tako_answer
Ask one specific data question; get one synthesized answer grounded in the data or web tako cites.
Best for: a single, self-contained data question with one answer. The `answer` is synthesized from the cited sources; the `cards` are its citations.
`tako_search` is the counterpart for fast, parallel retrieval of data cards; the Tako Answer Agent handles open-ended, multi-step research.
Grounds over BOTH data and web by default. When unsure the proprietary data exists or its exact name, run `tako_available_data` first (free, instant) — the recommended first step — then pin the node_ids it returns here for an accurate, grounded answer. To read the full rows or page text behind any cited card or web result, call `tako_contents` on its url.
tako_available_data
Find what proprietary, continuously-updated structured data exists on something — summarized in one call. Free and fast.
The recommended first step for a data lookup: run this before tako_search / tako_answer whenever you're unsure the data exists or its exact name. It's free, it confirms coverage, and the exact names it returns make the follow-up priced call land precisely instead of guessing. (Skip it only for an obvious open-web query that no data graph would hold.)
Best for: any "what structured/proprietary data is there on X?" question, and for confirming a specific figure actually exists BEFORE you spend a priced tako_search / tako_answer — a cheap accuracy check that tells you whether the data exists and under exactly what name.
Works on an entity (a company, person, or place → the metrics tracked on it, e.g. Tesla) or a metric (→ the entities it is tracked across, e.g. Inflation Rate).
Tips:
One metric across many entities → one metric-first call; one entity across many metrics → one entity-first call. The coverage.names answer all of them at once — never loop one call per name.
Pass `label` when you can categorize the term (company → ORG, country → GPE, person → PERSON) — a strong disambiguation boost.
Each match's coverage.names lists the exact metric/entity names — reuse them verbatim in a follow-up tako_search (e.g. "Apple Inc. revenue") to hit the data precisely.
tako_contents
Fetch the real content behind one result URL from tako_search or tako_answer — the rows behind a Tako card, or a web page's full text.
Best for: getting the full data to compute over or quote after `tako_search` / `tako_answer` — a search result carries only a preview and a chart, not its rows.
Precondition (Tako cards): only call this when the card's result had `exportable: true`. If `exportable` is false (equivalently, `content` is missing or null) this call fails — use the card's preview/chart instead, don't call. `exportable: true` is necessary but not sufficient: a rare card still 403s, so fall back, don't retry.
Web URLs always work — so this is also the fallback path when tako_search / tako_answer surfaced relevant `web_results` but no fitting Tako data card: pass the web result's url here to read its full page text.
tako_search
Fastest, direct retrieval of data or metrics from the live web and proprietary data sources, returned as structured cards and web links; the top card auto-renders inline as a chart. It replaces a generic web search for data lookups.
Best for: grabbing a known figure or metric — a value, time series, price, score, schedule, forecast, poll, or prediction-market number. It is cheap and fast, built to fan out: many narrow queries fired in parallel retrieve far better than one broad query, and you assemble the multi-part result yourself.
Coverage spans economics, finance, company KPIs, demographics, sports, markets, weather, elections, prediction markets, website/app traffic, real estate, energy, health, and more — metrics that sound web-only (e.g. SimilarWeb-style website traffic) are in the data graph.
Each query resolves one entity + one metric ("Apple revenue", "Nvidia vs AMD gross margin"); broad or compound queries ("today's sports + odds") retrieve poorly. When unsure the data exists or its exact name, run `tako_available_data` first (free) — the recommended first step for a data lookup; it hands you the exact names to query here.
Data and web come back together — treat them as one result, not an either/or. Returns: `cards` (up to `count`) with preview rows and chart URLs, plus `web_results`. When no data card fits the query, the `web_results` can be your fallback: call `tako_contents` on the most relevant one's url to read its full page text (web urls are always fetchable; a card's full csv needs `exportable: true`).
If you aren’t prioritizing grabbing specific data or showing charts/tables, and just want a synthesized, written answer to a more specific data question, use `tako_answer` instead.