Point it at your historical data, get back a prediction API. No training runs, no ML team. We’re in private alpha, working with a few design partners.
npx zerograd-mcp add claudeWorks with Claude, Cursor, GitHub Copilot, Windsurf, Gemini over MCP.1from zerograd import ZeroGrad23ai = ZeroGrad()45fraud = ai.create(6 objective="detect fraudulent transactions",7 examples="transactions.parquet", # past outcomes8 target="is_fraud",9)10fraud.deploy()1112fraud.predict({"amount": 1840, "country": "BR", "age_days": 9})
{ "prediction": "fraud", "probability": 0.94 }It's like ChatGPT, but for predicting: machine learning models over your data, and they're getting better fast.
You don’t retrain them every time you need a new task done. Show a few examples and they pick it up from context, without changing their weights: no feature engineering, no expensive retraining, no team tuning a model for each use case.
It was pretrained once, on millions of tables. Give it your history as examples and it predicts the next row: fraud, churn, demand. No training run, no tuning.
p(fraud) = 0.94
During 2026, tabular foundation models went from research papers to solid baselines. On many real datasets they already hold their own against gradient-boosted trees and deep learning.
Language went from one model per task to one pretrained model that gets better every few months. Tables are on the same path, and the next generation will likely beat what you run today.
Test them on your data, use them as your baseline, and be ready to adopt the next one the week it ships. That's the problem we solve.
Without depending on an expensive team. You ship fraud detection, not an XGBoost container: four verbs cover most of the predictive work products need. Some of what teams build with them:
You call one endpoint. We run the evaluation, deployment, monitoring and upgrades a data science team would spend quarters building.
Team timelines and costs are typical US ranges, for illustration.
We watch for drift and re-run a championship on your latest data. The winner waits in shadow, ready to promote.
Every version runs in shadow first, then canary or A/B behind your SLOs, with one-click rollback.
v8 10% of traffic
New foundation models keep landing. We prove them on your workload and swap them in behind the same endpoint.
Every capability is an MCP tool with a Skill that says when to call it. Claude, Cursor, Copilot, LangGraph or your own loop.
→ fraud_detection.predict
probability 0.94 · model v7
Blocking it: 94% is over the 0.90 threshold.
You do two things: describe the prediction and call the endpoint. Everything in between is ours, and every number comes with the model, the data window and the measured error behind it.
Point at past outcomes and say what you want predicted. Nothing gets trained.
Validation, candidate approaches and temporal backtests — on your history, not a benchmark.
An autoscaled endpoint, with monitoring and drift checks on from the first request.
predict · forecast · recommend · detect. Your application consumes an API; you don't manage models.
Quality, latency, data health and the business metric you actually care about.
A challenger beats production on your data, under your constraints.
Proven on live traffic, behind the SLOs and guardrails you set.
Promoted with instant rollback. You change zero code.
We're working with a few design partners who have real predictive workloads. Bring a historical dataset; leave with a benchmark.