TesseractHow it works
38 posts on the Tesseract blog.
- Gemini Omni 1.2 Flash in Tesseract: clips, edits, prices How Tesseract runs Gemini Omni 1.2 Flash: 3–15 s clips, 9:16 and 1080p, six reference pictures, edits in words, and a limit check before any charge. · 13 min read
- How an AI agent decides a command is too risky to run How Tesseract checks an agent's shell commands and file writes on 14 frameworks: 13 rules, a second opinion that can only raise, and what happens next. · 15 min read
- How AI agent tasks survive an app restart What happens to an agent's turn when the app restarts mid-task: liveness clocks, an on-disk turn record, dev server cleanup and conversation recovery. · 13 min read
- How one chat drives 14 AI agent frameworks the same way The agent contract behind Tesseract: three integration routes, 32 capability flags, tools the app serves itself, one event stream, and how each tool reaches the model. · 15 min read
- Updating an app without stopping its AI agents How Tesseract ships updates in kilobytes and applies them only when nothing is running: risk tiers, signed deltas, idle checks and automatic rollback. · 10 min read
- How an AI agent builds a PowerPoint deck: our design How Tesseract turns one sentence into a researched deck with speaker notes and a PowerPoint download: the pipeline, the tricks, the numbers and the results. · 12 min read
- How an AI makes a real font, and a brand kit around it How Tesseract turns a style prompt or your handwriting into an installable font, and one logo into a full brand kit: the pipeline, the checks, the kit. · 11 min read
- How Tesseract edits a PDF without breaking it Tesseract's PDF editor annotates, retypes, fills, signs and redacts. How saves keep the original bytes, why redaction rewrites, and what the agent can do. · 10 min read
- How an AI agent edits Word and Excel files in Tesseract Tesseract's Word and spreadsheet editors run every click, key and agent request through one command table. How that works, what it saves, and how it checks itself. · 10 min read
- How Tesseract's agent works in your own browser Tesseract Companion connects Chrome, Edge, Brave, Arc or Vivaldi to the app: how the bridge works, which tab the agent uses, and what keeps it visible. · 6 min read
- The Tesseract CLI: commands, MCP server and mesh What tesseract-agent does beyond hosting: installing without npm, browser sign-in, model accounts, connectors, an MCP server for Claude Code and Codex. · 8 min read
- The Remote Desktop app: share and connect, one window Tesseract Remote Desktop as a small free app: one window to connect out and share this screen, a small download, and how it works next to Tesseract. · 5 min read
- The Mesh app: a private network without the full app Tesseract Mesh as its own small app: what it does, how it opens, why it runs no background service, and how it hands over to Tesseract on one computer. · 7 min read
- Atrium as its own app: team chat without the AI app Tesseract Atrium as a free desktop app: what is in it, how it shares your sign-in with Tesseract, why it runs no service, and why it starts fast. · 5 min read
- How an AI game maker builds a game you can play at once Inside Tesseract's game maker: a design you approve, a browser game or an engine project, played beside the chat and tested by a headless checker. · 10 min read
- How an AI data analyst shows every change to your data Inside Tesseract's data maker: a first look at your file, a plan you approve, a re-runnable script, charts from its output and every change listed. · 10 min read
- How an AI research agent plans, cites and checks its work Inside Tesseract's research maker: a plan you approve, a checklist the agent ticks, a citation ledger, a coverage check and a report written from evidence. · 11 min read
- What is inside an AI assistant you make in Tesseract How Tesseract's assistant maker turns one sentence into a charter, a schedule, a first brief with no sign-in and rules that hold on every run. · 10 min read
- How the app maker checks a real app against its demo How Tesseract's app maker turns a tappable demo into Flutter, React Native or SwiftUI, then reads every built screen to prove it matches what you approved. · 9 min read
- How the website maker keeps, checks and publishes a site What happens to a site after Tesseract's website maker drafts it: the files it lives in, how edits reach the code, how it is checked and how it goes out. · 11 min read
- Present a deck live: notes, a phone remote and an AI How Tesseract runs a live talk: a presenter view with notes and a timer, your phone as the remote with no pairing, and an assistant that moves slides. · 11 min read
- How Tesseract's Enhance a photo maker fixes a photo Enhance a photo measures your picture, plans only the fixes it needs, runs them in a fixed order, mostly on your computer, and shows a before and after. · 9 min read
- How Tesseract generates images, transparency included How the agent makes pictures in the chat: your provider setting, four at a time, files kept with the session, and a real transparent PNG from any model. · 9 min read
- How the image editor in Tesseract works A layered image editor inside Tesseract: local tools, layers and filters, plus AI fills that change only the pixels you selected. How it all works. · 9 min read
- How a team chat app on your phone decides when to buzz How Atrium on the phone notifies only for messages you haven't seen, rings like a real phone call, and catches up after a lost connection. · 8 min read
- How a phone app steers AI agents on your computers Why the Tesseract phone app never runs an agent itself, how it stays connected through sleep and network changes, and which events reach your phone. · 10 min read
- How Tesseract writes one post seven ways Tesseract's social maker plans a set of posts, writes each caption for its platform, draws the pictures and saves a CSV your scheduler can import. · 7 min read
- How Tesseract designs a postcard you can print Say the occasion and who it is for: Tesseract designs both sides, draws the picture, picks readable words and saves a PDF at the card's real size. · 8 min read
- How Tesseract writes with you, not at you Write something asks only what you know, outlines first, looks facts up, keeps every version and your own edits, and hands over Word, PDF or mail. · 8 min read
- How Tesseract makes a skin for the whole app Describe a look or attach a picture, and Tesseract designs a theme for the app, keeps its text readable, makes its sounds and puts it on. · 10 min read
- Clean up a recording in Tesseract's audio editor Trim, de-noise, even out and bring a recording to a loudness target in Tesseract's audio editor, by hand or by asking the agent. The original stays untouched. · 7 min read
- Editing video with an AI agent: Tesseract's video editor Two video editors in one app: AI Edit for quick fixes to one clip, and a full timeline for a whole edit. The agent runs the same actions you click. · 8 min read
- How an AI video maker turns one sentence into a video How Tesseract's video maker works: questions, a script, a storyboard, every shot made at once, and a cut timed to the voice on your own computer. · 10 min read
- Violet: one AI conversation that never starts over How Violet gives each person a single conversation that never resets: a rolling window, pinned promises, hidden work sessions and long-term memory. · 10 min read
- How the Tesseract apps connect: one system map Desktop, web, CLI, Companion, Atrium, Remote Desktop and Mesh: which runs agents, what goes through the server, what goes direct, and what syncs. · 11 min read
- Can an AI-era vault replace your password manager? What Tesseract's vault stores, who can read it, how its AI reads forms without seeing your data, and how it fills on every device you use. · 13 min read
- We make complex things simple. Here's what's underneath Five things that take one sentence in Tesseract, and the engineering behind each: 14 agents in one chat, memory, computers by name, agent teams, decks. · 12 min read
- How AI agents remember across sessions: our design How Tesseract gives Claude Code, Codex and 12 other agents a long-term memory that finds the right past item without filling the context window. With numbers. · 10 min read