AI image generation / text to image: one image from a text prompt with Google Imagen 4, returned as base64 image bytes with its MIME type. Use it for illustrations, thumbnails, product mockups and social images. Input: prompt; optional aspect_ratio.

Deep web research with citations: search the live web for your question, read the top sources in full, and get a written answer with every claim tied to a numbered source. Use it when an agent needs a sourced answer, not a list of links. Input: question; optional max_sources. Quick answer with links: search.web.

Product availability and price check: is it in stock and buyable right now, at what price, in which options (sizes, colours, dates, rooms), which option matches the variant you name, quantity and ship-to eligibility where the page states them, with evidence and a checked_at stamp. Reads any product or booking page (schema.org, Open Graph, Shopify JSON; a browser only when needed). Input: url; optional variant, quantity, ship_to.

Ask a database in plain English (text to SQL): any BigQuery table, including Google's public datasets. The SQL is written for you, run under a byte ceiling, and the rows are read back into a written answer, returned with the SQL, columns and rows it used. Use it for analytics questions without writing SQL. Input: question, table; optional columns.

Time-series forecasting: forecast any BigQuery table with Google's pretrained TimesFM model (AI.FORECAST), nothing to train. Point it at the table, timestamp column and value column and get forecast rows with bounds for the horizon you set, per series with id_cols. Use it for sales, traffic, demand or metric forecasts. Input: table, timestamp_col, data_col; optional horizon, id_cols.

Website change detection, step two: re-fetch a page saved with monitor.snapshot and get whether it changed, how old the baseline is, and a written summary of what changed. Use it to track price changes, terms updates, new job posts or competitor moves. Input: url; optional language.

Trading and portfolio math: returns and CAGR, volatility, Sharpe and Sortino, max drawdown, VaR/CVaR, beta, alpha and correlation vs a benchmark, SMA/EMA, RSI, Bollinger bands, Black-Scholes option price with Greeks, and the Kelly fraction, over prices you send or a ticker fetched live. Deterministic, every formula named, no LLM. Input: prices or symbol; optional metrics, benchmark, option, kelly.

Anomaly detection on a time series: score recent periods against history with BigQuery AI.DETECT_ANOMALIES (TimesFM). Give one table and its latest periods are scored against its own history, or a history and a target table. Returns every checked row with bounds, anomaly flag and probability, plus the anomaly count. Input: history_table, target_table, timestamp_col, data_col; optional target_last, threshold, id_cols.

Company research report: what a company does, its products and anything notable, from the live web (official site and Wikipedia first) and recent GDELT news, every claim cited to a numbered source, thin or conflicting evidence disclosed. Use it for due diligence, sales prep and competitor research. Input: company name; optional max_sources. Structured firmographics in one cheap call: company.enrich.

Speech to text / audio transcription: transcribe up to 60 seconds or 10 MB of audio with Google Cloud Speech-to-Text and get the transcript with its confidence and detected language. Use it for voice notes, call snippets and clips. Input: audio_base64; optional language_code. Longer audio is refused before payment.

Fact check claims against sources: up to 10 claims checked against up to 4 source URLs you name; each comes back SUPPORTED, CONTRADICTED or UNSUPPORTED with the exact quote the verdict rests on. Use it to verify AI output, citations or statements before publishing. Input: claims[], sources[]. A question but no sources yet: research.web.

Crypto market sentiment in one call: live Coinbase spot price, live Polymarket probabilities and recent news headlines (GDELT) on the same subject, analysed together in a written summary that says what the numbers do and do not support. Use it for a quick read on Bitcoin, Ethereum or any coin. Input: optional product_id (BTC-USD...), query, question, limit. Market data only, not investment advice.

AI video generation / text to video: a short clip from a text prompt with Google Veo, returned as base64 video bytes (or a GCS URI when large) with its MIME type, optionally with generated audio. Input: prompt; optional aspect_ratio (16:9 or 9:16), duration_seconds (4, 6 or 8), generate_audio.

sitepeek (Tanod) OCR: the text of a public PNG, JPEG, WebP, GIF or TIFF image (tesseract), with word count and mean word confidence, run in an isolated, resource-limited process. Read-only.

pdfpeek (Tanod) PDF compression: lossless restructuring or image re-encoding.

Security headers check for a website: audits HTTPS, HSTS, Content- Security-Policy (CSP), X-Content-Type-Options, X-Frame-Options / clickjacking protection, Referrer-Policy and CORS from the real HTTP response of any live URL. Findings with severity for what is missing. Header posture only, not a penetration test.

Extract text from images using Tesseract OCR. Send image as image_base64 (PNG/JPEG/WebP/TIFF/BMP, max 10 MB decoded). Returns text and confidence (0-100, where 80+ is reliable). Set language to Tesseract code: 'eng' (default), 'deu' (German), 'fra' (French), 'chi_sim' (Chinese Simplified), 'jpn' (Japanese), 'ara' (Arabic). Use for invoices, receipts, scanned documents, or screenshots with text.

Transcribe the actual audio of a YouTube, TikTok or Instagram video with Whisper large-v3. Not caption scraping: the audio is downloaded and run through speech recognition, so it works on videos with no subtitles, in any language, and on TikTok and Instagram where no caption track…

Page speed and page weight audit of a web page from one real browser load: total bytes transferred, HTTP request count and DOM node count, measured rather than estimated, with findings when the page is heavy. Page-weight signals, not Core Web Vitals.

HubVibe Agent (quick): hand it any task in plain words and get the finished result in one call. Research a company or a person of interest, check a website, screen a name, read the markets, compare options, pull the facts: you get a headline, the key points, the full answer, its sources and every tool step. Up to 4 steps. Charged only when the answer is delivered. Input: task; optional context, fields for a JSON answer, language.