Prompt Flow Studio Launches a Multi-Model AI Workspace Built for Data Protection and Auditability

Compare, challenge and co-create across 400+ AI models — worldwide on the Web September 4, 2026; iOS and Android roll out next; enterprise controls in October.

TOKYO, JAPAN, September 4, 2026 /EINPresswire.com/ — EarthLink Network Co., Ltd. today announced that Prompt Flow Studio, a multi-AI workspace that lets teams put many AI models to work in a single environment, is officially launching worldwide — including the United States and Japan. The Web version is available from September 4, 2026, with the iOS and Android apps rolling out on the App Store and Google Play (https://prompt-flow.studio).

Prompt Flow Studio is more than a collection of AI models in one place. You choose the right model for the task, compare answers from several models side by side, have models debate one another, and co-write documents with multiple models. Uploaded internal documents can ground any AI you prefer, and the cost of every run is shown on the spot. It is a product built specifically to put models to work across vendors.

“One model gives one answer. Many models compare, challenge and co-write — surfacing disagreements before you decide. We bring that to enterprises, with data protection and auditability built in,” said Masayoshi Uehara, CEO of EarthLink Network Co., Ltd.

Enterprise data protection, rolling out beginning in October 2026. Organizations will be able to choose whether sensitive data is blocked before transmission or automatically masked before it is sent and restored when the response returns, with support for custom detection rules. The rollout also brings tamper-evident audit logging with hash-chained records, Legal Hold, retention of up to seven years, selective deletion of personal data, and governance over which AI services, models and standard personas an organization permits. Conversations are private by design: administrators are not granted day-to-day access to users’ prompts or AI responses, while legitimate legal requests are served through a controlled, fully audited disclosure process — everyday privacy and defensible evidence, both at once.

Debate — adversarial verification by multiple AIs. Assign different AI models (OpenAI, Anthropic, Google and others) — and personas — to different roles.

An advocate and a critic exchange claims and rebuttals, a facilitator keeps the issues organized, and a dedicated “Red Team” AI attacks the weaknesses of the conclusion before a synthesizer delivers the final synthesis, scored across multiple quality metrics.

Instead of accepting a single answer, you keep the entire argument that led to the conclusion.

Personas — decide who is answering. Personas define the expertise, stance and tone an AI should answer with — “a strict auditor,” “a hands-on practitioner,” “your brand’s editor.” Personas can be layered onto comparison and debate, not just chat. You can even pit the same model against itself with opposing personas — isolating the structure of a disagreement from differences between models. Organizations will be able to distribute standard personas, such as “our legal reviewer,” across whole teams.

Co-Draft — collaborative writing. Multiple AIs propose outlines in parallel, review one another’s work, and an integrator AI produces successive versions of a single document — a specification, a proposal, a concept paper — separating drafting from reviewing to avoid self-review.

Model-agnostic RAG. Ground answers in your own uploaded documents with the AI of your choice. Unlike tools that fix grounded answering to a single vendor’s models, the same grounded context can be answered by OpenAI, Anthropic (Claude), Google (Gemini) and more — so you can compare models on the shared foundation of the same source material, in strict (documents-only) or blended mode.

Compare — text and images. Send one prompt to several models at once and compare the results side by side. That includes simultaneous image generation: the same prompt sent to multiple image models, with results in a single grid and per-model costs displayed.

Cost transparency. Every response shows its confirmed cost in PFC credits, with line-item breakdowns for AI responses, knowledge retrieval and web search, plus monthly usage by mode, AI service and day. An estimate is shown before you send.

Free to try, priced to scale. The free plan includes Chat, Compare (text and image), Debate, Branch, Co-Draft, Workflow automation (currently in beta), model-agnostic RAG (1 GB), personas, model handover and web search on lightweight models, with 300 PFC credits granted at signup. Paid plans unlock all models.

Prompt Flow Studio supports more than 50 AI providers and over 400 models spanning text, image, audio and video, with a model gallery showing each model’s strengths and cost, an intent advisor that suggests the right model for the task at hand, and model handover that carries full context — objectives, constraints, conclusions — when you switch models mid-conversation.

Availability

Prompt Flow Studio is available worldwide on the Web from September 4, 2026, with the iOS and Android apps rolling out on the App Store and Google Play. Visit https://prompt-flow.studio for details and sign-up.

About EarthLink Network Co., Ltd.

EarthLink Network Co., Ltd. (Shirokane, Minato-ku, Tokyo; founded March 2000) provides IT solutions including systems development, infrastructure, AI consulting and education services. Corporate site: https://www.eln.ne.jp/

Media inquiries: https://prompt-flow.studio/en/contact

Masayoshi Uehara
EarthLink Network Co., Ltd.
support@eln.ne.jp
Visit us on social media:
LinkedIn
Facebook

Media gallery