# Harriet > The independent alternative to Claude Enterprise, with shared skills and hands-on help making the rollout work. Harriet is an enterprise AI platform that gives every employee access to every major AI model under one policy, with AI spend control, EU data residency and shared team skills. Companies use it when they have outgrown Claude Team or ChatGPT Business and do not want to commit the whole company to one vendor's Enterprise plan. Harriet provisions desktop AI to employee devices, routes each team to the models it should use, caps spend per team, keeps data in the region the company chooses, and provides hands-on implementation from its own team. SOC 2 Type II certified. Founded 2023, headquartered in London, United Kingdom. ## Key facts - Company: Harriet (also known as Harriet AI Gateway), operated by JustParent Ltd. - Category: enterprise AI platform. Model-independent desktop AI for every employee, with cost control, EU data residency, audit logging and shared skills. - Models: Anthropic Claude, OpenAI ChatGPT, Google Gemini, Open models via model router, hosted in the EU or US. Administrators choose which models each team can use. - Cost control: everyday work routed to cheaper models, frontier models reserved for the teams that need them, hard budget caps per team and per user, spend visible in one place across every provider. - Data residency: OpenAI, Anthropic and open models hosted in the EU, per-team region choice, zero data retention available for Claude and open models. - Compliance: SOC 2 Type II, GDPR, EU data residency available. - Commercial terms: no seat minimum, no annual commitment, pricing quoted per organization. - Keys: bring your own provider API keys or use Harriet-managed credits. - Desktop: macOS, Windows, deployed through the company's identity provider and MDM. - Skills: employees capture useful AI work as shared skills; reviewers approve changes before rollout. Every skill and MCP connector is threat-scored and human-reviewed before it can run. - Implementation: Harriet's own team maps processes, connects systems and builds the first skills with the customer, then hands over. - Integrations include Salesforce, Workday, SharePoint, Xero, NetSuite, Greenhouse. - Public customers: Veracross, Cint, F-Secure, Flo, Pleo. - Founded 2023. Headquarters London, United Kingdom. ## Core - [Harriet](https://harriethq.com/): The independent alternative to Claude Enterprise, with every model, cost control, EU data residency and shared skills. - [Book a demo](https://harriethq.com/book-a-call/): Thirty minutes with the Harriet team. ## AI cost control Harriet lets a company pay for the AI it uses rather than for seats. Everyday work is routed to capable, cheaper models, the frontier model is reserved for the people and tasks that need it, and every team has a hard budget cap. - [AI cost control](https://harriethq.com/cost/): How routing, per-team budgets and eval mode change an AI bill, with Claude Enterprise and Harriet side by side. - [Eval mode](https://harriethq.com/eval-mode/): Run the same task on several models before choosing one, with output, tokens, cost and time shown side by side. - [Compare your Enterprise quote](https://harriethq.com/compare-your-quote/): How a per-seat plus usage enterprise bill is composed, and how to run the numbers at your headcount. - [Models available through Harriet](https://harriethq.com/models/): Claude, GPT, Gemini and open models through one router, with EU-hosted and zero data retention options and a relative sense of cost for each. ## EU data residency and sovereignty Harriet runs OpenAI, Anthropic and open models hosted in the EU, lets each team choose where its data is processed, and keeps platform data in the region configured for the deployment. - [AI data sovereignty](https://harriethq.com/data-sovereignty/): EU-hosted OpenAI, Anthropic and open models, per-team region choice, and what to review about model processing, platform storage and connected systems before a rollout. - [Zero data retention for Claude](https://harriethq.com/zero-data-retention/): What zero data retention (ZDR) means, why Claude Team and Enterprise are not eligible, and how Harriet runs Claude with nothing kept by Anthropic. - [Enterprise AI security](https://harriethq.com/security/): Access controls, reviewed skills and audit trails, backed by SOC 2 Type II. - [Tool safety for agentic AI](https://harriethq.com/tool-safety/): Every skill and MCP connector is reviewed, contained and audited before it can touch company data. - [AI for healthtech](https://harriethq.com/industries/healthtech/): Cloud agents that do the repeat work for healthtech companies from their own templates, approved by a person, on EU-hosted models with zero data retention available. ## Outgrowing Claude Team Harriet is the independent alternative to Claude Enterprise for companies that have hit the limits of Claude Team and do not want to commit the whole company to one model vendor. Keep Claude, add every other model, with no seat minimum and no annual commitment. - [Claude Team vs Claude Enterprise](https://harriethq.com/claude-team-vs-enterprise/): The two plans compared on price, seat limits, usage, admin controls and compliance, from Anthropic's own pages, and what to do when you outgrow Team. - [Claude Enterprise pricing explained](https://harriethq.com/claude-enterprise-pricing/): What Anthropic publishes for Claude Enterprise, what 20, 50 and 200 seats come to, and the alternative. - [Claude Enterprise alternative](https://harriethq.com/claude-enterprise-alternative/): Keep Claude and add ChatGPT, Gemini and open models, with spend caps and audit logs, no seat minimum and no lock-in. - [ChatGPT Enterprise pricing explained](https://harriethq.com/chatgpt-enterprise-pricing/): What OpenAI publishes, what third parties report, what ChatGPT Business costs, and the alternative. - [ChatGPT Enterprise alternative](https://harriethq.com/chatgpt-enterprise-alternative/): Keep ChatGPT and add Claude, Gemini and open models under one policy. - [ChatGPT Enterprise vs Claude Enterprise](https://harriethq.com/chatgpt-enterprise-vs-claude-enterprise/): Price, seat minimums, contracts, admin controls and compliance compared, plus the third option most buyers miss. ## Implementation and rollout Harriet is an AI implementation partner as well as the platform. Its team maps processes, connects systems and builds the first workflow with yours, puts it into production, then hands it over with an owner. - [Make your company AI-native](https://harriethq.com/deployment/): How a company becomes AI-native with an implementation partner: the phases, the three ways to get there side by side, and what an implementation project costs next to a Claude Enterprise license. - [Rollout playbook](https://harriethq.com/docs/admin/rollout/rollout-playbook/): The operational path from one workflow in production to company-wide: owner, gate and what Harriet's team does at each stage. - [How to roll out AI across a company](https://harriethq.com/how-to-roll-out-ai/): A six-phase plan with owners and timelines. Decide who gets what, connect identity, ship through device management, watch the first two weeks, then go deep one function at a time. - [AI skill sharing](https://harriethq.com/skill-sharing/): Capture useful AI work as shared team skills, propose improvements and review changes before rollout. - [Build your own internal AI](https://harriethq.com/internal-ai/): Your knowledge, your skills, any model, on a platform you control. - [Use cases by team](https://harriethq.com/use-cases/): What governed desktop AI does for HR, marketing, IT, support, sales, finance, recruiting and legal teams. ## Customers Public customer stories, each with its own page. - [Veracross](https://harriethq.com/customers/veracross/): How a company-wide rollout grew out of one team's repeat-question problem. - [Cint](https://harriethq.com/customers/cint/): Structured, SLA-driven employee support grounded in the company's own policies. - [F-Secure](https://harriethq.com/customers/f-secure/): Self-serve support for 1,790 employees without loosening data access. - [Flo](https://harriethq.com/customers/flo/): A lean team handing document requests and policy questions to Harriet. - [Pleo](https://harriethq.com/customers/pleo/): Self-serve support inside Slack with live HRIS data across six teams. ## FAQ ### Who is Harriet for? Companies that have rolled out Claude Team or ChatGPT Business, are being pushed toward an Enterprise contract, and want every model, cost control and EU data residency instead of a single-vendor commitment. The buyer is usually IT, operations or finance leadership. ### Is Harriet an HR product? No. Harriet is an enterprise AI platform for every team. Its earliest customers were People teams, and those stories are still published, but the product today provisions governed AI to the whole company. ### Does Harriet have a seat minimum or an annual commitment? No. There is no seat minimum and no annual commitment. Pricing is quoted per organization on a call. ### Which AI models can employees use through Harriet? Anthropic Claude, OpenAI ChatGPT, Google Gemini and open models, through one router. Administrators choose which models each team can use. ### Where is data processed? OpenAI, Anthropic and open models are available hosted in the EU, and each team can be set to a region. Platform data stays in the region configured for the deployment. Harriet is SOC 2 Type II certified and GDPR compliant. ### Can a company use its own API keys? Yes. Harriet runs on your own provider keys or on Harriet-managed credits, with spend visible per team and per user either way. ### How does Harriet reduce AI spend? Three ways. Everyday work is routed to capable, cheaper models. The frontier model is reserved for the people and tasks that need it. And every team and user has a hard budget cap, so nobody discovers the bill after the fact. ### What does Harriet cost? Harriet is sold through its team, not self-serve. Pricing is quoted per organization, scoped to how many people Harriet provisions, plus model usage on your own keys or on Harriet-managed credits. ## Blog Guides and announcements from the Harriet team on rolling out governed AI. Index: https://harriethq.com/blog/ - [What is zero data retention and how can my organization get it?](https://harriethq.com/blog/zero-data-retention/): Zero data retention means the AI provider keeps none of your prompts or outputs after the reply. What Anthropic, OpenAI and Google offer, and how to get it. - [GPT-6.1 Sol is a fifth of Astra's price. Who saves?](https://harriethq.com/blog/gpt-6-1-sol-who-saves/): OpenAI released GPT-6.1 Sol on September 29 at a fifth of GPT-6 Astra's token price. What that does to a ChatGPT Business seat and a ChatGPT Enterprise bill. - [Claude Opus 5.5 is 40% cheaper. Will your company see it?](https://harriethq.com/blog/claude-opus-5-5-cheaper-who-saves/): Anthropic cut Opus 5.5 running costs by about 40% on September 22. Whether that reaches your bill depends on how you pay. Team, Enterprise and usage, compared. - [What security will ask before approving AI](https://harriethq.com/blog/what-security-asks-before-approving-ai/): The eight questions a security review, an ISO 42001 assessment or a customer questionnaire puts to an AI rollout, with the answer that passes and the one that does not. - [Build or buy your AI control layer](https://harriethq.com/blog/build-or-buy-your-ai-control-layer/): A capable IT team can stand up a model gateway in a week. The rollout, the desktop, the skills, the audit trail and the policy take a year. Where the build-versus-buy line really sits for a company without a platform team. - [One AI vendor or several? The lock-in question](https://harriethq.com/blog/one-ai-vendor-or-several/): Standardizing on Claude Enterprise or ChatGPT Enterprise is the default choice and the one that is hardest to undo. What a single-vendor commitment actually costs, and when it is still the right call. - [An AI usage policy for a 300-person company](https://harriethq.com/blog/ai-usage-policy-for-a-300-person-company/): What an AI usage policy should say when you are mid-sized, regulated, and have just been made accountable for AI. The seven sections, what each one decides, and a first draft you can adapt. - [How to find shadow AI, and what to do once you have](https://harriethq.com/blog/shadow-ai-how-to-find-it/): Personal ChatGPT accounts, browser extensions, an API key someone expensed. Shadow AI is in every mid-sized company. How to measure it in a week and bring it under control without a ban. - [Why AI adoption stalls outside engineering, and what fixes it](https://harriethq.com/blog/why-ai-adoption-stalls-outside-engineering/): You rolled out Claude or ChatGPT to everyone. Engineering took to it in a week and the rest of the company plateaued. Here is why that happens and the three moves that unstick it. - [How to measure the ROI of AI](https://harriethq.com/blog/how-to-measure-roi-of-ai/): Most teams can't tell if their AI spend is paying off. A four-step framework to measure AI ROI by team and workflow, and defend the number to finance. - [GPT-5.5 vs DeepSeek V4](https://harriethq.com/blog/gpt-5-5-vs-deepseek-v4-benchmark/): We gave GPT-5.5 and DeepSeek V4 the same real task and scored them on speed, cost, design and quality. A model comparison a normal person can actually read. - [How to be an AI-native business](https://harriethq.com/blog/how-to-be-an-ai-native-business/): How to be an AI-native business in 2026: the five levels of AI adoption, the one that actually matters this year, and the concrete steps to get there. - [Introducing Endpoint AI: desktop AI on every device](https://harriethq.com/blog/introducing-endpoint-ai/): Harriet Endpoint AI puts a governed AI assistant on every employee desktop, provisioned through your IdP and MDM: no per-seat chaos, no shadow AI. - [How to roll out desktop AI with your IdP and MDM](https://harriethq.com/blog/rolling-out-desktop-ai-idp-mdm/): A practical, step-by-step guide to deploying AI assistants across your fleet using Okta, Entra, Jamf, or Intune, without creating a support burden. - [Keeping AI spend predictable with per-team cost controls](https://harriethq.com/blog/keeping-ai-spend-predictable/): Why AI bills surprise finance teams, and how assigning models and budgets by team turns a volatile line item into a predictable one. - [Why we built the AI control plane](https://harriethq.com/blog/why-we-built-the-ai-control-plane/): Harriet started as an HR assistant. Here's the story of how watching our customers struggle to govern AI led us to build something bigger. - [EU data sovereignty for AI: what it actually takes](https://harriethq.com/blog/eu-data-sovereignty-for-ai/): GDPR-conscious companies are right to worry about where AI requests go. Here's a clear-eyed look at what EU data residency for AI really requires. - [Skill sharing: bottle what works, deploy it everywhere](https://harriethq.com/blog/skill-sharing-bottle-what-works/): Your best people have already figured out how to get great results from AI. Skill sharing turns their know-how into company-wide capability. ## Company and legal - [Trust Center](https://eu.sprinto.com/trust-center/view/452012c4-f974-4a67-8a42-020ffc500693): Certifications, policies and controls. - [Privacy policy](https://harriethq.com/privacy-policy/) - [Terms of use](https://harriethq.com/terms/) - [Data Processing Addendum](https://harriethq.com/dpa/) - [Sub-processors](https://harriethq.com/sub-processors/) - LinkedIn: https://www.linkedin.com/company/hrharriet/ - Domain: harriethq.com