAV; interpretability-based startup could command a premium |
The wide ranges reflect real uncertainty. Andrew Yao's valuation spans from $280M (US-China capital decoupling limits his investor base) to $5.3B (backed by sovereign wealth as a "national champion"). Alec Radford's range depends on whether he'd build a small research lab or a frontier competitor.
Of the researchers forecasted, Noam Brown was flagged as most likely to actually leave — mostly because people keep leaving OpenAI. The full analysis is in Which AI Researchers Have the Most Valuable Skills?.
Built with FutureSearch. See the forecast documentation for all parameters and output formats. Related guides: Turn Claude into an Accurate Forecaster, Find Profitable Polymarket Trades, Value a Private Company.