Gillie.AI for Home Care
Gillie.AI for Home Care is a Finnish cloud AI platform that reads care notes, measurements and diagnoses to show changes in home care clients' functional ability and well-being, alerting care units before a deviation turns into an emergency.
What is Gillie.AI for Home Care?
Gillie.AI for Home Care is one of five products published by the Finnish company Gillie.io Company Oy, alongside Nursing Homes, Medication, Emergency Monitoring and IoT. The vendor describes it as a cloud-based healthcare artificial intelligence platform that integrates with existing healthcare and nursing information systems, and its promise is deliberately narrow: artificial intelligence shows changes in customers' functional ability and well-being.
The intended population is people receiving regular home care, which the site puts at roughly one to two per cent of the population, namely those who live in their own homes but need regular support. It pulls in the free-text entries care workers write, plus measurements, medication and diagnoses, and runs them through self-learning models. Rather than reading one value in isolation, it looks for weak signals and predicts deviations before they occur, on the reasoning that early intervention improves well-being and lowers the need for care. It follows cognition, pain, speech production and comprehension, breathing, pulse, nutrition and movement, and predicts changes in physical, cognitive and social capacity. It is compatible with the interRAI, RAVA, Omaha System and IFSIII assessment standards.
The vendor reports hundreds of AI models, a typical prediction error of two to three per cent and an average diagnostic odds ratio of about 8,000. Its home page claims 388 million observations collected, 96 per cent AI accuracy, 95,000 patients and 67 care organisations; these are the publisher's own figures, not independently audited.
Buying is a management decision at care unit level, because the system analyses the data of every client in the unit, whether that unit is a municipality, a joint municipal authority, a wellbeing services county or a private care company. There is no hardware to buy: it is a cloud service and the client mainly needs access rights. The vendor states there is no start-up fee and no separate delivery project, and recommends a four to eight week deployment.
Two limits are stated plainly. The solution has no CE mark under 2017/745/EU and is not used for the treatment or diagnosis of diseases; it is decision support for professionals. It is entered in the Valvira register of software processing social and health client data.
What it does
- Assesses each client's well-being automatically from care data, measurements and free-text nursing notes
- Raises an alert whenever a client's well-being deviates from that client's own normal pattern
- Learns what is normal for an individual in two to three weeks, so irrelevant changes stop triggering alerts
- Ranks clients by criticality, for example by pain or by severity of depression
- Highlights the risk factors associated with emergency department visits
- Assesses in real time whether a client can still cope at home
- Scores quality of life and activities of daily living, and compares care teams objectively
When to use Gillie.AI for Home Care / When not to
A quick filter to help you decide if Gillie.AI for Home Care is the right fit.
When to use Gillie.AI for Home Care
- Home care units in wellbeing services counties, municipalities and joint municipal authorities that follow ageing clients living in their own homes
- Care organisations that already record their observations electronically, which is the setting the vendor says the AI works best in
- Private care companies and unit management aiming to cut unnecessary emergency visits, ambulance call-outs and institutional care days
- Nurse supervisors and care coordinators who want to compare their care teams objectively rather than by impression
- Physicians and registered nurses who need risk factors and criticality ranking surfaced automatically across a large client base
When not to use Gillie.AI for Home Care
- Clinicians looking for a regulated medical device: the vendor states the solution has no CE mark under 2017/745/EU
- Anyone expecting a diagnostic or treatment tool, as the site says the application is not used for the treatment or diagnosis of diseases
- Home care clients themselves and their families as buyers: the vendor notes that customers themselves do not usually use the solution
- Buyers who want a self-service purchase, because there is no pricing page, no announced free trial and no sign-up route outside a Service Agreement
- Organisations unwilling to redesign care processes or fund training: the vendor warns that lack of training and support for nursing staff prevents efficient use
How to use Gillie.AI for Home Care
A typical end-to-end flow, from setup to results.
- Approach the publisher through the enquiry form or the sales address on the contact page, as there is no self-service sign-up
- Review your care unit's procedures with the supplier before deployment, because the system changes how observations are acted upon
- Plan a four to eight week deployment project covering configuration, integration, care process redesign and training
- Connect the platform to your patient information system, ERP, measurement and safety devices and alarm centre through the open interface; professionals generally do not log in separately, as the platform sits inside the systems they already use
- Use the public REST API documented at apidoc.gillie.ai to send datapoints, integrate back-end systems, embed the Gillie interface in a third-party system or query clients, datapoints and AI findings, authenticating with a public and private key pair plus a hash
- Train the organisation: the vendor budgets four to eight hours for administrators and one to two hours per group of care workers
- Decide how monitoring is organised, either through a dedicated coordination centre or by decentralised follow-up among care workers; the vendor says both work provided staff are trained
- Allow a few weeks of supervision at go-live so the AI's findings can be compared with what the care workers observe themselves
- Respond to flagged deviations within one to three days of the observation, the cadence the vendor considers necessary
- Have an expert switch off anomaly types that are not relevant for a given client, for example someone in palliative care
Pros & Cons
Pros
- An unusually complete public technology assessment: target population, mechanism, safety, clinical effectiveness, costs, organisational and legal aspects are all set out on the product page, limitations included
- Strong legal hygiene for a vendor of this size: a published data processing agreement, sub-processors named individually (Amazon Web Services, MongoDB, Nets Denmark and Zendesk), a designated data protection officer, and no transfer of personal data outside the EU or the EEA
- Detects long-term change that professionals do not always notice, opening the way to preventive measures with a real effect on quality of life
- The vendor reports fewer routine visits, fewer ambulance call-outs and unnecessary emergency visits and fewer institutional care days, keeping clients at home longer
- Security measures are described rather than implied: personal data encrypted in the database, encrypted connections to and between servers, an audit trail of access and modifications, and DDoS protection
- A public, documented REST API and an open interface, so the platform can be embedded in existing systems instead of adding another login for staff
- Measured acceptance and no hidden licensing: in one unit of 2,000 clients about 90 per cent allowed their data to be analysed and in another of 400 clients all relatives welcomed it, and the vendor says no licence or other fees are payable to third parties
Cons
- No published price of any kind: no pricing page, no tiers, no starting figure. The general terms simply refer to a price list attached to the Service Agreement, which is not public
- No free plan and no free trial are announced, and there is no self-service route; every engagement runs through a sales conversation
- The benefit is conditional on the organisation changing its care processes, and the vendor itself warns that lack of training and support for nursing staff prevents efficient use
- Results depend on how completely and how regularly observations are recorded electronically, which the supplier does not control
- False positives and false negatives both carry a cost: staff may acknowledge an alert without knowing the whole situation, and waiting for the AI to flag something may delay care
- The AI does not weight its sources differently, so a relative's observation counts as much as a professional's, and some assessments stay subjective; nutrition is the weakest model, with a diagnostic odds ratio close to 400
- The commercial centre of gravity is Finnish, from the five case studies (Etelä-Pohjanmaa, Päijät-Häme, HoviCare, Etelä-Karjala and Etelä-Savo) to the Valvira registration, and the help centre at support.gillie.ai returns a 403 to anyone not logged in
Pricing & Plans
No free plan and no free trial are announced, and the publisher discloses no price. No pricing page exists on the site, and the general terms state that the price of the Service shall be charged in accordance with the supplier's price list attached to the Service Agreement, a document that is not published. The contract does set the framework: amounts are payable in euros excluding VAT, usage is invoiced monthly in arrears with payment at fourteen days net, prices may be adjusted on sixty days' written notice, and consultancy or additional work is charged by time spent. The vendor states there is no start-up fee, no separate delivery project and no licence or other fees payable to third parties, and that material faults in the Service are corrected at no additional cost. The main outlay on the client side is internal effort, from three person-days to two person-months depending on the size of the organisation. Two other amounts appear on the site and are not prices for the tool: a median cost of 120 EUR per day for one home care client in Tampere, which is a cost borne by the care provider, and claimed savings of 1,000 to 2,000 EUR per client per year, of which more than 300 EUR is attributed to avoided emergency visits. Both are reported outcomes from client studies, not fees charged by the publisher.
Data, GDPR & hosting
A consolidated view of how Gillie.AI for Home Care handles your data.
GDPR overview
GDPR implementation is documented rather than merely claimed. The product page answers "Does the solution comply with GDPR requirements?" with "Yes", and the privacy page names the controller as Gillie.io Company Oy in Espoo, Finland, with a Data Protection Officer designated under Article 37, Jussi Peltoniemi. A data processing agreement published as a PDF appendix to the Service Agreement sets out rights of access, rectification, erasure, restriction, portability and objection, mirrored on the privacy page alongside the right to lodge a complaint with a supervisory authority and to withdraw consent. Personal data is not transferred outside the EU or the EEA. The client may audit the supplier once a year on thirty days' written notice at its own cost, and both parties must report breaches without undue delay; processor assistance is chargeable by time spent. No Article 27 representative applies: the publisher is based in the EU.
Who owns the data?
Under the published general terms, the client organisation is the data controller and the supplier is the processor, handling personal data on the client's behalf and by its order. The client keeps ownership and the intellectual property rights in its Customer Material, and obtains a perpetual, royalty-free right to use the material the Service produces. One clause deserves attention: the client grants the supplier, free of charge, a perpetual, irrevocable and unlimited right to use and exploit generalised information and data derived from that Customer Material. The product page adds that Gillie owns no customer information and cannot sell or divulge it to anyone.
Reuse rights
The platform ingests home visit notes, diagnoses, medication, laboratory results, readings from measurement and safety devices, and alarm centre records, drawn from the patient information system, the ERP, connected devices and the alarm centre. It uses them to catch weak signals of change and to predict deviations that have not happened yet. The privacy page is explicit that profiling takes place: an identifier is created so the data can be combined and compared with the profiles of other data subjects, for the stated purpose of detecting and anticipating deviations in a person's health and well-being. The same page states the file is not used for direct marketing. The processing agreement limits the supplier to the client's documented written instructions, and the general terms give the client a perpetual, royalty-free right to use the material the Service produces, so its own output can be reused without asking again. Data categories cover service users, including name, card billing details and usage history, and patients, including device measurements, patient entries, records made by the client and observations produced by the AI.
Data retention & training
Hosting summary
The publisher is consistent across three first-party sources: the product page states that the database and servers are located in Amazon's data centres in the EU, the privacy page states that all data is stored in the area of the EU, and the data processing agreement commits that the supplier does not transfer personal data outside the EU and the EEA. No individual country is ever named, so the region is as specific as the publisher gets. Four sub-processors are named in Schedule 1.1 of that agreement: Amazon Web Services for the cloud platform, MongoDB for the database, Nets Denmark for card billing and Zendesk for the support centre. The security measures described include personal data encrypted in the database, encrypted connections to the servers and between servers and database, an audit trail of access and modifications, and DDoS protection. One point of attribution matters: the ISO 27001 compliance, PCI certification and SOC reports mentioned on the privacy page are maintained by the hosting environment for its data centres, not by Gillie, which publishes no certification of its own.
Things to keep in mind
Risks and trade-offs to weigh before adopting Gillie.AI for Home Care.
- The ISO 27001 compliance, PCI certification and SOC reports quoted on the privacy page belong to the environment that hosts the services, not to Gillie. The publisher advertises no certification of its own, and the two must not be conflated when comparing suppliers
- The solution carries no CE mark and therefore sits outside the medical device regulation 2017/745/EU. Treating its output as a clinical verdict rather than as decision support puts the risk entirely on the professional who acts on it
- The general terms grant the supplier a perpetual, irrevocable and unlimited right to use and exploit generalised information and data derived from Customer Material, a right that outlives the contract and is easy to sign past
- No opt-out from model training is documented anywhere on the site or in the contractual documents, and the publisher takes no explicit position on whether client data feeds its models
- Retention is not expressed in months or years: the privacy page says data is kept until the grounds for processing have ended, measured from the last update of that person's data, so a buyer cannot derive a deletion date from what is published
- The published general terms and data processing agreement both carry the version date 20 December 2018 and the privacy page shows no effective date at all, so the legal block may have aged relative to current practice
- Alert fatigue and deskilling are the human risks: staff may acknowledge alerts without knowing the whole situation, or wait for the system to flag a change instead of forming their own judgement. The headline figures of 388 million observations, 96 per cent accuracy, 95,000 patients and 67 organisations, and the savings of 1,000 to 2,000 EUR per client per year, are vendor claims drawn from client studies rather than independent measurements
Setup & Integrations
Technical difficulty
Technically light, organisationally demanding. It is a cloud service with no hardware investment and no start-up fee: the client mainly needs access rights, and staff usually do not log in separately, the platform sitting inside the systems they already use. Integration with the patient information system, ERP, devices and alarm centre uses an open interface and a REST API. The vendor recommends a four to eight week deployment project, four to eight hours of administrator training and one to two hours per care worker group. The hard part is redesigning processes so deviations are answered within one to three days.
Deployment
Supported languages
Behind Gillie.AI for Home Care
Social
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
Who is Gillie.AI for Home Care designed for?
Is it a medical device?
What data does it collect?
Where is the data hosted?
How much does it cost?
How long does deployment take?
How accurate is the AI?
Is there an API?
Who is the data controller?
Which assessment standards and systems does it work with?
Should you pick Gillie.AI for Home Care?
Gillie.AI for Home Care is a vertical product that makes no attempt to be anything else. It addresses home care units (municipalities, joint municipal authorities, wellbeing services counties and private care companies) rather than the general public, and the buying decision belongs to unit management, because the system analyses the data of every client in the unit.
Its most unusual quality is documentary. The product page reads like a technology assessment rather than a brochure: target population, mechanism, safety, clinical effectiveness, costs, organisational and legal aspects are all set out in public, limitations included. The legal hygiene matches it: a data processing agreement published in full, sub-processors named one by one, a designated data protection officer, and a commitment that personal data does not leave the EU or the EEA. Few vendors of this size publish as much.
The reservations are just as clear. There is no pricing transparency at all: no page, no tiers, no starting figure, only a price list annexed to the contract, so no buyer can size the investment without talking to sales. The commercial footprint is essentially Finnish, from the five case studies to the Valvira registration. And the value is conditional on organisational change: the vendor says so itself, warning that without training and support the system will not be used efficiently, and that flagged deviations need an answer within one to three days.
Finally, what it is not. This is not a medical device, carries no CE mark under 2017/745/EU, and is not used to treat or diagnose; it is decision support that leaves the judgement with the professional. The headline numbers on the home page are the publisher's own claims rather than audited measurements. Read on those terms, it is a serious and well-documented tool for a narrow, demanding market.
- Choosing a selection results in a full page refresh.
- Opens in a new window.