ODYSS takes $199 reservations for a meal-tracking necklace backed only by in-house tests

In a hands-on demonstration, The Gadgeteer confirmed that a coffee entry in the N1's companion app had been recognized automatically ^[11]. Noodles sat in the same meal history, and the report did not establish whether the necklace logged them or a person added them by hand ^[11].

The N1 has three jobs: noticing that someone is eating, naming the food, and estimating how much of it there is. The headline accuracy figure covers the first, and ODYSS warns that even that one varies with the type of food, the environment and how the device is worn ^[9][10]. For naming, ODYSS matches what the camera sees against open food databases plus its own product-image and ingredient data ^[18]. A co-founder told The Gadgeteer that packaged foods can be identified precisely, a statement TechRepublic says is not backed by published independent testing ^[12]. Portion size and calories come from visual and depth information, according to the company's demonstration ^[13]. ODYSS has not published independent validation of how well the N1 identifies foods, estimates portions or calculates calories ^[14].

The pitch is a wearer who never types a meal again ^[5]. TechRepublic's assessment is that the N1 may reduce logging without ending manual input, particularly when it is not being worn or cannot identify a meal confidently ^[2].

The three scores lean on different jobs. Rhythm considers meal timing and eating speed, so it rests mostly on event detection, the one task ODYSS has put a number on ^[16]. Energy compares intake against targets and Quality looks at nutritional balance ^[16]. Both need the food named and the calories estimated. Research on that step says the approach can work, with accuracy that varies considerably. A 2025 systematic review of 47 studies covering 22 image-based dietary apps found an average per-meal calorie error of 18.7% across the 31 studies with usable data ^[15]. The other 16, about a third, did not feed that average ^[20].

The VIP price is $100 below list, about a third off, and includes lifetime access to the N1's AI features ^[19][6]. TechRepublic's report does not say whether list-price buyers get the same access.

In my view the N1 is for someone who wants a rough record of when and how fast they eat, and who is willing to open the app and fix what it gets wrong. ODYSS says its analyses, scores and AI guidance are for general wellness and should not replace medical or professional dietary advice ^[17].

Anyone recommending it can sort the decision on two axes: whether the action they would take depends on calorie numbers or on meal timing, and whether the wearer will correct entries. Meal timing with no corrections sits closest to what ODYSS has measured. Adding corrections to a timing use changes little. On the calorie side, a wearer who corrects is getting a first draft of a food diary, worth the typing it saves minus the fixing it adds. The fourth quadrant is calorie targets with nobody correcting, where raw estimates go straight into the Energy score against a field average error of 18.7% ^[15]. That quadrant is the one to hold until an independent test on production units exists.