Guide series
Automated Insulin Delivery Systems
Understand how AID systems work, what improvement to expect on average, and how to explore which system may suit you best.
GNL Grace
Not sure which AID system to explore, or want to understand how the algorithms compare? Ask Grace and she will take you to the most relevant evidence.
How this guide works
This page explains what AID systems are, how they work on average, and what improvement is typically seen. It also links to individual system pages for Control-IQ, MiniMed 780G, CamAPS FX, and Omnipod 5.
Recommended approach
- Read this overview page first to understand the foundations
- Then explore the individual system page that applies to you
- Use the resources and tools to deepen your understanding
- Discuss direction of change with your diabetes care team, not just settings in isolation
Adhering to the principle of skin in the game, this guide was written after spending from 2017 to 2026 gaining personal and professional experience. Every commercially available AID system has been used personally, and more than 300 children and young people have been supported through AID starts in clinical practice.
Short version: all four are clinically meaningful step-changes over MDI or pump-without-AID. Choosing between them is like choosing a supercar, there is no single best system, only the right one for you.
What this guide will help you understand
- What AID systems are and how they work on average
- Simple decision criteria for exploring which system may suit you
- Practical principles to get the most from whichever system you are on
What are AID systems and how do they work?
AID systems combine three components:
- An insulin pump
- A CGM device
- An algorithm (in the pump or a smartphone app) that adjusts insulin to reduce highs and lows
Current commercially available systems in the UK (January 2026):
How AID systems operate (in plain English)
- When glucose is predicted to rise above target, the algorithm tends to increase insulin (extra basal and/or an auto-correction bolus)
- When glucose is predicted to fall, the algorithm reduces or suspends insulin delivery
- You still need to enter carbohydrates eaten and bolus for meals, AID cannot yet replace meal boluses
Foundations still apply. AID is not a replacement for the basics; it is a substantial layer on top:
- Three balanced meals
- Accurate carbohydrate counting
- Bolus insulin 10 to 20 minutes before eating
How does the algorithm increase insulin?
Overnight, the system typically increases basal insulin when glucose rises above target. Later in the day, many systems will also deliver automatic corrections, for example, a small correction dose when glucose is predicted to remain high.
How does the algorithm decrease insulin?
To help prevent lows, basal insulin is reduced and may be suspended entirely when glucose is predicted to fall. Meal boluses still show as separate doses, the system is supporting your decisions, not replacing them.
What improvement tends to happen with AID systems?
- CGM alone: often around 40 to 60% time in range (TIR 3.9 to 10.0 mmol/L / 70 to 180 mg/dL)
- AID systems: commonly around 60 to 90% TIR
- Typical gain: +10 to 30 percentage points versus baseline, larger gains are often seen when starting from a lower TIR
This improvement is typically achieved with less micromanagement, not more.
Without AID, pushing beyond around 70% TIR often requires:
- Checking CGM repeatedly across the day
- Multiple correction boluses
- Frequent hypo treatments
- Ongoing trial-and-error learning
AID systems tend to let people achieve more by doing less.
The hidden superpower: sleep
One of the most consistent benefits reported is getting eight hours back. Fewer overnight hypos. Fewer wake-ups high. Flatter lines overnight, for the person with diabetes and for parents of children with Type 1.
Exploring which system may suit you
All AID systems tend to improve glucose management and quality of life. If you cannot access a specific system, the evidence suggests any of them delivers meaningful benefit.
The question is not “Which is best?” The more useful question is: Which is best for you?
Worth knowing: AID delivers a substantial uplift in TIR but it still depends on CGM performance, insulin timing, meal boluses, and the basics. If you are exploring exercise on AID, the AID and exercise key guide is a good starting point.
Find your best match
Pick the three things that matter most to you. The selector scores each system against your priorities and recommends your top two.
Before choosing, check CGM compatibility. All four UK AID systems require a CGM, and not every system works with every CGM. If you already use a CGM, or plan to, that will narrow the field straight away. The selector is a starting point for a conversation with your diabetes team, not a substitute for it.
Select 3 priorities (0 of 3 chosen)
How do the systems compare head to head?
Multiple head-to-head comparisons have now been published across all four UK-available systems. The consistent finding across all of them is the same:
No system is consistently better than another in matched populations. When baseline TIR is accounted for, no head-to-head study shows one system achieving significantly higher TIR than another.
What the evidence shows:
- Omnipod 5 vs Control-IQ (youth, n=272), no significant TIR difference at 90 days (p=0.08). Both achieved around 62 to 64% TIR from a similar baseline. The key finding was demographic: female participants, non-Hispanic Black participants, and those with public insurance were more likely to choose OP5. Gera 2025, multicentre RCT.
- Control-IQ vs MiniMed 780G (adults, n=97), treatment satisfaction equivalent (p=0.60). No TIR difference after 12 weeks. User experience and device ecosystem were more important than algorithm differences. Navas Moreno 2023, crossover RCT.
- Control-IQ vs CamAPS FX (real-world registry), no significant difference in TIR after propensity matching. Beato-Vibora 2024.
- Omnipod 5 vs MiniMed 780G (real-world registry, n=8,540), no significant TIR difference in matched populations. Khan 2026.
- DPV registry (n=10,000+, all systems), no consistent TIR advantage for any one system across age groups after matching. Baseline TIR was the strongest predictor of outcome, not system choice. Karges 2024.
What this means in practice: The factors that drive outcomes most are baseline glucose management, how well the settings are configured, and how consistently the system is used, not which brand is on the pump. Wearability preferences, CGM compatibility, lifestyle fit, and the support available through a local centre are all legitimate and evidence-backed reasons to prefer one system over another.
Explore the individual systems
Tandem t:slim X2 with Control-IQ
One of the most customisable mainstream AID systems. User settings strongly shape how the algorithm behaves, highly flexible when well set up.
Medtronic MiniMed 780G
The most aggressive system for tackling high glucose levels. Highly automated, with strong overnight control and continuous background learning.
CamAPS FX
The most adaptable system, with customisable glucose targets and continuous algorithm learning. Well-researched across all age groups, including pregnancy.
Omnipod 5
A tubeless, highly automated system with the algorithm inside the Pod itself. Discreet, simple to operate, and requiring minimal user input day to day.
Further AID resources
Use these pages to go deeper, troubleshoot, or optimise:
From the GNL Podcast
Three episodes go deeper on AID systems and the people using them.
- Episodes 1-6, AID Systems series, how to choose, how each algorithm works, and how to optimise time in range.
- Episode 30, Educating on the algorithms behind diabetes devices, Dr Inge Van Boxelaer on why AID without education fails.
- Episode 37, Dexcom G7 and the AID revolution, the CGM behind every modern AID.
Does switching to ultra-rapid insulin improve AID outcomes?
Ultra-rapid formulations (Lyumjev and Fiasp) absorb faster and clear sooner than standard rapid-acting analogues. But the largest pooled analysis to date (Rakab et al. 2025, systematic review and meta-analysis of 12 randomised controlled trials across six AID systems) found that ultra-rapid insulin produced less than 1 percentage point improvement in time in range compared with standard insulin (not statistically significant).
The AID algorithm compensates for insulin speed so effectively that the formulation barely changes overall glycaemic control. The benefit of faster insulin in AID is in reduced glycaemic variability (fewer sharp spikes and drops) and improved exercise safety margins, not in overall TIR. A simulation study of Fiasp vs NovoLog in the 670G (Grosman et al. 2021, 7,485 virtual patients) predicted a 2.2% TIR improvement, but the larger real-world evidence (Rakab 2025, 12 RCTs) shows this overstates the actual benefit; simulations assume perfect adherence that real life does not deliver.
If you are considering a switch, discuss with your diabetes care team whether variability reduction or exercise safety are priorities for you. Read more in The IOB Trade-Off.
Evidence: Rakab et al. 2025, Frontiers in Endocrinology (SR/MA of 12 RCTs); Leohr et al. 2021, Clinical Pharmacokinetics (pooled PK/PD, N=190); Pinsker et al. 2024, Diabetes Technology and Therapeutics (URLi in Control-IQ, n=179); Grosman et al. 2021, Computer Methods and Programs in Biomedicine (Fiasp vs NovoLog 670G simulation, 7,485 virtual patients).
Pairs with the CGM Guide
AID requires CGM. If you are choosing both at once, start with the five-question framework and the CGM device guides.
Pairs with the Adjunctive Therapy Guide
CGM made type 1 diabetes visible, AID automated much of it, and adjunctive therapy (GLP-1RA and GLP-1/GIP) is the third advanced lever for adults. Off-label, specialist-led, and covered end to end in the four-part guide.
Once you have the system: which TIR are you optimising for?
Two AID users can run identical 70% TIR and land at HbA1c values up to 10 mmol/mol apart, because of biological differences in how their haemoglobin glycates (the HGI signal). Two AID users on identical glucose patterns can read approximately 5 percentage points apart on different CGMs (central estimate; up to 10 at the upper bound of the head-to-head accuracy data), because of calibration zone differences. The 70% target is an ensemble figure; your personal target depends on your CGM zone and your glycator status. The personalised matrix that combines both sits in the What Gets Measured Gets Managed guide.