Palo Alto, California, Aug. 03, 2026 (GLOBE NEWSWIRE) -- TinyFish, Inc. (Headquarters: Palo Alto, California; CEO: Sudheesh Nair; hereinafter “TinyFish”), the company behind Mako, a web-native AI model built to operate the live web, has entered into a strategic partnership with NEC Networks & System Integration Corporation . Together, the two companies are aiming to co-create a data bank for AI agents and accelerate the adoption of AI agents across enterprises and local governments in Japan, as part of NESIC’s AI Agent Ready Project.

As generative AI spreads through the enterprise, expectations are rising for AI agents that decide and act autonomously alongside business systems and SaaS. At the same time, the web carries vast amounts of information of uncertain authenticity and judging the freshness and accuracy of that information is extremely difficult. Decisions delayed by stale or inaccurate data, or by AI hallucinations*1, translate directly into lost opportunities and serious risk, so gathering information quickly, accurately, and reliably has become a central challenge. For AI agents to advance real business operations, a data bank built only from comprehensive and accurate data is indispensable.

TinyFish addresses this at the level of the model. Instead of adapting a general-purpose model to web tasks, TinyFish built Mako, a web-native AI model trained on production enterprise web task data and designed to operate the live web. Web execution does not require frontier-scale reasoning, and Mako is sized accordingly: 35 billion total parameters, with 3 billion active at inference, a fraction of the scale of a general-purpose frontier model. The same web workload costs materially less to run, so enterprises can operate web agents continuously across the business instead of limiting how often they are used.

Within the TinyFish platform, enterprises pair the reasoning model of their choice with Mako: the reasoning model understands the goal and plans the work, and Mako carries it out on the page. Acting on the user’s behalf, it acquires information even from sources without APIs, from dynamic pages where retrieval is difficult, and from sites that require logins, completing the multi-step workflows that general web agents fail to finish.

TinyFish also verifies what it gathers. It correlates and analyzes the freshness and accuracy of the information it collects and prevents the hallucinations specific to AI. By recognizing “unknown information” on its own and independently conducting further research, it extracts and structures only data of high reliability.

As the first initiative under the partnership, the two companies will deploy use cases in Business Continuity Planning (BCP), where accuracy and speed are decisive. During a disaster, for example, the system will support rapid decision-making by extracting only accurate facts, in real time, from the conflicting information found on social media and other sources. From there, the partnership will build use cases across a wide range of domains, including supply chain, marketing, and finance. Combining the high-precision world and web data acquired by TinyFish with the data enterprises already hold, NESIC is aiming to build a data bank through its AI Agent Ready Project and drive the adoption of AI agents that contribute to fast, high-level decision-making and business transformation.

◉ Comment from Sudheesh Nair, CEO of TinyFish, Inc.

“We are delighted to enter this strategic partnership with NESIC. As enterprises adopt AI worldwide, the difficulty has been less in reasoning than in acting on the live web accurately, without guesswork or hallucination. We built Mako for that work: a model designed for the web, able to reach information other systems cannot and to return only what it has verified. NESIC’s AI Agent Ready Project helps companies adopt AI agents safely and with confidence. Through this partnership we expect to co-create a data bank that searches and analyzes public information alongside data of every other kind at scale, and to support Japanese enterprises in making faster and better-informed decisions.”

■ About TinyFish

Company Name: TinyFish, Inc.

Established: 2024

Location: 4410 W El Camino Real, Los Altos, CA 94022, United States

CEO: Sudheesh Nair

Website: https://www.tinyfish.ai/

■ About the AI Agent Ready Project

The AI Agent Ready Project aims to “verify the optimal environment for using AI agents and quickly introduce it to the Japanese market” in order to support companies in adopting AI agents. In addition to accumulating knowledge through internal implementation, NESIC will build an optimal environment through co-creation with customers and partners, driving the new era of business pioneered by AI agents.

*1 Hallucination: a phenomenon in which AI generates answers that differ from the facts.

Company names and product names mentioned herein are trademarks or registered trademarks of their respective companies.

Image: https://www.globenewswire.com/newsroom/ti?nf=OTgwMjU5MiM3NzU1OTI3IzUwMDE2NDQzMg==

Image: https://ml.globenewswire.com/media/ZDUwMWY1ZGUtZjhmMC00MDU0LThkOGUtMmY2MWQ2YmE3MzA4LTUwMDE2NDQzMi0yMDI2LTA4LTAzLWVu/tiny/TinyFish.png Sky Zhang, Growth at TinyFish

[email protected]

https://www.tinyfish.ai/

Source: TinyFish