elvex
Enterprise platform for building, sharing and governing AI agents. It lets any employee assemble no-code agents on any model, wire them to company data and tools, and gives IT full audit, permission and spend control.
What is elvex?
elvex is an enterprise platform for building, sharing and controlling AI agents, published by Elvex, Inc. in New York. Its pitch is deliberately narrow: not another chat window, but a company-wide layer that makes generative AI usable by every employee while giving the business the governance it needs. The three claims repeated across the site are any model, any integration, any employee.
Model neutrality is the foundation. Teams select Gemini, Claude, GPT, Llama, Cohere or custom models, reaching them directly or through Amazon Bedrock and Azure OpenAI, and switching model is a configuration change rather than a migration, because agents, context and workflows sit on elvex rather than on any one provider. The same neutrality applies to software: with no competing productivity suite to protect, the vendor treats every connector as first class, from Salesforce and Slack to Snowflake, Notion and Jira.
The second pillar is organisational context. elvex builds a multi-layer picture of the company, its roles, agents, datasources, shared Spaces and ongoing projects, so that a correction made by one person sticks for everyone. On top of that sits a no-code agent builder aimed at business teams, an Agent Library of prebuilt agents sorted by function, and Flows for multi-step processes with validation between steps, branching, loops and a different model per step if needed.
Governance is built in rather than bolted on: custom roles, control over which models each workspace or user may reach, complete audit logs, automated removal of PII, PHI and financial data, and analytics covering adoption, feedback and token-level cost. The vendor holds SOC 2 Type II and HIPAA reports, available on request through a public trust center.
Human services come with the product: a dedicated success partner, support in Slack, Teams and in-app chat, tailored training and help building the ROI case. The company was founded in 2023 by Sachin Kamdar, Mike Sukmanowsky and John Levitt, the team behind Parse.ly, which Automattic acquired in 2021. Customers named on the site include The Boston Globe, McClatchy, Movable Ink, Embark, Stacker and 1Kosmos.
What it does
- Pick and swap the underlying model, from OpenAI, Anthropic, Google, Cohere, AI21 or open-weight options, without rebuilding agents
- Build an agent in plain language, with editable rules, business context and full version history
- Deploy ready-made agents from the Agent Library, then tailor them to your own data and workflows
- Connect live data sources: SQL warehouses, documents, web content and the SaaS tools already in use
- Orchestrate multi-step processes with Flows, including validation between steps, branching and loops
- Share agents and datasources through team Spaces with role, team and individual permissions
- Track adoption, audit every action and monitor spend down to token cost, with per-model limits
When to use elvex / When not to
A quick filter to help you decide if elvex is the right fit.
When to use elvex
- Large organisations rolling AI out to every department rather than to a handful of power users
- IT and security teams that must curb shadow AI while still letting staff work with generative models
- Companies unwilling to lock their AI strategy to a single model vendor or productivity suite
- Non-technical business teams in marketing, sales, HR, support, finance, legal and operations
- Regulated or compliance-heavy sectors such as insurance, financial services, legal, healthcare and publishing
When not to use elvex
- Individuals and freelancers: there is no personal plan, and access starts with a sales demo and a signed order form
- Buyers who need to compare list prices before talking to anyone, since no rate is published anywhere on the site
- Teams that want to trial a product themselves first: no free plan and no self-serve free trial are advertised
- Organisations requiring European data residency, as customer data is stored exclusively in the United States
- Anyone expecting mobile or offline use: no iOS app, no Android app and no browser extension exist
How to use elvex
A typical end-to-end flow, from setup to results.
- Request a demo through the Get a demo form, which is the only entry point to the product
- Take the onboarding call, where the team builds a working agent live against your own workflow
- Leave the call with a scoped proof of concept, a timeline and the elements needed to decide
- Sign an order form, since fees and terms are set there rather than on a public price list
- Get provisioned and start with the prebuilt assistants that are available immediately, with no technical setup
- Connect your tools through OAuth using each user's own credentials, which preserves source-system permissions
- Attach datasources: documents, web content, SaaS tools and SQL warehouses, kept in live sync
- Build or clone an agent in plain language, writing its rules and business context, then version it
- Chain agents into Flows when a task needs several steps, with human approval on writes to external systems
- Configure governance: roles, permitted models, audit logging and spend limits, then extend through the API or MCP servers
Pros & Cons
Pros
- Genuine model neutrality: switching model is a configuration change, so a better release does not mean rebuilding agents
- Governance designed in from the start, with role, team and individual permissions, audit logs and per-workspace model control
- SOC 2 Type II and HIPAA reports available on request through a public trust center
- Subprocessors published by name, role and country, which is rare enough to be worth noting
- Written commitment that customer data is never used to train models, backed by the SOC 2 System Description
- Usage-based billing rather than per-seat licensing, so rolling the tool out widely does not scale cost with headcount
- Human services included: a dedicated success partner, Slack and Teams support, tailored training and ROI work
Cons
- No public pricing at all: the cost only becomes visible after going through the sales process
- No free plan and no self-serve free trial; the only free element is a ten-minute agent-building session
- Purchase runs through a signed order form, which makes for a long cycle unsuited to small organisations
- Customer data is stored exclusively in the United States, with no European residency option announced
- The Terms of Use restrict the website to persons located in the United States
- No mobile app, no browser extension and no published list of interface languages
- The vendor keeps a perpetual, irrevocable licence over customer data folded into its aggregated statistics, and termination gives no right to a refund
Pricing & Plans
No price is published. The vendor states that it bills on usage rather than per seat, but the site carries no rate card, and the /pricing page serves an onboarding sequence instead of a tariff. Fees are set in the order form signed between the parties, payable in US dollars and exclusive of taxes. Neither a permanent free plan nor a free trial is advertised, and the Terms of Sale specify that termination or expiry gives no right to a refund. The only figures quoted on the site, around 30 USD per user per month, describe competitors rather than elvex.
Data, GDPR & hosting
A consolidated view of how elvex handles your data.
GDPR overview
GDPR handling is concrete and documented. A public Data Processing Addendum names Elvex, Inc. as processor, lists the GDPR and the CCPA among the applicable data protection laws, and incorporates the European Commission's Standard Contractual Clauses of 4 June 2021 (Implementing Decision 2021/914), module 2 for customer-to-vendor transfers and module 4 for the reverse. A second schedule adds the UK Addendum issued by the ICO. The vendor commits to Article 32(1) security measures, to assisting with Article 35 and 36 impact assessments, to notifying personal data breaches without undue delay, and to giving prior notice before adding or replacing a subprocessor. The DPA states that a data protection officer has been appointed, without naming them. No Article 27 EU representative is designated. Data subject rights are exercised through privacy@elvex.ai; the policy was last modified on 22 January 2026.
Who owns the data?
The Terms of Sale draw a clean line: elvex owns the Cloud Services, and the customer owns all right, title and interest in Customer Data. In exchange the customer grants elvex a non-exclusive, royalty-free, worldwide licence to reproduce, distribute, use and display that data strictly as needed to deliver the service. One clause reaches further: the licence covering Customer Data incorporated into the vendor's Aggregated Statistics is perpetual and irrevocable. The customer also warrants it holds the rights to the data it uploads. Under the Data Processing Addendum, elvex acts as processor on the customer's documented instructions, and returns or deletes the data on request when the contract ends.
Reuse rights
Customers keep and may freely reuse their own data and the outputs they generate; nothing in the Terms of Sale requires the vendor's permission for that. On the vendor's side, use is bounded. The Security page states that customer data is never used to train AI models, and that prompts and content sent to the integrated model providers are processed for that request only, never retained or used for training, a commitment documented in the SOC 2 System Description. The Privacy Policy adds that elvex does not permit GenAI providers to train their general-purpose models on customer data unless the customer has expressly enabled that in its own agreement with the provider. Account details such as username and email are not shared with model providers, though the customer's API key is passed to establish the connection. elvex does keep interaction logs and metadata for service delivery, security, abuse prevention, legal compliance and support, and reserves a perpetual licence over customer data folded into its aggregated statistics.
Data retention & training
Hosting summary
Jurisdiction is the United States, without ambiguity. The Data Processing Addendum states that all customer data is permanently stored in the USA and backed up for disaster recovery, and that services and data are hosted in Amazon Web Services facilities in the USA. The Security page details encryption at rest with AES-256 and KMS, enforced SSL and TLS in transit, daily database backups, multiple availability zones and recovery objectives of one hour RPO and four hours RTO. Subprocessors are published by name, role and country: AWS, Cloudflare, the model providers OpenAI, Anthropic and Cohere, Unified, Daytona, Twilio and SendGrid, Auth0 by Okta, Google Workspace and Vercel, all in the USA, plus Sentry in Germany and the USA. The same list is mirrored on a Notion page referenced by the DPA. No European or other regional hosting option is announced anywhere on the site.
Where elvex works
Country-level availability.
Available in
Not available in
Things to keep in mind
Risks and trade-offs to weigh before adopting elvex.
- Aggregated statistics carry a perpetual, irrevocable licence over the customer data folded into them: read that clause before signing
- The no-training promise partly depends on the customer's own contract with its GenAI provider, which the customer could itself loosen
- Binding arbitration and a class action waiver apply; opting out means emailing support@elvex.ai within 30 days of acceptance, under Delaware law
- Official addresses sit on the elvex.ai domain rather than elvex.com, a mismatch worth checking before trusting an email that claims to come from the vendor
- Making agent building effortless also makes it easy to industrialise a flawed prompt: the version history and human diff review exist for a reason, and skipping them is tempting
- Proactive, context-aware guidance is convenient, but a workforce that stops questioning suggested workflows loses exactly the judgement the tool was meant to amplify
- The vendor may publicly name the customer and use its logo unless the customer objects in writing, and termination gives no right to a refund
Setup & Integrations
Technical difficulty
Low for the business side, moderate for IT. The vendor states that no technical setup is required and that prebuilt assistants are available as soon as an account is provisioned, with most teams operational within a day and an enterprise-wide rollout with full governance configuration taking one to two weeks. Agents are written in plain language, and tools connect through OAuth with each user's own credentials. The heavier work is optional and falls to IT: SSO and role design, data warehouse connections, crawler configuration, API use and MCP servers. Solutions engineers, training and debugging support are included.
Deployment
Integrations
Behind elvex
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Alternatives
Tools that compete with or complement elvex.
Frequently asked questions
Is elvex available to individuals or small teams?
Can we choose which AI models our teams use?
Is our data used to train AI models?
What compliance certifications does elvex hold?
Where is customer data hosted?
Is a Data Processing Addendum available?
Does elvex offer an API?
How long does it take to get started?
How much does elvex cost?
Is there a mobile application?
Should you pick elvex?
elvex is one of the more coherent answers to a problem most large organisations now recognise: giving every employee useful AI without losing sight of what is happening. Its two structural choices, model neutrality and usage-based billing, reinforce each other. Because agents and context live on the platform rather than on a provider, a better model is a configuration change; because billing follows usage rather than seats, spreading the tool across a workforce does not scale cost with headcount. Governance is not an afterthought either: permissions, audit logs, model controls and token-level spend tracking are part of the product, and the vendor publishes its subprocessors by name and country, holds SOC 2 Type II and HIPAA reports, and puts a full DPA online with the EU Standard Contractual Clauses and the UK Addendum.
The reservations are as clear as the strengths. Nothing about pricing is public, and there is no free plan or self-serve trial, so evaluating elvex on your own is simply not possible: the road runs through a sales demo and a signed order form. Customer data is stored only in the United States, which a European buyer will want to weigh, and the Terms of Use formally restrict the site to persons located in the United States. A perpetual licence over customer data folded into aggregated statistics deserves a read before signing.
The company is young, founded in 2023, but the founding team previously built and sold Parse.ly to Automattic, and the customer names on the site are real and checkable. For an organisation that has already decided to standardise its AI usage and wants control alongside adoption, elvex belongs on the shortlist. For anyone hoping to try before talking to a salesperson, it does not.
- Choosing a selection results in a full page refresh.
- Opens in a new window.