Each is a real, deployed product with an honest
capability list. Click through and its own email→PIN gate lets you in.
Setu
Real-time voice translation
Two people, two languages, one conversation.
You speak; a beat later the other person hears it in their language, as text and voice.
English and five Indian languages — any pair, both ways, about 1.5 seconds a turn.
Turn-based, and nothing is recorded.
● 8 live
1 out of scope
3 on the roadmap
LensLab
Local image search
Ask questions of your own image folders.
Point it at a folder and search your photos in plain language — "the red bike by the
lake" — with a local CLIP model. Runs on your own machine; nothing about your images
leaves it.
● Private beta
On-device CLIP
Zero egress
FridgeAI
Cook from what you have
A photo of your fridge, and dinner has a plan.
Snap your fridge or groceries; FridgeAI detects the ingredients, you confirm and edit,
then it suggests meals you can actually make — ranked by what you already own, with a
step-by-step cook mode and an assistant that knows your recipe.
● Vision detection
Photo → recipes
Cook mode + chat
DClaw Forecast
Time-series forecasting
See what comes next — and how certain it really is.
Turn historical business numbers into forecasts built from trend and seasonality, with an
uncertainty range around every prediction. Compare models, backtest on your own data, run
scenarios, and ask a grounded copilot — nothing invented, nothing hidden.
● 3 models + Auto
Uncertainty + backtest
Accounts + saved forecasts
Data connectors