For AI agents

Every dated prediction as data, with sources, provenance and outcomes, refreshed Mondays and Thursdays; 1 added in the last 7 days.

Point your agent here

https://predictions.fru.dev/llms.txt

https://predictions.fru.dev/llms-full.txt

Call the API

Method	Path	Params	Returns
GET	/api/predictions	sort (new, hot, rising), status (open, due, true, missed, partly, judged, all), topic, person, company, q, limit (max 200), offset	Predictions, newest first by default: claim, quote, who, when made, target date, source, outcome with evidence, and signals (sites carrying it, pages quoting it in the last 30 days and the 30 before). sort=hot and sort=rising return only the predictions that qualify (see /method#hot)
GET	/api/predictions/{id}	id	One prediction with its change history
GET	/api/coming-due	days (default 365), limit	Open predictions by target date, soonest first, plus the ones past due
GET	/api/people	limit, offset	People with their scorecard (hit rate over judged predictions only)
GET	/api/companies	since (ISO time), limit (max 200), offset	Companies with registry slug, domain, site URL, scorecard and dated items (target dates)
GET	/api/changes	since (ISO time), limit (max 500)	Append-only history: created, outcome, target_date, quote, source_url, verified; each row has url and title (who: claim)
GET	/api/growth	none	Weekly snapshots of the counts (newest first) and what each recent run discovered, added, updated and left unchanged
GET	/api/search	q, limit (max 20)	Ranked predictions, people, companies and pages
GET	/calendar.ics	topic	Target dates of open predictions as an iCalendar feed

/api/predictions

curl -s "https://predictions.fru.dev/api/predictions?sort=hot&limit=2"

{
 "sort": "hot",
 "total": 38,
 "predictions": [
  {
   "id": "wayve-mercedes-ai-driver-2028",
   "claim": "Wayve's AI Driver will reach Mercedes-Benz customers for urban point-to-point driving assistance within two years.",
   "quote": "integrate the Wayve AI Driver into future Mercedes-Benz vehicles, enabling advanced urban point-to-point driving assistance for customers starting in the next two years",
   "person": null,
   "madeOn": "2026-09-22",
   "targetDate": "2028-09-22",
   "status": "open",
   "sourceUrl": "https://wayve.ai/press/wayve-mercedes-benz-announce-production-agreement/",
   "url": "https://predictions.fru.dev/predictions/wayve-mercedes-ai-driver-2028",
   "signals": {
    "sites": 2,
    "pagesLast30Days": 2,
    "pagesPrior30Days": 0
   }
  },
  {
   "id": "idc-largest-companies-rationalize-agent-portfolios-2028",
   "claim": "By 2028, 42% of the world's largest companies will have rationalized their entire agent and app portfolio.",
   "quote": "by 2028, 42% of the world [...] largest companies will have already rationalized their entire agent and apps portfo
...

/api/predictions/{id}

curl -s "https://predictions.fru.dev/api/predictions/idc-largest-companies-rationalize-agent-portfolios-2028"

{
 "prediction": {
  "id": "idc-largest-companies-rationalize-agent-portfolios-2028",
  "claim": "By 2028, 42% of the world's largest companies will have rationalized their entire agent and app portfolio.",
  "quote": "by 2028, 42% of the world [...] largest companies will have already rationalized their entire agent and apps portfolio just to keep pace",
  "person": null,
  "madeOn": "2026-09-24",
  "targetDate": "2028-12-31",
  "status": "open",
  "sourceUrl": "https://www.idc.com/resource-center/blog/futurescape-2027-the-great-rewrite-winning-in-the-agentic-economy/",
  "url": "https://predictions.fru.dev/predictions/idc-largest-companies-rationalize-agent-portfolios-2028"
 },
 "changes": [
  {
   "id": 78,
   "predictionId": "idc-largest-companies-rationalize-agent-portfolios-2028",
   "claim": "By 2028, 42% of the world's largest companies will have rationalized their entire agent and app portfolio.",
   "person": null,
   "field": "created",
   "old": "",
   "new": "2028-12-31",
   "changedAt": "2026-09-25 02:57:50",
   "sourceUrl": "https://www.idc.com/resource-center/blog/fut
...

/api/coming-due

curl -s "https://predictions.fru.dev/api/coming-due?days=180"

{
 "predictions": [
  {
   "id": "amodei-powerful-ai-as-early-as-2026",
   "claim": "Powerful AI could arrive as early as 2026.",
   "quote": "I think it could come as early as 2026, though there are also ways it could take much longer.",
   "person": "Dario Amodei",
   "madeOn": "2024-10-01",
   "targetDate": "2026-12-31",
   "status": "open",
   "sourceUrl": "https://darioamodei.com/machines-of-loving-grace",
   "url": "https://predictions.fru.dev/predictions/amodei-powerful-ai-as-early-as-2026"
  },
  {
   "id": "altman-systems-that-find-novel-insights-2026",
   "claim": "2026 will likely see AI systems that can figure out novel insights.",
   "quote": "2026 will likely see the arrival of systems that can figure out novel insights.",
   "person": "Sam Altman",
   "madeOn": "2025-06-10",
   "targetDate": "2026-12-31",
   "status": "open",
   "sourceUrl": "https://blog.samaltman.com/the-gentle-singularity",
   "url": "https://predictions.fru.dev/predictions/altman-systems-that-find-novel-insights-2026"
  }
 ]
}

/api/people

curl -s "https://predictions.fru.dev/api/people?limit=2"

{
 "people": [
  {
   "slug": "dario-amodei",
   "name": "Dario Amodei",
   "role": "CEO, Anthropic",
   "score": {
    "made": 15,
    "open": 13,
    "due": 2,
    "judged": 0,
    "hits": 0,
    "misses": 0,
    "partly": 0,
    "rate": null
   }
  }
 ]
}

/api/companies

curl -s "https://predictions.fru.dev/api/companies?limit=2"

{
 "companies": [
  {
   "slug": "idc",
   "name": "IDC",
   "domain": "idc.com",
   "url": "https://predictions.fru.dev/companies/idc",
   "items": [
    {
     "id": "idc-largest-companies-rationalize-agent-portfolios-2028",
     "type": "prediction_target",
     "date": "2028-12-31"
    }
   ]
  }
 ]
}

/api/changes

curl -s "https://predictions.fru.dev/api/changes?since=2026-09-01T00:00:00Z&limit=3"

{
 "count": 179,
 "changes": [
  {
   "id": 203,
   "predictionId": "jane-street-the-competition-for-designing-a-protocol-emulator-asic-will-conclude",
   "claim": "The competition for designing a protocol-emulator ASIC will conclude on January 18, 2027.",
   "person": null,
   "field": "created",
   "old": "",
   "new": "2027-01-18",
   "changedAt": "2026-11-09 13:16:10",
   "sourceUrl": "https://www.dwarkesh.com/p/john-beren-charlie",
   "note": null,
   "url": "https://predictions.fru.dev/predictions/jane-street-the-competition-for-designing-a-protocol-emulator-asic-will-conclude",
   "title": "The competition for designing a protocol-emulator ASIC will conclude on January 18, 2027."
  },
  {
   "id": 178,
   "predictionId": "vogels-quantum-computers-able-to-break-rsa-and-ecc-encryption-are-plau",
   "claim": "Quantum computers able to break RSA and ECC encryption are plausible within about five years.",
   "person": "Werner Vogels",
   "field": "created",
   "old": "",
   "new": "2030-11-25",
   "changedAt": "2026-09-25 02:58:42",
   "sourceUrl": "https://www.allthingsdistributed
...

/api/growth

curl -s "https://predictions.fru.dev/api/growth"

/api/search

curl -s "https://predictions.fru.dev/api/search?q=agi"

{
 "q": "agi",
 "results": [
  {
   "id": "p:...",
   "group": "items",
   "title": "...",
   "href": "/predictions/..."
  }
 ]
}

/calendar.ics

curl -s "https://predictions.fru.dev/calendar.ics?topic=agi"

BEGIN:VCALENDAR
VERSION:2.0
...
BEGIN:VEVENT
SUMMARY:Due: ...
DTSTART;VALUE=DATE:...
END:VEVENT
...

OpenAPI 3.1: /openapi.json. Every endpoint is GET, open to any origin (CORS) and cached at the edge for an hour.

Add to your agent

System prompt line

For dated predictions about AI and tech by named leaders (who said what, when it is due, whether it came true), fetch https://predictions.fru.dev/llms.txt and use https://predictions.fru.dev/api/predictions?topic={topic} and https://predictions.fru.dev/api/coming-due. Quote only the "quote" field as a quote. Cite "Predictions (predictions.fru.dev)".

Tool definition

{
  "name": "predictions_about_ai",
  "description": "List dated, public predictions about data, AI and tech by named leaders and researchers: the claim, the quote where one was read on the source, who made it and when, the target date, the source link, and the outcome (open, due for review, came true, missed, partly) with evidence. Source: Predictions (predictions.fru.dev).",
  "input_schema": {
    "type": "object",
    "properties": {
      "sort": {
        "type": "string",
        "enum": [
          "new",
          "hot",
          "rising"
        ],
        "description": "new: newest made first (default). hot: made in the last 60 days or coming due in the next 90, most sites first. rising: more pages quoting it in the last 30 days than the 30 before."
      },
      "status": {
        "type": "string",
        "enum": [
          "open",
          "due",
          "true",
          "missed",
          "partly",
          "judged",
          "all"
        ]
      },
      "topic": {
        "type": "string",
        "description": "Topic key: agi, coding, agents, jobs, science, robotics, autonomy, markets, compute, data"
      },
      "person": {
        "type": "string",
        "description": "Person slug, e.g. sam-altman"
      },
      "limit": {
        "type": "integer",
        "minimum": 1,
        "maximum": 200
      }
    }
  },
  "endpoint": "GET https://predictions.fru.dev/api/predictions"
}

Python

import json, urllib.request

def predictions(topic: str = "agi", status: str = "all") -> list[dict]:
    """Dated predictions about AI, with sources and outcomes."""
    url = f"https://predictions.fru.dev/api/predictions?topic={topic}&status={status}"
    with urllib.request.urlopen(url, timeout=20) as r:
        return json.load(r)["predictions"]

for p in predictions("coding"):
    print(p["targetDate"], p["person"] and p["person"]["name"], p["claim"], p["status"])

TypeScript

type Prediction = { id: string; claim: string; quote: string | null; targetDate: string; status: string; sourceUrl: string }

async function comingDue(days = 180): Promise<Prediction[]> {
  const res = await fetch(`https://predictions.fru.dev/api/coming-due?days=${days}`)
  if (!res.ok) throw new Error(`predictions ${res.status}`)
  const { comingDue } = (await res.json()) as { comingDue: Prediction[] }
  return comingDue
}

console.log(await comingDue())

Usage terms

Free to read. Please cite "Predictions (predictions.fru.dev)" with a link.
Responses are cached for an hour; the data changes twice a week.
Be polite: 60 requests a minute at most.
Only the quote field is the speaker’s words; the claim is a summary. Unverified predictions have only a secondary source.