Grounding for AI agents · facts only

The facts your AI agent
needs, in one call.

Regime, macro, technicals, fundamentals, insider flows, news and a live document corpus — consolidated, timestamped, returned as plain facts. So your agent stops making numbers up.

Battle-tested: an autonomous paper-trading bot and a 45,000-strategy research lab run on this exact API, every day.
helix — terminallive
REGIME CALM · CRASH 2/18 · VIX 15.8 · 10Y 4.21% · 2Y 3.88% · HY OAS 3.12 · F&G 44 NEUTRAL · WTI 71.2 · GOLD 2981 · DXY 101.1
OverviewMarketsMacroFlowEarningsRiskExplorer
Nasdaq 100
722.8
▼ 2.4% / 20d
S&P 500
751.3
▼ 0.8%
VIX
15.8
calm
Core PCE
3.4%
vs 2% target
10Y-2Y
+0.33
normalized
Sector rotation 20d vs S&P 500
XLE+8.1
XLF+4.2
XLV+2.9
XLI+1.8
XLK-2.4
SMH-3.8
World markets Asia · EU · US
Nikkei▲ 1.2%
DAX▲ 0.4%
FTSE▼ 0.2%
Hang Seng▼ 1.1%
CAC 40▲ 0.3%
Ticker click for the full card
NVDA$182.44▲ 3.1%
RSI 63 · 60d +21% · insiders +$2.1M
Movers covered universe
TSLA +6.7%AMD +6.6%QCOM +5.8%UBER -2.7%PFE -2.5%ABBV -2.4%
Ask Helix facts only
"What did the Fed say about inflation?" → answered from the actual speeches on record, quoting the source. No guessing.
Every module you see is one grounded API call away.

The problem

LLM agents hallucinate numbers.

Ask an agent for a P/E, a drawdown, or what the Fed said, and without grounded data it invents them. Stitching ten vendors and a RAG pipeline to fix that is weeks of plumbing.

Helix

One endpoint. Grounded facts.

A single call returns the consolidated context your agent needs — every field from a named primary source, timestamped, no narrative, no advice. You keep the reasoning; Helix removes the guessing.

Ground an LLM in 8 lines

# one call gives the agent the facts; you keep the reasoning
ctx = requests.get(
    "https://<host>/agent/context?symbols=NVDA",
    headers={"X-API-Key": "YOUR_KEY"}).json()

msg = anthropic.Anthropic().messages.create(
    model="claude-haiku-4-5", max_tokens=300,
    system="Use ONLY the facts in HELIX_DATA. Never invent a number.\n" + str(ctx),
    messages=[{"role":"user","content":"What does the data say about NVDA?"}])

What you get

/agent/context

Consolidated context

Regime, macro, sectors, and per-ticker technicals + fundamentals + insiders + news + earnings, in one grounded blob. Missing data is marked missing — never faked.

/search · /read

Live document corpus

Fed speeches, SEC filings, BEA, ECB and market press — read, indexed and searchable. RAG for finance, without building the pipeline.

/brief

Plain-language daily brief

A readable recap of what moved and why, compiled from primary sources every market day. Also delivered by email.

/data-sources

Transparent & timestamped

Every field shows its source, cadence and live freshness. Delayed, not tick-level — honest by design, built for context not execution.

Pricing — early access

available now
Free
€0
to explore & prototype
  • 250 requests / day
  • Core data: quotes, technicals, macro, sectors
  • Daily brief (on the web)
  • Agent context endpoint
  • Document corpus & search
Get my key
most popular
Developer
€15/mo
run your agent in production
  • 25,000 requests / day
  • /agent/context — the grounded blob
  • Document corpus, search & read (RAG)
  • Daily brief by email
  • Usage dashboard & priority freshness
Start — request access
contact
Team
Custom
volume & reliability
  • Custom volume
  • SLA & support
  • Dedicated endpoints
  • On-prem / private deploy
Talk to us
Early access: paid tiers are being onboarded now — the free key works today.

API docs  ·  Data sources  ·  Status

Helix — facts only, no advice, no predictions. Data is informational only. Not financial advice.