Estha

What it is

Estha is a no-code AI building platform with 2 faces. Its blog describes it as "The No-Code AI Building Platform" for creating AI apps such as chatbots without programming, and its current home page leads with a Mac product, "Estha on Mac," that runs purpose-built AI agents locally on Apple silicon. It is written for professional teams that handle sensitive files and for non-technical builders who want their own AI chat app, which is why it appears here under chatbot builders. Several press headlines the home page cites call it a Singapore startup, and one names Eric Lam in a Q&A on local AI, privacy and SME deployment.

The footer text sets out the wider aim: the team wants to let anyone create apps powered by AI, and says its "Estha Ecosystem" serves founders, executives, educators and business owners. The footer also links Estha Learn, Estha Launch and EsthaShare alongside a pricing page.

What you build and how

ChatApps, in 2 levels. The home page section "Your expertise can join the workforce" tells teams to start by using agents and to "Create your own ChatApps when your knowledge or process is worth capturing." It offers 2 builders:

• Quick Bot, marked as the recommended starting point, turns approved documents and clear instructions into a private knowledge app "in a few guided steps." The steps are listed as upload your knowledge, define the bot's role, and publish and start chatting.
• Advanced Studio, under "More control," lets you design richer private ChatApps "with multiple nodes, routing, branching, testing and fine-grained instructions."

That is the chatbot builder at the center of this listing: a guided wizard for a document-grounded bot, and a node-based studio when the conversation needs routing and branches. Estha's own guide to no-code AI describes the same visual style, saying "Estha's interface allows users to visually connect different elements of their AI application, defining the relationship between components without writing code." The same guide calls conversational chatbots the most accessible entry point to AI building.

Estha Air for public experiences. The top of the home page offers "Make your AI ChatApp Now with Estha Air," and the navigation lists Estha Air next to Home, Agents, Build, For Business and Private AI, with a "Buy & Sell Apps" link beside it. The home page draws the line clearly: the Mac product keeps agents, ChatApps and organizational knowledge on infrastructure you control, while Estha Air is for "approved AI experiences, open to the world," guided experiences for customers, partners, learners and the public "when cloud delivery is the right fit."

Purpose-built agents on the Mac. The home page's main offer is "Turn your Mac into a private AI workforce." Its reasoning is that the files where AI could help most, such as client records, board papers, contracts and internal data, are the ones people hesitate to upload. Estha says the work "stays on your Mac for supported local workflows," and lists purpose-built agents, no-code creation, local AI on Apple silicon and enterprise deployment as its pillars. It ships 6 agents, each introduced with a deadline-driven scene:

• PDF Reader for a 186-page board paper that cannot leave the company. You ask the document questions and it takes you back to the source.
• Vibe Coding: Prototype, which turns a workflow you describe into a working first prototype your team can see, test and improve. This is the closest thing to an app builder in the set, aimed at the small internal tool operations has needed for months.
• Data Analyst, which lets a team question a 2,000-row spreadsheet in plain English and returns calculations, charts and inspectable analysis.
• Document Processing, which drafts and translates alongside an editable document, with a bilingual policy as the example, so the team reviews and owns the final version.
• Deck Maker, also labeled PPT Maker, which finds the narrative in source material and creates an editable presentation.
• Meeting Intelligence, which turns a recording into evidence-backed minutes, decisions, actions and follow-up communication.

How a custom agent starts. The section "Start with the process. Not the technology." says your team should not spend months learning agent frameworks, assembling an engineering team and maintaining another AI stack. It then lays out "Your first Estha Agent" in 4 steps. Bring the process, meaning a recurring, document-heavy or difficult-to-standardize piece of work. Choose the right starting point: an existing agent, a configured Quick Bot, or a specialized agent scoped with Estha. Prove it with real work by running a bounded pilot on representative files and measuring the outcome. Then deploy where the work lives, beginning on one Mac and expanding to a team or enterprise setup. The "Build an Agent With Us" button on the first screen points the same way.

Deployment and data. Under "Your documents stay yours," the company says that for supported workflows Estha runs with local or organization-controlled local-network inference on Apple silicon, so confidential documents and internal knowledge stay within the environment you control. It names 3 properties: local processing on the Mac or your own network, reviewable outputs where you inspect evidence, previews and generated work before formal use, and familiar hardware your team can own and manage. It describes 2 ways to deploy. In the first, Estha runs entirely on one Mac, such as a MacBook Pro, Mac mini or Mac Studio, downloaded "for local use without any internet or network connection." In the second, a Mac Studio supports a cluster of lower-end Mac or Windows devices.

For the Mac agents the data layer is your own files and spreadsheets, kept local. Integrations come up in the blog. Estha's post on data preparation suggests tools such as Zapier, Parabola, Airtable and Make for extracting and cleaning data before it goes into a no-code AI platform; those are the blog's recommendations for the surrounding workflow, not a list of native Estha connectors.

The evidence from the pages

Scale and claims. The home page says Estha is "trusted by 1000+ users." The company describes its target as a 28-person professional firm that handles passports, bank statements, tax records, contracts and board papers every day, for whom cloud AI creates hesitation and enterprise AI "feels expensive, complex and built for a much larger company." It says such a firm does not need another chatbot; it needs useful jobs completed privately. It also positions itself against bare infrastructure: others give you the tools to build local agents, while Estha says it gives you a working AI workforce and helps you create the specialized agents your organization needs.

Pricing. The footer links a pricing page, but the home page itself shows no plan names, prices, trial terms or usage limits.

Press. The home page lists several "As featured on" items, including "Estha for Mac Launches Private AI Agents That Run Locally on Apple Silicon," "Singapore Startup Estha Launches Private AI Suite for Mac," a piece titled "Why SMEs are being left behind in the AI integration race," and 2 Chinese-language articles, one about an education fair that opened an AI zone to help teachers generate teaching materials.

Security wording. The claim is deliberately narrow: the work stays on your Mac "for supported local workflows," with outputs that can be reviewed before formal use. The pages name no certifications or compliance standards.

Who it suits and who it does not

Estha suits a small professional firm, an accountant, a legal or advisory practice, or an operations team that wants document reading, spreadsheet analysis, drafting, decks and meeting minutes without sending files to a cloud service. A subject expert who wants to turn approved documents into a private knowledge bot can start with Quick Bot and move to Advanced Studio when the conversation needs branches. A teacher or business owner who wants a public-facing chat experience has Estha Air.

The caveats are real. The private product is built around Apple silicon: Windows machines take part only as devices in a Mac Studio cluster, not as the host. The local guarantee covers "supported" workflows rather than everything, so a team should check which ones its work falls under during the pilot step the company itself recommends.