SimplAI
SimplAI is an enterprise agentic AI operating system for building, orchestrating, evaluating and governing production-grade AI agents. It runs across managed cloud, private VPC, on-premises, sovereign and fully air-gapped environments, with role-based governance and step-level execution tracing.
What is SimplAI?
SimplAI describes itself as an operating system for agentic AI, built by Datafuse Technology Inc., a Delaware corporation with offices listed in San Francisco, Dubai and Gurugram. The pitch is narrow and consistent: take AI pilots that stalled at the demo stage and turn them into governed, observable enterprise operations.
The platform rests on five components, an Agent Builder, a Workflow Builder, knowledge bases, evaluation and tracing. Agents are assembled on a canvas: you pick a model, write instructions, attach knowledge and tools, and apply guardrails. Eleven capabilities can then be switched on individually, among them persistent memory, citations, tool citations, file upload, browser use, artifacts, voice, and a reflection step that makes the agent check its own work before answering. Memory is backed by a knowledge graph, so agents reason over entities and relationships rather than isolated document chunks.
Model choice is deliberately open. Anthropic Claude, OpenAI, Google Gemini, Cohere, Mixtral, Ollama and TogetherAI are supported for generation, with OpenAI, Cohere, Gemini and Voyage for embeddings, and bring-your-own-key throughout. Enterprise customers can bring their own models and fine-tune on text, vision or audio.
Around the builder sits the part aimed squarely at large organisations. Governance covers RBAC and ABAC, isolation between development, staging and production, audit trails at both execution and decision level, and spend or rate limits enforced per agent at runtime. Security adds runtime isolation between agents and tenants, a native secrets vault, fine-grained control over which tools an agent may reach, and private networking. An MCP Gateway exposes internal APIs as schema-driven agent tools without backend rewrites, while agent-to-agent communication handles delegation across systems.
Deployment is the real differentiator. The same agents run on SimplAI's managed cloud, in a customer VPC, on-premises, in sovereign cloud, or in fully air-gapped networks with no internet connectivity. SimplAI advertises a 99.9% uptime SLA, point-in-time recovery, cross-region replication and zero-downtime rolling updates. A library of ready-made vertical agents covers claims intake, fraud detection, underwriting, credit analysis, financial spreading, accounts payable and mortgage workflows.
What it does
- Build modular AI agents with defined roles, memory, tools and guardrails
- Orchestrate multi-agent, multi-step workflows with planning, routing and delegation
- Ground agents in enterprise knowledge bases with parsing, embedding and re-ranking
- Evaluate agents against test datasets and catch regressions before promoting a version
- Trace and audit every execution at step and decision level
- Expose internal APIs and services as governed agent tools through the MCP Gateway
- Deploy the same agents to managed cloud, private VPC, on-premises or air-gapped environments
When to use SimplAI / When not to
A quick filter to help you decide if SimplAI is the right fit.
When to use SimplAI
- Regulated enterprises in banking, insurance, healthcare or government whose data cannot leave their own perimeter
- AI and platform teams juggling several disconnected LLM pilots who want one place to build, evaluate and govern agents
- Organisations with no in-house AI engineering team that still need production-grade agents
- Insurance and lending operations teams after ready-made agents for claims intake, underwriting, fraud detection and credit analysis
- Compliance, risk and governance functions that need decision-level audit trails and per-agent spend controls
When not to use SimplAI
- Individuals looking for a personal assistant, since the terms restrict SimplAI to business use only
- Small teams needing a shared workspace, because the Free and Starter tiers are both single-user
- Buyers who want published pricing for enterprise capabilities, all of which sit behind a custom quote
- Mobile-first users, as there is no iOS or Android application
- Direct competitors of SimplAI, who are contractually barred from accessing the platform without written consent
How to use SimplAI
A typical end-to-end flow, from setup to results.
- Create an account on the SimplAI app: the Free tier needs no credit card and supports Google or Microsoft single sign-on
- Define your agent on the builder canvas by picking a model, writing instructions, attaching knowledge and tools, and setting guardrails
- Switch on the capabilities the use case actually needs, such as memory, citations, file upload, browser use, voice or reflection
- Connect your enterprise data sources so the agent is grounded in your own context rather than generic knowledge
- Chain agents into a multi-step workflow in the Workflow Builder when a single agent is not enough
- Run test datasets in the playground and review evaluation scores before promoting anything
- Publish a version of the application, since the Deployment tab stays locked until at least one published version exists
- Open the Deployment tab and click Enable integration, which may count against your plan's limit on integrable applications
- Wire the agent into your stack through the REST API, the Python SDK, webhooks, the embed script tag, or agent-to-agent invocation
- Watch run history and traces in real time, then iterate: continuous evaluation flags regressions and version control lets you roll back
Pros & Cons
Pros
- Covers the whole agent lifecycle, from building and evaluation through deployment, observability and governance, on one platform
- Unusually wide deployment range, reaching on-premises, sovereign and fully air-gapped environments
- On-premises mode commits explicitly to no external API calls, no model training on your data and no cloud egress
- Genuine model neutrality across Anthropic, OpenAI, Google Gemini, Cohere, Mixtral, Ollama and TogetherAI, plus bring-your-own-key
- Substantial public documentation, with the REST API and Python SDK spelled out including working curl examples
- Free tier is usable without a credit card and already includes Google and Microsoft single sign-on
- Concrete operational commitments: 99.9% uptime SLA, point-in-time recovery, canary releases and instant rollback
Cons
- The terms grant a very broad, irrevocable and sublicensable licence over Customer Data, extending to improving SimplAI's other offerings
- No training opt-out is documented for cloud deployments; the no-training commitment covers on-premises only
- No subprocessor list, no Article 27 EU representative and no named data protection officer, despite an advertised GDPR posture
- The pricing grid is inconsistent: the Enterprises card refers to a Scale-up tier that does not exist on the page
- Free and Starter are both single-user, so any team collaboration requires an enterprise quote
- Total liability is capped at 100 USD and no refunds are issued on any plan, monthly or annual
- No hosting country or region is published for SimplAI's own cloud, and there is no mobile application
Pricing & Plans
SimplAI operates a freemium model. A free tier is available and requires no credit card, providing 5,000 credits and 50 runs per day. The lowest published paid price point is the Starter plan at 99 USD per month, billed monthly, which raises the allowance to 40,000 credits per month and 100 runs per day. Enterprise pricing is not published and is quoted on request. No refunds are granted on any plan once a subscription fee has been charged.
- up to 7 applications
- single-user access
- 100 MB of dataset storage
- 50+ pre-built apps
- tracing for debugging
- basic admin controls
- version control
- and single sign-on with Google and Microsoft
- everything in the free tier plus up to 14 applications
- single-user access
- 500 MB of dataset storage
- integration through SDK
- webhooks and embed code
- access to Voice AI
- basic admin controls
- and email support
- flexible deployment on customer cloud or on-premises
- advanced compliance (SOC 2 Type 2
- ISO) and data isolation
- bring your own AI models
- model fine-tuning for text
- vision and audio
- custom contracts including BAA and DPA
- observability
Data, GDPR & hosting
A consolidated view of how SimplAI handles your data.
GDPR overview
GDPR implementation is documented but incomplete. The privacy policy names four lawful bases, namely consent, contractual necessity, legal obligation and legitimate interest, and enumerates eight data subject rights including erasure, portability, objection and the right not to be subject to automated decision-making. International transfers rely on appropriate safeguards, explicitly the European Commission's standard contractual clauses. The cookie policy defines applicable law as including the EU GDPR and requires consent for anything beyond strictly necessary cookies, while the deployments page carries a GDPR badge alongside SOC 2 Type II, ISO 27001 and HIPAA. Three gaps stand out: no Article 27 EU representative is designated, no data protection officer is named, and no subprocessor list is published even though the policy admits using third parties. All requests go to a single address, legal@simplai.ai.
Who owns the data?
Under section 10 of the terms you keep your Customer Data, but you grant Datafuse Technology Inc. a nonexclusive, worldwide, royalty-free, irrevocable, sublicensable and fully paid-up right to access, collect, use, process, store, disclose and transmit anything you upload. That right serves three stated purposes: running the platform, exercising the contract, and improving SimplAI's platform and other offerings. The platform itself, its source code and all supplied materials remain SimplAI's property and are licensed, never sold. Feedback you send is explicitly not treated as confidential. On-premises deployment reverses the picture: execution stays on your own hardware, with no external API calls and no cloud egress.
Reuse rights
The terms grant end users no reuse rights over SimplAI's own materials. Section 11 forbids modifying, copying, reproducing, publishing, transmitting, distributing, decompiling or reselling the platform, its software or its content, and bars granting access to any third party. Section 13 additionally forbids using the platform to build competing products or services, to mine data, or to reverse engineer it. You may only upload data you hold all required rights to, and you remain responsible for export-control and sanctions compliance. In the opposite direction, SimplAI may reuse your Customer Data broadly, including to improve its platform and other offerings; the site never states whether that extends to training models on cloud deployments.
Data retention & training
Hosting summary
SimplAI does not publish where its own managed cloud hosts customer data: no country, no region and no cloud provider is named. What the site documents instead is the choice it hands to the customer, between SimplAI Cloud, the customer's own cloud or VPC, on-premises including fully air-gapped networks, sovereign cloud and hybrid edge, framing data residency as the customer's decision. In on-premises mode, inference, data processing and agent orchestration all run on customer hardware, with no external API calls and no cloud egress. Isolation controls include per-agent and per-tenant runtime isolation, private VPCs, private endpoints, and separated development, staging and production environments. Resilience relies on multi-zone deployment with automatic failover, continuous backups with point-in-time recovery, and cross-region geographic replication with customisable RPO and RTO targets. The privacy policy warns that personal data may be transferred outside your jurisdiction, under the European Commission's standard contractual clauses. Note that the domain resolves to a Cloudflare anycast address, which says nothing about where data actually lives.
Things to keep in mind
Risks and trade-offs to weigh before adopting SimplAI.
- The licence you grant over Customer Data is irrevocable and sublicensable and covers improving SimplAI's other offerings, so read section 10 of the terms before uploading anything sensitive
- Because no training opt-out is documented for cloud deployments, teams may wrongly assume the on-premises no-training promise applies to them
- Agents that act autonomously on internal APIs can do real damage if guardrails, spend limits and tool permissions are configured casually
- Automating claims, underwriting and credit decisions carries discrimination and compliance risk; audit trails record decisions but do not make them fair
- Confidence in a polished agent erodes human review, and citations and reflection reduce hallucinations without eliminating them
- With no published subprocessor list and no Article 27 representative, European buyers cannot easily assess third-party exposure
- Documentation inconsistencies, including contradictory minimum ages, a phantom Scale-up tier and terms pointing to a privacy URL that does not exist, mean the legal pages deserve a careful read rather than a skim
Setup & Integrations
Technical difficulty
Difficulty depends entirely on the path taken. Getting started is easy: the free tier needs no credit card, offers Google or Microsoft single sign-on, and agents are assembled on a visual canvas with capabilities toggled on, backed by more than fifty pre-built apps. Moving to production is developer work, since you must publish a version, enable the integration, then wire up the REST API, webhooks or the embed script; the Python SDK removes the polling, retry, streaming and scheduling plumbing. Enterprise deployments to VPC, on-premises or air-gapped environments are vendor-led, with hands-on go-live and white-glove onboarding.
Deployment
Integrations
Supported languages
Behind SimplAI
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
Who is behind SimplAI?
Is there a free plan, and what does the first paid tier cost?
Does SimplAI expose an API?
Which AI models can I use?
Can I run SimplAI on my own infrastructure?
Does SimplAI train its models on customer data?
What certifications and compliance does SimplAI claim?
How do I contact SimplAI?
Should you pick SimplAI?
SimplAI is a coherent, credible answer to a problem many enterprises actually have: agent projects that work in a demo and never reach production. Putting building, evaluation, deployment, observability and governance on one platform is well executed, and the deployment range is genuinely unusual. Very few agentic platforms will run the same agents inside a fully air-gapped network, and for banking, defence, healthcare and government buyers that alone narrows the shortlist.
The documentation supports the claim: seventeen sections, a documented REST API with working curl examples, a Python SDK, and a model layer spanning Anthropic, OpenAI, Google, Cohere, Mixtral, Ollama and TogetherAI with bring-your-own-key. The free tier is usable without a credit card, and enterprise onboarding is vendor-led, with a promised go-live in two days or less.
The reservations are contractual rather than technical. Section 10 of the terms grants an irrevocable, sublicensable licence over your data that extends to improving SimplAI's other offerings, and the reassuring no-model-training commitment appears only in the description of on-premises deployment. No subprocessor list is published, no Article 27 EU representative is designated, and no data protection officer is named. The pricing grid is inconsistent, the enterprise card pointing at a Scale-up tier that does not exist. Liability is capped at 100 USD and nothing is refundable.
This is also a young company: the domain was registered in January 2024, the first web archive capture dates from March 2024, and trade press reports a 2.5 million USD seed round. Treat SimplAI as promising and well worth a scoped pilot, with the contract read closely before anything sensitive goes into it.
- Choosing a selection results in a full page refresh.
- Opens in a new window.