The only AID system that makes 100% of insulin dosing decisions — no carb counting, no correction factors, no basal rate programming. Just enter your weight and announce meals qualitatively.
Device specifications change frequently — always verify current information directly with the manufacturer before making any decisions. Full disclaimer.
New to the iLet? Start here.
Every other automated insulin delivery system requires you to program basal rates, set an insulin-to-carb ratio, dial in a correction factor, and count carbohydrates at every meal. The iLet eliminates all of that. It starts with one number — your body weight — and figures everything else out on its own.
To start the iLet, you enter one number: your body weight. No basal rates, no carb ratios, no correction factors. The algorithm calculates everything else and starts automating immediately — no warm-up period.
The algorithm continuously learns your actual insulin needs and refines its dosing in real time — 24 hours a day, 288 automatically determined basal segments per day. It adapts to your changing needs without you adjusting anything.
When you eat, you don't enter grams of carbohydrates. You just tell the iLet roughly how much you're eating relative to your typical meal. That's it. The system doses accordingly.
Works with Dexcom G6, Dexcom G7, or FreeStyle Libre 3 Plus. The iLet is the only AID system that offers choice across both major CGM platforms — Dexcom and Libre.
Instead of counting carbs, you pick one of three options relative to what's typical for that meal — breakfast, lunch, or dinner. The algorithm does the rest.
This is a typical meal for you — your usual breakfast, a normal lunch. The algorithm applies what it has learned about your typical insulin needs for this meal.
Smaller than usual — a light snack, half a meal, or a day when you're not that hungry. The algorithm doses more conservatively.
Larger than usual — a big holiday dinner, extra portions. The algorithm doses more aggressively. You don't estimate exactly how much more.
Ever since my son developed type 1 diabetes as an infant almost 22 years ago, I had hoped that technological advancements would bring even better glycemic control to people with diabetes — and without the relentless demand that insulin-dosing decisions fall to them.
— Dr. Ed Damiano, Founder & Executive Chair, Beta Bionics · Juicebox Podcast Episode #934 · juiceboxpodcast.com/episodes/jbp934Scott has interviewed the Beta Bionics team multiple times on the Juicebox Podcast: Episode #934 with Founder Ed Damiano · Episode #1217 with CMO Dr. Steven Russell · Episode #1401 with Clinical Services Director Kelly Postiglione Cook
People burned out by the math of diabetes management. Those not meeting goals on current AID systems. Patients in primary care settings without specialist access. Youth whose families want fewer daily decisions.
Unlike all other AID systems, the iLet requires no ICR, ISF, or basal rate programming from the clinician or patient. Initiation requires only body weight. No diabetes-specific settings needed at start.
The iLet is CGM-agnostic between Dexcom and Libre. This is valuable for patients with insurance that covers one but not the other, or for those with strong CGM preferences.
Users cannot override the algorithm — no manual correction doses, no user-set ICR. Patients who want to be "in control" of every dose may find this frustrating. Discuss expectations before initiating.
Transitioning to the iLet requires a mental shift more than a technical one. The hardest part for most people isn't setting it up — it's trusting it to do its job without intervention.
The Learning CurveThe iLet's algorithm is designed to learn your specific needs and improve over time. Resist the urge to intervene — the system cannot be overridden for corrections, and trying to "help" it by changing behavior can slow its learning.
The algorithm learns continuously. Early weeks may show more variability as it calibrates. Studies show meaningful TIR improvements often visible within just one day of use — and the system keeps refining itself for months.
The default target is "Usual" (120 mg/dL). You can shift it "Lower" (110 mg/dL) or "Higher" (130 mg/dL) in 10 mg/dL increments, and set a different target for part of the day. This is the main lever available to users.
The iLet is a tubed pump — infusion sets are changed every 3 days as with standard pumps. Most common hyperglycemia events in the pivotal trial were traced to infusion set issues. Consistent site rotation and set changes matter.
Simpler than any other AID system. Enter body weight in the iLet. Choose CGM. No basal rate programming, no ICR, no ISF. No run-in period — automated delivery starts immediately. Training focuses on meal announcements and target adjustment, not settings.
Review iLet data. Common early issues: infusion set failures causing hyperglycemia, user anxiety about "not doing enough." Reinforce trust in the algorithm. Check that meal announcements are being used consistently.
A 2025 RCT in Clinical Diabetes showed iLet improved glycemia when deployed in primary care settings and via telehealth — comparable results to endocrinology care. This is a meaningful differentiator for patients without specialist access.
Counsel patients that they cannot manually correct and should not expect to adjust ICR or basal rates — there are none. The algorithm's job is to handle what they used to manage. This is a feature, but requires buy-in.
The iLet has more published clinical trial data behind it than almost any other AID system — including a pivotal trial published in the New England Journal of Medicine. Here's what the science actually shows.
Pivotal Trial ResultsThe Insulin-Only Bionic Pancreas Pivotal Trial (IO BPPT) enrolled 440 adults and children ages 6–79 with T1D — the most diverse AID pivotal trial population ever, including participants on MDI, standard pumps, and existing hybrid closed-loop systems.
Participants randomized to the iLet bionic pancreas had on average a lower hemoglobin A1C, a lower average glucose, and more time in range — despite the reduced requirement for input from users in the iLet group compared to participants in the standard-of-care group.
— Dr. Steven J. Russell, Trial Chair, Harvard Medical School · Russell et al., NEJM 387(13):1161–1172, 2022The idea of having a system this automated was something that was scary to other pump manufacturers — they preferred a system that still left more of the responsibility in the hands of the user, so that if something went wrong, they could say they were not responsible. That's why Beta Bionics had to get started.
— Dr. Steven Russell, CMO, Beta Bionics · Associate Professor of Medicine, Harvard Medical School · Juicebox Podcast Episode #1217 · juiceboxpodcast.com/episodes/jbp1217The algorithm calculates a new basal segment every 5 minutes — 288 per day — automatically, without any user input. Each segment is determined based on current CGM data, trend, learned insulin sensitivity, and time of day patterns.
Unlike other AID systems that use fixed programmed rates as a baseline, the iLet has no programmed baseline to revert to. It continuously updates its model of your insulin needs using every CGM reading and meal announcement as data points.
When you announce a meal, the algorithm delivers approximately 75% of its estimated meal dose immediately — a dose the user cannot modify. The remaining insulin is delivered adaptively based on the observed glucose response.
The algorithm continuously delivers correction insulin when glucose is elevated or trending up, and reduces or suspends delivery when glucose is trending low — without any user action. All correction decisions are automatic.
The C|A|R|E|S Framework from the PANTHER Program at the Barbara Davis Center for Diabetes. Updated through 2025.
Lifelong learning algorithm. 288 automatically determined basal segments/day. Delivers basal, correction, and meal-announcement doses — 100% of all insulin doses. Initialized on body weight only. No basal rates, ICR, or ISF programmed. Meal bolus: ~75% delivered immediately on announcement; remainder delivered adaptively based on glucose response.
Glucose target: "Lower" (110), "Usual" (120), or "Higher" (130) mg/dL — can be set differently for part of the day. Meal announcements: Usual / Less / More. Cannot adjust: correction doses, basal rates, ICR, ISF. No extended bolus option. Users cannot override or modify algorithm-determined doses.
If CGM signal is lost, a blood glucose meter can be used for manual BG input to continue automated dosing for a limited period. The system is designed for continuous CGM input — persistent signal loss will require management via finger sticks and user judgment until signal is restored.
Counsel patients that they cannot manually correct. Meal announcements should be made for all meals and snacks. Hyperglycemia = check infusion set first. For exercise: shift target to "Higher" before activity, return to "Usual" after. No warm-up period — automation starts immediately at setup.
Compatible CGMs: Dexcom G6, Dexcom G7, FreeStyle Libre 3 Plus. Remote monitoring: Bionic Circle App — up to 10 followers see glucose, meal announcements, insulin doses, and receive alerts. Data: iLet Mobile App + Beta Bionics Cloud. Unique: followers see dosing decisions, not just glucose.
Pivotal trial (Russell et al., NEJM 2022): n=440, ages 6–79. HbA1c reduced 0.5% vs. standard of care; TIR +11% (2.6 hrs/day) vs. SoC; no increase in hypoglycemia. Extension study (PMC 2022): Control group crossover to iLet: HbA1c 7.7% → 7.1%, TIR 53% → 65%. Primary care RCT (Oser et al., Clinical Diabetes 2025): iLet effective in primary care and telehealth settings, comparable to endocrinology care.
Source: PANTHER Program · Barbara Davis Center for Diabetes · University of Colorado · Russell et al. NEJM 2022 · Oser et al. Clin Diabetes 2025