Cookiy AI
Cookiy AI is an agentic user research platform that recruits participants, runs AI-moderated video, voice and survey interviews, then delivers a structured report within 24 hours. Built for product, marketing, UX and agency teams.
What is Cookiy AI?
Cookiy AI presents itself as the world's first agentic AI user researcher: a platform that, in its own words, recruits, interviews, and reports back. It is published by YCD Corp., trading as Cookiy, with a declared address on El Camino Real in Palo Alto, California.
The platform covers the whole research chain. You describe a business question in natural language and answer a few clarifying prompts; a discussion guide is generated in roughly five minutes. Cookiy then sources and screens participants across several verified panels, with automated screening, identity verification and engagement quality checks. An AI moderator conducts the sessions autonomously, probing when a respondent hesitates and tracking sentiment in real time. Synthesis follows automatically, and most studies return their insights within 24 hours.
Four interview formats are offered: video interviews of 15 to 30 minutes, written surveys of 5 to 10 minutes, voice conversations of 10 to 20 minutes and live focus groups of 45 to 90 minutes. Remote in-home visits are supported at scale. The deliverables are a ten-section written report covering research background, central business question, objectives, key outcomes, respondent profile and size, field work period, executive summary, key takeaways, recommendations and appendix, plus a stakeholder-ready deck with charts and time-stamped, tagged video clips with transcripts.
The publisher claims a pool of more than 170 million respondents, over 190,000 interviews already run, 100+ interviews in a single day and about four hours from objective to insight. Those figures come from Cookiy itself and no independent source confirms them. Participants are always told they are speaking to an AI, and the terms invoke the ESOMAR codes.
Alongside the web app, Cookiy publishes an MIT-licensed Skill and CLI on GitHub (cookiy-ai/user-research-skill, around 1,500 stars when this file was compiled), compatible with Claude Code, Cowork and Desktop, Codex, Cursor and OpenClaw. Studies, participants and results are shared between both routes, and a study can use real or synthetic participants. A public Discoveries library shows real studies free of charge. Editorially, the company argues methodology over models, and says the tool is made for experts rather than as a substitute for their judgement.
What it does
- Turn a research objective written in plain language into an expert discussion guide in about five minutes
- Recruit and screen participants across several verified channels, with no agency and no diary management
- Let an autonomous AI moderator run video, voice, written survey or live focus group sessions around the clock
- Watch interviews live, flag moments and inject follow-up topics mid-session
- Receive a ten-section written report, a shareable deck and time-stamped video clips, usually within 24 hours
- Run mixed qualitative and quantitative programmes, including multilingual surveys with conditional logic
- Drive the whole pipeline from an AI agent through the open-source Cookiy Skill and CLI
When to use Cookiy AI / When not to
A quick filter to help you decide if Cookiy AI is the right fit.
When to use Cookiy AI
- Product managers who need customer evidence in days rather than weeks, without a dedicated research budget
- UX and design researchers running exploratory studies, concept tests, journey mapping or message testing across 20 to 100+ interviews
- Marketing and brand teams looking for qualitative depth on a quantitative sample, across several time zones and with no scheduling constraint
- Startups and B2B SaaS teams with no in-house research capacity
- Agencies and consultancies that need extra interview capacity, with white-label options mentioned for partnerships
When not to use Cookiy AI
- Anyone looking for a quick survey generator: the About page states plainly that Cookiy is likely not for them
- Teams hoping to automate the thinking itself; the publisher's line is that AI should amplify researchers, not replace them
- Personal or consumer use: the terms reserve the service solely for business use
- Anyone under 18, and any project involving minors, which requires a separate written agreement and parental consent
- Studies built on special categories of personal data, which the DPA says are not intentionally processed; participants are instructed not to disclose them
How to use Cookiy AI
A typical end-to-end flow, from setup to results.
- Sign up on s.cookiy.ai: the publisher advertises a two-minute setup with no credit card required
- Describe your research objective in plain language and answer a few clarifying questions
- Review the expert discussion guide generated for you, in roughly five minutes
- Define your target audience and let Cookiy source and screen participants across several channels
- Let the automated screening, identity verification and engagement quality checks run before and during the interviews
- Launch the study and let the AI moderator conduct the interviews autonomously, around the clock
- Observe sessions live if you wish: flag moments and inject follow-up topics mid-interview, or review them afterwards
- Collect the automatic synthesis, the shareable deliverables and the video clips, usually within 24 hours
- Alternative agent route: install the Cookiy Skill (Claude Code: /plugin marketplace add cookiy-ai/user-research-skill then /plugin install user-research@cookiy-ai; Codex, Cursor or OpenClaw: npx cookiy-ai), allow the s-api.cookiy.ai domain, then describe your objective or call /user-research-cookiy
- For a guided walkthrough, book a demo through the Calendly scheduler embedded on the home page
Pros & Cons
Pros
- Speed: about four hours from objective to insight, synthesised results within 24 hours and a setup announced at two minutes
- Scale: 100+ interviews in a single day, a claimed pool of 170 million+ respondents and active participants in 120 countries
- No entry cost: a free tier of 50 interviews with no credit card required
- Share-ready deliverables (report, deck, video clips) with no manual synthesis work
- Documented security posture: SOC 2 Type II, ISO 27001, GDPR and HIPAA claims, a public DPA and a published subprocessor list
- An MIT-licensed open-source building block for AI agents, so the web interface is not a dependency
- Transparency: participants are told they are speaking to an AI, and a public library of real studies is freely readable
Cons
- No public price list: cost per interview is calculated inside the platform and the site has no pricing page
- Demanding subscription terms: automatic annual renewal, four months' notice of non-renewal and a 3-5% increase each year
- Extra charges: minimum project fee, project management fee and a 500 EUR/GBP/USD cancellation fee once fieldwork has started
- No documented way to exclude customer data from the training of the publisher's models
- Hosting and subprocessors are exclusively in the United States, with no region choice and no Article 27 EU representative named
- No list of interface or interview languages published despite the multi-language promise, and no public API documentation although the terms define buy-side and supply-side APIs
- Performance figures (95% accuracy, 92% completion) are publisher claims with no independent source, and no official mobile app was found
Pricing & Plans
A free plan is available. Cookiy AI advertises a free tier of 50 interviews with a two-minute setup and no credit card required, and its own comparison tables describe the tool as “From $0 (free tier)” and “Free to start”. A 14-day free trial, cancellable at any time, is also mentioned on the product page. Beyond that, no firm price is published: the site has no pricing page, and paid usage is billed on a cost-per-interview (CPI) basis calculated by the platform for each target group, participant incentive included (terms 6.2). The product page gives orders of magnitude per study only: USD 50-200 for 1 to 10 participants, USD 200-500 for 10 to 50, USD 500-1,500 for 50 to 200, USD 1,500-5,000 for 200 to 1,000 and USD 5,000-20,000+ above 1,000 participants; startups are said to be able to run professional research for free or under USD 100 per month. Minimum project fees, project management fees, a 500 EUR/GBP/USD mid-fieldwork cancellation fee, 30-day payment terms and 12% annual late interest may also apply.
- 50 free interviews
- two-minute setup
- no credit card required
- 14 days
- cancel at any time (mentioned on the product page)
- Cost-per-interview (CPI) pricing calculated inside the platform for each target group
- participant incentive included
- no named commercial plan is published
- Managed Services
- ordered by email or telephone outside the platform
- initial term set in the order confirmation
- automatic annual renewal
- four months' notice of non-renewal and a 3-5% increase at renewal
Data, GDPR & hosting
A consolidated view of how Cookiy AI handles your data.
GDPR overview
GDPR commitments are documented rather than merely asserted. Terms section 4.2 states that Cookiy AI maintains an information security management system in accordance with ISO/IEC 27001 certification, SOC 2 compliance and GDPR compliance; SOC 2 Type II, GDPR, ISO 27001 and HIPAA badges appear on the home page. A public DPA v1.0, effective 15 July 2026, is incorporated into the terms: controller and processor roles, EU Standard Contractual Clauses (Decision (EU) 2021/914, Module Two) plus the UK ICO addendum, breach notification to the customer within 48 hours of discovery, assistance with data subject requests (Articles 12-23) and DPIAs (Articles 35-36), and audit rights met through SOC 2 reports with an annual on-site option on 30 days' notice. One gap: no Article 27 EU representative and no data protection officer are named anywhere on the site. Privacy contact: info@cookiy.com.
Who owns the data?
Section 11.5 of the terms sets out an ownership table: the survey questionnaire and associated content belong to the client, and so does the survey data collected, unless either was provided by Cookiy AI. The techniques used by Cookiy AI remain its exclusive property. The Data Processing Agreement makes the customer the controller and Cookiy the processor for participant data. When the contract ends, Cookiy deletes or returns all personal data processed on the customer's behalf within 30 days, at the customer's choice, with written confirmation on request. The privacy policy does not govern data processed on behalf of business customers: those are governed by the client contracts.
Reuse rights
The client owns the survey data collected and, under the DPA, Cookiy processes it only on the customer's documented instructions, international transfers included, so a client can reuse its own study material without asking permission. On the publisher's side, the privacy policy in force since 28 July 2026 allows Cookiy to create aggregated, de-identified or anonymised data from personal information and to share it with third parties for lawful business purposes, including improving and promoting the service and, in its own words, to train our AI models and for other machine learning purposes. The declared AI subprocessors are Google Cloud Vertex AI, OpenAI, OpenRouter and the Cloudflare AI gateway, with content described as minimised. The product FAQ states that participant information is anonymised by default and that participant data is never shared or sold. No opt-out from model training is documented anywhere, and the privacy policy and the DPA do not cover the same perimeter, so the picture remains unclear.
Data retention & training
Hosting summary
The DPA states plainly that Cookiy processes personal data primarily in the United States. AWS supplies the cloud infrastructure, storage and interview recordings, and every subprocessor on the published list is located in the United States: AWS, LiveKit, Google Vertex AI, OpenAI, OpenRouter, Cloudflare (United States and global), Firebase, Stripe, PostHog, Google Analytics and Google Workspace. No European hosting region is offered and no data residency choice is documented. Transfers out of the EEA and the UK rely on the EU Standard Contractual Clauses (Decision (EU) 2021/914) together with the UK ICO addendum. Data is encrypted at rest with AES-256 through AWS KMS and in transit with TLS 1.2 or above. One technical point worth clearing up: the domain resolves to 3.165.190.4, an Amazon address geolocated in Zurich, Switzerland. That is a CDN edge node and says nothing about where the data is stored; the DPA remains the reference. What the site does not say is which AWS regions are used, nor whether a customer can ask for storage outside the United States.
Things to keep in mind
Risks and trade-offs to weigh before adopting Cookiy AI.
- No public pricing: the real cost of a study is only known after a quote or once you are inside the platform
- Binding contract clauses: automatic annual renewal with four months' notice of non-renewal and a 3-5% yearly increase, a 500 EUR/GBP/USD mid-fieldwork cancellation fee, termination for convenience by Cookiy AI on seven days' notice, liability capped at the fees paid over the last 12 months, and a service delivered on a best commercial effort basis with no availability commitment
- Model training: the privacy policy allows anonymised data derived from personal information to be used to train the publisher's AI models, with no documented opt-out
- Identity inconsistency: the privacy policy, DPA and cookie policy name YCD Corp. d/b/a Cookiy while every footer reads © 2026 Cookiy AI Inc.; there is no legal notice page to settle it, the postal address is published without a street number and the contact emails sit on the neighbouring cookiy.com domain
- Special categories of personal data are not covered: participants are simply instructed not to disclose them, with no filtering commitment on the platform's side
- Cognitive risk: handing interviews and synthesis to an agent can quietly erode a team's own listening and analytical practice; the publisher itself warns the tool is not there to automate the thinking
- Taking the numbers at face value: accuracy, completion and panel-size claims come from the publisher alone, and an AI moderator may still miss what an experienced human interviewer would follow up on
Setup & Integrations
Technical difficulty
Low on the web route. Sign-up on s.cookiy.ai is announced at two minutes with no credit card, there is nothing to install, and recruitment, scheduling and transcription are handled by the publisher, so no tooling has to be assembled. No coding skill is required; methodological judgement to frame the objective matters more than technical ability. The agent route is a step up: install a plugin or run npx cookiy-ai, then allow the s-api.cookiy.ai domain in your agent's network settings. That stays straightforward for anyone already using Claude Code, Codex, Cursor or OpenClaw.
Deployment
Integrations
Behind Cookiy AI
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
Can an AI interviewer match a human one?
Do participants know they are speaking to an AI?
Where do the participants come from?
What kinds of research can be run?
How long does a study take?
Is there a free tier?
What security measures are in place?
Who owns the data collected?
Can Cookiy be driven from an AI agent?
Who can use the service, and where is the data hosted?
Should you pick Cookiy AI?
Cookiy AI automates the whole qualitative research chain, from recruiting participants to delivering the report. You describe an objective in plain language; the platform generates a discussion guide, sources respondents, runs AI-moderated video, voice or survey sessions and returns a ten-section report, a slide deck and tagged video clips, usually within 24 hours.
The strengths are easy to check on the site: speed, scale, share-ready deliverables, a free tier of 50 interviews, an MIT-licensed Skill and CLI for AI agents, and a documented security posture with a public DPA and a published subprocessor list. The freely readable Discoveries library of real studies is an unusual gesture of transparency.
The reservations are just as concrete. No price list is published: paid usage runs on a cost-per-interview basis computed inside the platform, and the terms attach an automatic annual renewal, four months' notice of non-renewal, a 3-5% yearly increase and a 500 EUR/GBP/USD cancellation fee once fieldwork has started. Hosting and every declared subprocessor sit in the United States, with no region choice and no Article 27 EU representative named. The privacy policy allows anonymised data derived from personal information to be used to train the publisher's models, and no opt-out is documented.
The company is young: the domain was registered on 12 August 2025, the first Wayback capture dates from 6 October 2025 and a pre-seed round of over USD 7 million was announced that same day. The legal identity is not fully settled either, since the privacy documents name YCD Corp. d/b/a Cookiy while the footer reads © 2026 Cookiy AI Inc. Every performance figure quoted comes from the publisher and is not independently verified. Worth testing on the free tier, with the contract read closely before any commitment.
- Choosing a selection results in a full page refresh.
- Opens in a new window.