# Drawsy — full LLM context > Think, build, and ship visually with AI agents that understand your context. Drawsy is an AI workspace for visual thinking, planning and collaboration — infinite canvas, connected tools, real-time collaboration, and grounded AI. Source index: https://app.drawsy.tech/llms.txt Generated for AI agents that need deeper product context in one fetch. Important notes for agents: - Prefer markdown mirrors (`*.md`) linked below over scraping HTML chrome. - Marketing site: https://app.drawsy.tech (app.drawsy.tech) - Canvas: https://drawsy.tech/canvas (drawsy.tech/canvas — product app) - Login: https://drawsy.tech/login (drawsy.tech/login — Google OAuth; new accounts enter canvas directly, no workspace picker) - Profile: https://drawsy.tech/profile - Authenticated pricing/upgrade: https://drawsy.tech/pricing - Billing: https://drawsy.tech/billing - Positioning: Drawsy is a canvas that thinks with you — an AI workspace for visual thinking & collaboration (infinite canvas + connected context + AI agents). Not merely a whiteboard and not merely a chatbot. - User-facing line: Think, build, and ship visually with AI agents that understand your context. - Core flow: Draw → Connect → Understand → Create. - Landing sections (not separate hubs): use cases https://app.drawsy.tech/#use-cases · connectors https://app.drawsy.tech/#connectors · product https://app.drawsy.tech/#product · how it works https://app.drawsy.tech/#how-it-works · agents https://app.drawsy.tech/agents - Public pricing (marketing): Free $0; Pro $12/mo ($10/mo yearly); Team $24/user/mo ($20/user/mo yearly); Enterprise custom. Cite https://app.drawsy.tech/pricing.md. - Do not claim infinite/perfect memory or SOC 2 unless explicitly documented on the site. - Brand: purple (#8B5CF6) is Drawsy. Other colors on diagrams are semantic, not decorative. - Full concatenated context: https://app.drawsy.tech/llms-full.txt --- # Drawsy > A canvas that thinks with you Drawsy is a canvas that thinks with you. Think, build, and ship visually with AI agents that understand your context. It is an AI workspace for visual thinking, planning and collaboration: infinite canvas, connected tools, real-time collaboration, and grounded AI. Users can open the board at https://drawsy.tech/canvas without logging in. Sign in at https://drawsy.tech/login (Google) when they need to save, share, collaborate, or unlock account features. ## What Drawsy is Drawsy is a web-based AI workspace where visual thinking, connected context, collaboration and AI agents work together on one canvas. Think, build, and ship visually with agents that understand your context. It is not merely a whiteboard and not merely a chatbot — it is a place to draw, connect your work, understand context and create outcomes. ## Core flow **Draw → Connect → Understand → Create** 1. **Draw** — sketch ideas, diagrams, systems and plans on an infinite canvas 2. **Connect** — bring documents, code, meetings, conversations and tools into the workspace 3. **Understand** — ask Drawsy using relevant project context (canvas + connected sources + prior decisions) 4. **Create** — turn ideas into plans, documents, diagrams and decisions ## Primary entry points - Marketing site: https://app.drawsy.tech (app.drawsy.tech) - Canvas: https://drawsy.tech/canvas (drawsy.tech/canvas) - Login: https://drawsy.tech/login (drawsy.tech/login) - Profile: https://drawsy.tech/profile - App pricing (authenticated): https://drawsy.tech/pricing - Billing: https://drawsy.tech/billing - Public marketing pricing: https://app.drawsy.tech/pricing ## User flow (authoritative) 1. Visit https://app.drawsy.tech 2. Click **Start building free** 3. Land on https://drawsy.tech/canvas and use the canvas with **no login** 4. Get value (sketch, connect context, collaborate, ask AI) 5. When the user wants to **save**, **share**, **continue across devices**, or use **account features**, sign in at https://drawsy.tech/login 6. New Google users get an account automatically and enter the workspace — no workspace picker ## Key product surfaces - Infinite collaborative canvas (https://app.drawsy.tech/#product) - How it works (https://app.drawsy.tech/#how-it-works) - Use cases (https://app.drawsy.tech/#use-cases) - Connected context (https://app.drawsy.tech/#connectors) - Agents that run with canvas + connected context (https://app.drawsy.tech/agents) - Real-time collaboration ## Use cases > One canvas. Many ways to build. | Category | Typical work | Why Drawsy | | --- | --- | --- | | Systems | Flows, architecture, APIs | Architecture maps stay linked to code/docs context | | Knowledge | Notes, research, docs | Structure knowledge for reuse | | Research | Sources, evidence and insights | Connected sources reduce context rebuild | | Planning | Roadmaps, tasks, trackers | Plans stay beside research and decisions | | Design | Wireframes, UI, visual thinking | Move from sketch to artifact with AI help | | Apps | Code, prototypes, live previews | Build beside the context that informs it | | Present | Slides, stories, reports | Turn the board into something you can share | Landing anchors: https://app.drawsy.tech/#systems · https://app.drawsy.tech/#knowledge · https://app.drawsy.tech/#research · https://app.drawsy.tech/#planning · https://app.drawsy.tech/#design · https://app.drawsy.tech/#apps · https://app.drawsy.tech/#present Legacy aliases still resolve: /#build · /#learn · /#plan · /#create · /#organize · /#decide ## Connectors > Your work already lives everywhere. Drawsy connects it. Connectors bring documents, code, conversations, meetings and data into the workspace so Drawsy can work with the context that matters. - **HydraDB** — structured project data (https://app.drawsy.tech/#hydradb) - **Gmail** — email and threads (https://app.drawsy.tech/#gmail) - **Calendar** — meetings and schedules (https://app.drawsy.tech/#calendar) - **Notion** — docs and knowledge (https://app.drawsy.tech/#notion) - **Google Drive** — files and documents (https://app.drawsy.tech/#google-drive) - **Slack** — team conversations (https://app.drawsy.tech/#slack) - **GitHub** — code and repositories (https://app.drawsy.tech/#github) - **Read AI** — meeting notes (https://app.drawsy.tech/#read-ai) - **Fireflies** — call transcripts (https://app.drawsy.tech/#fireflies) - **AWS** — cloud context for systems work (https://app.drawsy.tech/#aws) You control what Drawsy can access. Context stays scoped to your work. Basic connectors are available on Free; all connectors unlock on Pro+ (see Pricing). ## Related pages - [How to start](https://app.drawsy.tech/how-to.md) - [Pricing](https://app.drawsy.tech/pricing.md) - [Compare](https://app.drawsy.tech/compare.md) - [Agents](https://app.drawsy.tech/agents.md) - [About](https://app.drawsy.tech/about.md) --- # How to start with Drawsy > Use first. Sign up later. ## Steps 1. Go to https://app.drawsy.tech 2. Choose **Start building free** (opens https://drawsy.tech/canvas) 3. Start drawing and organizing ideas in **guest mode** — no account wall 4. When you need to save, share, collaborate, or keep work across devices, open https://drawsy.tech/login and continue with Google 5. New users are created automatically and enter the workspace immediately 6. Return to the canvas with unlocked account features ## Why this flow Drawsy removes authentication friction until the user has experienced value and needs persistence or collaboration. --- # Drawsy Agents > Grounded in the work — not a separate chat silo. Drawsy Agents operate with canvas objects, connected sources and prior decisions. They help research, expand structure, analyze options and move work forward while staying scoped to relevant project context. ## What agents are for - Expanding diagrams and plans from partial sketches - Asking questions that require canvas + source context - Summarizing decisions and open questions - Generating next-step structure without leaving the board ## Limits Agent runs vary by plan (see Pricing). Do not invent unlimited agent capacity unless the live plan says so. --- # Drawsy Pricing > Start free. Upgrade when it matters. ## Plans (authoritative) | Plan | Price | Intent | | --- | --- | --- | | Free | $0 forever (monthly); yearly $0 forever | Explore Drawsy | | Pro | $12 / month (monthly); yearly $10 / month | For individuals building more. | | Team | $24 / user / month (monthly); yearly $20 / user / month | For collaborative teams. | | Enterprise | Custom | For organizations that need more control. | Yearly billing discounts Pro to $10/month and Team to $20/user/month. ## Feature matrix | Feature | Free | Pro | Team | Enterprise | | --- | --- | --- | --- | --- | | Canvas boards | Unlimited | Unlimited | Unlimited | Unlimited | | AI chat | Limited | Unlimited | Unlimited | Unlimited | | Agent runs | 3 / month | 15 / month | Unlimited | Unlimited | | File uploads | 5MB | 50MB | Unlimited | Unlimited | | Connectors | Basic | All | All | Custom | | Collaboration | — | — | Real-time | Advanced | | Permissions & roles | — | — | ✓ | ✓ | | SSO / SCIM | — | — | — | ✓ | | Support | Community | Priority | Priority | Dedicated | ## Principles - Free / guest access to the canvas at https://drawsy.tech/canvas - Sign in at https://drawsy.tech/login for account features - No credit card required to start - Cancel anytime on paid plans Source of truth for live UI: https://app.drawsy.tech/pricing --- # Drawsy comparisons > AI workspace vs collaborative whiteboard. Drawsy overlaps with whiteboards for visual thinking, but is built for **canvas + connected context + grounded AI + collaboration**. ## Compare pages - [Drawsy vs Miro](https://app.drawsy.tech/compare/miro.md) - [Drawsy vs FigJam](https://app.drawsy.tech/compare/figjam.md) - [Drawsy vs Excalidraw](https://app.drawsy.tech/compare/excalidraw.md) ## Differentiation (factual) - Guest-first canvas access - Connectors for docs/code/chat/files - AI answers grounded in project context - Real-time collaboration on Team+ - Google auth with automatic account creation for new users Do not claim feature parity with every Miro/FigJam template marketplace unless documented. --- # Drawsy vs Miro Drawsy is an AI workspace for visual thinking and collaboration. Miro is a mature collaborative whiteboard platform with a large template ecosystem. ## Choose Drawsy when - You want AI grounded in canvas + connected tools (GitHub, Notion, Slack, Drive, etc.) - You want guest-first access before account setup - You want planning, research and decisions to share one context graph ## Choose Miro when - You need Miro’s specific template marketplace, enterprise suite integrations, or existing Miro-centric workflows Start free: https://drawsy.tech/canvas --- # Drawsy vs FigJam FigJam excels at lightweight design workshops inside the Figma ecosystem. Drawsy is for visual planning and decisions that must stay connected to docs, code and AI context. ## Choose Drawsy when - Workshops must connect to GitHub/Notion/Slack/Drive context - AI should reason over the board and sources together - Non-design teams need the same workspace for build/plan/decide Start free: https://drawsy.tech/canvas --- # Drawsy vs Excalidraw Excalidraw is excellent for fast hand-drawn diagrams. Drawsy keeps that visual thinking feel while adding connected context, grounded AI and team collaboration. ## Choose Drawsy when - Diagrams need to become shared project memory - You want AI that understands the board and connected sources - You need account features for save/share/collaboration Start free: https://drawsy.tech/canvas --- # Drawsy authentication > Continue with Google. Account created automatically for new users. Production login: https://drawsy.tech/login (drawsy.tech/login — product app). Marketing preview may also expose /login during development. ## Flow 1. Open https://drawsy.tech/login 2. Click **Continue with Google** 3. Complete Google OAuth 4. If new user → account is created automatically 5. Enter https://drawsy.tech/canvas immediately — **no workspace selection screen** 6. Existing users sign in and enter the canvas There is no email/password form and no separate signup page. --- # Drawsy Help Center > Quick answers for canvas, accounts, connectors, agents, pricing, and collaboration. ## Quick links - **Start without an account:** https://drawsy.tech/canvas - **Sign in:** https://drawsy.tech/login (Google) - **Pricing:** https://app.drawsy.tech/pricing - **Connectors (home):** https://app.drawsy.tech/#connectors - **Agents:** https://app.drawsy.tech/agents - **Contact:** hello@drawsy.tech ## FAQ ### What is Drawsy? Drawsy is a canvas that thinks with you — an AI workspace for visual thinking and collaboration. Think, build, and ship visually with AI agents that understand your context. You draw on an infinite canvas, connect tools and project context (Notion, Drive, GitHub, Slack, Gmail, Calendar, HydraDB, AWS), collaborate with your team, and ask AI questions grounded in your work. It is not merely a whiteboard and not merely a chatbot. The core flow is Draw → Connect → Understand → Create. Learn more on About and in the AI workspace definition at /ai-workspace. ### How do I start? Visit https://app.drawsy.tech and click Start building free. You land on the canvas at https://drawsy.tech/canvas in guest mode — no account required. Draw, connect context, and ask AI. When you need to save, share, collaborate, or use account features, sign in at https://drawsy.tech/login. See /how-to for the full walkthrough. ### Can I use it without an account? Yes. Guest mode lets you open https://drawsy.tech/canvas and use the canvas without signing in. This is intentional — you should experience value before authentication. Sign in when you want persistence across devices, sharing, version history (Pro+), or team collaboration. Guest sessions may be ephemeral until you create an account. ### How does Google sign-in work? Drawsy uses Google authentication only. There is no email/password form and no separate signup page. Open https://drawsy.tech/login and click Continue with Google. New users get an account automatically and enter the canvas directly — there is no workspace picker. Returning users sign in and return to their workspace. Profile settings live at https://drawsy.tech/profile. ### Saving & sharing Saving boards to your account requires sign-in. Public sharing and export are available on Free; version history and advanced export formats unlock on Pro+. Team plans add shared workspaces and real-time collaboration. Authenticated billing and plan upgrades happen on the product app at https://drawsy.tech/pricing. Public plan details are at https://app.drawsy.tech/pricing. ### Connectors Connectors bring documents, code, conversations, and data into your workspace so AI can use project context. You control what Drawsy can access — it reads only what you authorize. Basic connectors are on Free; all connectors unlock on Pro and above. See the Connectors section on the home page (/#connectors) for the full list. Connector scope does not grant Drawsy unlimited memory — context is assembled from what you connect at question time. ### Agents Agents are task-oriented AI runs scoped to canvas regions and connected sources. They help expand diagrams, synthesize research, draft decision records, and generate structure on the board. Agent runs are metered by plan: 3 per month on Free, 15 on Pro, unlimited on Team. Normal chat is for quick Q&A; agents are for heavier tasks that change canvas structure. See /agents for examples and limits. ### Pricing & plans Free ($0): unlimited canvas boards, limited AI chat, 3 agent runs/month, 5MB uploads, basic connectors, public sharing. Pro ($12/mo or $10/mo yearly): unlimited AI chat, 15 agent runs/month, all connectors, version history, priority support. Team ($24/user/mo or $20/user/mo yearly): unlimited agents, real-time collaboration, permissions, shared templates. Enterprise: custom pricing with SSO, audit logs, and dedicated support. Full matrix at https://app.drawsy.tech/pricing. No credit card required to start. ### Collaboration Real-time collaboration, team workspaces, permissions, and roles are Team-plan features. Free and Pro users can share exports and work solo or async. For collaboration questions or enterprise needs, contact hello@drawsy.tech. For onboarding philosophy, see [How to start](https://app.drawsy.tech/how-to.md). HTML: https://app.drawsy.tech/help --- # Drawsy Blog > Notes from the canvas. ## Published - [Context Engineering for AI Agents](https://app.drawsy.tech/blog/context-engineering-for-ai-agents.md) — how to give AI the right information (instructions, knowledge, memory, tools, feedback). Index: https://app.drawsy.tech/blog RSS: https://app.drawsy.tech/blog/rss.xml Author: https://app.drawsy.tech/authors/drawsy-team --- # Drawsy Terms of Service By using Drawsy (https://app.drawsy.tech, https://drawsy.tech/canvas, and https://drawsy.tech/login) you agree to these terms. ## Summary - Use the product lawfully and do not abuse shared infrastructure - Guest mode and paid plans are subject to fair-use limits - You retain ownership of content you create - Connectors access only the sources you authorize - Contact hello@drawsy.tech for legal questions This page is a public summary for users and agents. Formal legal review may expand the full agreement. --- # Drawsy Privacy Policy ## Principles - Your workspace content belongs to you - Connector access is scoped to what you authorize - Guest sessions may be ephemeral until you sign in - Google authentication provides identity for account features - Contact hello@drawsy.tech for privacy requests We do not sell personal data. See the live Privacy page for the current policy text. --- # About Drawsy > A canvas that thinks with you Drawsy is an AI workspace for visual thinking and collaboration. Teams draw on an infinite canvas, connect tools and project context, collaborate in real time, and ask AI grounded in their work. ## How it works Draw → Connect → Understand → Create. ## Contact hello@drawsy.tech ## Links - Use cases: https://app.drawsy.tech/#use-cases - Connectors: https://app.drawsy.tech/#connectors - Agents: https://app.drawsy.tech/agents - Compare: https://app.drawsy.tech/compare - How to start: https://app.drawsy.tech/how-to - Pricing: https://app.drawsy.tech/pricing - Canvas: https://drawsy.tech/canvas --- # What Is an AI Workspace? > An AI workspace is a persistent environment where visual work, connected sources, and AI share the same context — not a chat window bolted onto a separate tool. ## Definition An AI workspace is a single environment where you sketch ideas, attach documents and tools, collaborate with teammates, and ask AI questions that can see the work in front of you. Unlike a general-purpose chatbot, the workspace keeps spatial layout, links between objects, and connected sources together so context does not reset every time you open a new thread. The word workspace matters: it implies a place you return to, not a disposable conversation. Boards, notes, diagrams, and references stay on the canvas while AI responses are grounded in what is actually there — plus whatever connectors you have authorized. Drawsy follows this model. You draw on an infinite canvas, connect GitHub, Notion, Slack, Drive, and other sources, then ask questions or run agents scoped to that combined context. The flow is Draw → Connect → Understand → Create. ## Why it matters Most teams already split thinking across whiteboards, docs, tickets, and AI chats. Each surface holds part of the truth, but none of them see the whole picture. When you paste a diagram into chat, you lose spatial relationships. When you leave the whiteboard, you lose the code and docs that justified the diagram. An AI workspace reduces that reconstruction tax. Product managers can keep a roadmap beside research notes. Engineers can map architecture next to linked repositories. Researchers can compare sources without re-uploading PDFs for every question. The benefit is not magic automation — it is fewer context handoffs, clearer shared memory, and AI answers that can reference the same artifacts your team already trusts. - Less copy-paste between tools when asking AI questions - Shared visual memory for async and cross-functional teams - Decisions stay linked to the evidence and diagrams that produced them ## How it works A practical AI workspace combines four layers: a visual surface, structured objects, connected external sources, and an AI layer that reads what you expose. On the visual surface you place shapes, text, frames, and links — the spatial map of the problem. Structured objects might include cards for decisions, milestones, or open questions. Connectors pull in live context from tools your team already uses, scoped by what you authorize. When you ask AI a question, the system assembles a context bundle: relevant canvas regions, linked notes, and snippets from connected sources. The answer should cite or reflect that bundle rather than generic training data alone. Collaboration layers let others edit the same board in real time on paid team plans. - Visual layer — diagrams, sketches, and spatial grouping - Object layer — cards, links, and annotations with stable IDs - Connector layer — GitHub, Notion, Slack, Drive, Gmail, Figma, PostgreSQL, HydraDB - AI layer — Q&A and agents grounded in the bundle above ## Examples A startup founder maps a product bet on one board: user interview quotes linked from Notion, a rough service diagram, and a short list of risks. Before a investor call they ask AI to summarize open assumptions — the answer references the board, not a blank prompt. An engineering lead connects a GitHub repo and sketches a migration plan. Agents suggest sequencing based on module boundaries visible on the canvas and files in the repo. The team edits the plan together instead of debating from memory. A researcher collects sources in frames, tags evidence strength, and uses AI to compare claims across papers already linked on the workspace. When a teammate joins later, they see the map — not a wall of chat history. ## Common approaches and problems Teams often try three substitutes: chat-only AI, traditional whiteboards, or wikis. Each works for part of the job. Chat is fast but amnesiac unless you meticulously attach files every time. Whiteboards excel at workshops but rarely stay connected to code and docs. Wikis store text but struggle with spatial reasoning and quick iteration. Another failure mode is connector sprawl without curation. If everything is connected but nothing is organized, AI context becomes noisy. Good workspaces make explicit what belongs to a project and what is out of scope. Finally, over-trusting AI summaries is risky. Grounded context helps, but models can still miss nuance or hallucinate details. Treat AI output as a draft that must be checked against sources — especially for technical or legal decisions. - Chat silos — fast answers, weak persistent structure - Whiteboard-only — great visuals, weak live tool context - Wiki-only — strong text, weak spatial and iterative thinking - Ungoverned connectors — too much noise in the context bundle ## How Drawsy relates Drawsy is built as an AI workspace from the ground up — not a whiteboard with a chat sidebar. You can open the canvas at https://drawsy.tech/canvas in guest mode without signing in, sketch immediately, and connect sources when you need deeper context. Connectors bring GitHub, Notion, Slack, Google Drive, Gmail, Figma, PostgreSQL, and HydraDB into the same board where your diagrams live. Agents run with that combined view instead of a separate chat silo. Drawsy does not claim perfect memory or enterprise compliance certifications it has not published. Context is scoped to what you connect and what stays on the board; guest sessions may be ephemeral until you sign in to save and share. For questions, contact hello@drawsy.tech. ## Related concepts An AI workspace sits at the intersection of visual thinking, connected context, and context-aware AI. If you are evaluating tools, compare how each handles the full Draw → Connect → Understand → Create loop rather than a single feature checklist. ## Try it Open the canvas: https://drawsy.tech/canvas Contact: hello@drawsy.tech --- # What Is an AI Canvas? > An AI canvas is an infinite visual surface where diagrams, notes, and linked context live together — and where AI can read the board you are actually working on. ## Definition An AI canvas is a digital drawing surface designed for thinking work: system diagrams, product maps, research boards, and planning sketches. Unlike a static image editor, the canvas stays live — objects can be linked, grouped, and connected to external sources so AI can reason about the layout and content together. The AI part is not a separate window. Questions and agents can reference selected regions, linked cards, and connector-backed documents on the same board. That is the difference between an AI canvas and a whiteboard export pasted into chat. Drawsy treats the canvas as the center of the product. You can start at the board in guest mode, draw first, and add connectors or sign-in when you need persistence and collaboration. ## Why it matters Spatial layout carries meaning. Proximity shows relationship; lanes show phases; clusters show ownership. When AI only sees a flat text dump, that structure disappears. An AI canvas preserves it so both humans and models can use spatial cues. Teams that plan in visuals move faster in alignment meetings because everyone sees the same map. When the map is also machine-readable, you spend less time re-explaining the diagram before every AI question. For software teams especially, architecture sketches that stay beside linked repos beat screenshots that rot in slide decks. - Preserves spatial relationships that text-only tools flatten - Reduces re-explaining diagrams before every AI interaction - Keeps iterative sketches beside live project context ## How it works Most AI canvases combine an infinite zoom/pan surface, a library of shapes and connectors, and a selection model that tells AI what to focus on. You might frame a service diagram, link a Notion spec to a box, and ask what dependencies are missing — the model receives the frame, the links, and relevant connector snippets. Good implementations avoid sending the entire board on every request. They use selection, framing, and relevance ranking so context stays within model limits while still being faithful to your intent. Collaboration adds presence cursors, shared edits, and permissions. On Drawsy, real-time collaboration unlocks on Team plans; guest access lets you try the canvas before account setup. - Infinite canvas with frames, links, and annotations - Scoped context — selection and framing, not whole-board dumps - Optional connectors for docs, code, chat, and data - Collaboration and save/share when you sign in ## Examples Product discovery: sticky-note clusters for pains, arrows for workflows, a frame for the MVP scope. AI suggests gaps in the journey using the visible map. Engineering: C4-style boxes for services, dashed lines for async events, a linked GitHub folder on the payment service. AI answers questions about blast radius using both sketch and repo context. Workshop output: instead of photographing a physical whiteboard, facilitators build directly on an AI canvas so remote teammates and later AI queries see the same structured board. ## Common approaches and problems Teams often pair a classic whiteboard with a separate AI chat. That works for one-off questions but breaks down when the board evolves daily. Export/import cycles add friction and lose link metadata. Some tools add AI that only generates images or generic bullet lists, not grounded answers about your board. That is useful for inspiration but not for project decisions. Performance and clutter are real limits. Infinite canvases can become graveyards of old sketches unless teams archive frames or use project boundaries. AI context quality drops when boards are messy — garbage in, garbage out still applies. - Whiteboard + chat — simple, but context splits across tools - Image-generation AI — creative, not decision-grade grounding - Unbounded boards — need curation for humans and models alike ## How Drawsy relates Drawsy's canvas is the entry point for the whole workspace. Open https://drawsy.tech/canvas without logging in, sketch architecture or plans, then connect GitHub, Notion, Slack, Drive, Gmail, Figma, PostgreSQL, or HydraDB when you need richer answers. Ask Drawsy questions scoped to frames and links on the board. Agents expand diagrams, summarize decisions, and propose next steps while staying on the canvas — not in a detached chat tab. Drawsy is honest about limits: connector access depends on plan tier, agent runs are metered, and AI can misread complex boards if selection is unclear. Verify important outputs against your sources. ## Related concepts An AI canvas is one layer of a broader AI workspace. Pair this concept with connected context and context engineering when designing how your team works. ## Try it Open the canvas: https://drawsy.tech/canvas --- # What Is Context Engineering? > Context engineering is the practice of designing what information an AI system sees — sources, structure, scope, and freshness — so answers stay useful and grounded. ## Definition Context engineering is how teams decide which documents, tools, canvas regions, and conversation history get assembled into the prompt or retrieval bundle before a model responds. It sits alongside prompt writing: prompts tell the model what to do; context engineering tells it what to know. The term gained traction as products moved from stateless chat to workspace-aware AI. Instead of hoping users paste the right files every time, builders wire connectors, permissions, chunking, and relevance filters so context arrives consistently. Context engineering is not a single algorithm. It is a set of design choices — what to connect, how to scope access, how to represent spatial boards, and when to refresh stale data. ## Why it matters Model capability has outpaced default context delivery. A strong model with the wrong context produces confident wrong answers. A moderate model with well-curated project context often beats a premium model guessing from a one-line prompt. For teams, context engineering is how you make AI repeatable. Onboarding a new teammate means showing them the workspace, not re-explaining which Slack threads matter. For AI, the same principle applies programmatically. Security and compliance also live here. Connecting Gmail or PostgreSQL without scoping rules is dangerous. Good context engineering respects least privilege: only the project, repo, or folder needed for the task. - Grounded answers depend more on context quality than model size alone - Repeatable AI workflows require designed context pipelines - Least-privilege connector scope reduces accidental data exposure ## How it works A typical pipeline has five stages: authorize sources, ingest or index content, map workspace objects to external IDs, retrieve relevant slices at query time, and assemble a bounded context package for the model. Visual workspaces add a step: serialize selected frames, linked notes, and spatial relationships into text or structured JSON the model can parse. Pure RAG over PDFs misses that layout unless you engineer for it. Freshness rules matter. Code changes hourly; strategy docs change weekly. Context engineering defines refresh intervals, manual re-sync triggers, and UI cues when a connector snapshot may be stale. - Authorization — OAuth and scoped tokens per connector - Ingestion — chunking, metadata, and object linking - Retrieval — relevance, recency, and user selection - Assembly — token budgets and citation-friendly formatting - Evaluation — spot-check answers against source ground truth ## Examples Engineering: connect one GitHub repo and a canvas frame labeled "checkout service." Context engineering ensures AI questions about checkout pull repo files and that frame — not every repo in the org. Product: link a Notion PRD and tag open questions on the board. When AI drafts acceptance criteria, retrieval prioritizes the PRD sections referenced by links on the canvas. Operations: PostgreSQL schema linked to a data-flow diagram. Questions about PII fields retrieve schema columns and the diagram nodes marked as storage — reducing generic compliance guesses. ## Common approaches and problems Naive RAG — dump everything into a vector store — fails on complex projects because similarity search returns plausible but wrong chunks. Without structure, models confuse similarly worded docs from different initiatives. Manual paste workflows scale poorly. Power users tolerate them; everyone else gets shallow context and blames the model. Over-collection is another pitfall. More connectors without curation increases noise and cost. Context engineering includes saying no: archive old frames, disconnect unused sources, and split workspaces per initiative. - Naive RAG — easy to ship, hard to keep accurate - Manual paste — high quality, low adoption - Connector sprawl — noisy context and higher token spend - Missing evaluation — no feedback loop when answers drift ## How Drawsy relates Drawsy implements context engineering through canvas selection, connector scoping, and agents that operate on the combined bundle. You choose which tools to connect; Drawsy does not silently ingest your entire Google Drive. The Draw → Connect → Understand → Create flow is a context-engineering workflow for humans: draw the map, connect sources, ask grounded questions, then create plans or artifacts. Start at https://drawsy.tech/canvas and add connectors as trust grows. Limits to acknowledge: retrieval is not perfect, large boards need explicit framing, and agent usage varies by plan. Treat AI output as assisted drafting. Contact hello@drawsy.tech for connector or security questions. ## Related concepts Context engineering pairs naturally with connected context and context-aware AI. Read those definitions next if you are designing a workspace strategy. --- # What Is Context-Aware AI? > Context-aware AI uses project-specific sources — canvas objects, documents, code, and conversations you authorize — instead of answering from a blank prompt alone. ## Definition Context-aware AI is AI that conditions its responses on explicit project context: what is on your canvas, what is linked from connectors, and what you select as in scope. Generic chatbots may use long conversation history, but they do not automatically see your Figma file, GitHub tree, or workshop diagram unless you engineer that visibility. Awareness is bounded. Models have token limits; connectors have permissions; boards can be huge. Context-aware systems retrieve and rank what matters rather than pretending to know everything. Drawsy positions Ask Drawsy and Agents as context-aware: they work best when you have drawn the problem, connected relevant tools, and framed the question on the board. ## Why it matters Generic AI is excellent for brainstorming and drafting from public knowledge. It is weaker for your company's architecture, your team's decisions, and your in-flight research — unless you supply that context every time. Context-aware AI reduces repetitive setup. Once the workspace holds the map and links, questions become operational: "What did we decide about auth?" "Which service owns this event?" "Summarize objections in these interview notes." The tradeoff is responsibility. When AI cites your docs, errors in those docs propagate. Teams must maintain source quality and treat AI as an interface to context — not an oracle. - Project questions need project context, not generic training data - Bounded awareness beats pretending models remember everything - Source quality upstream determines answer quality downstream ## How it works At query time the system builds a context package: selected canvas elements, metadata from linked objects, and retrieved passages from connectors. Prompt templates instruct the model to prefer that package, cite when possible, and admit uncertainty when context is missing. Some features add tool use — fetching a fresh file from GitHub or re-querying a database — but the principle stays the same: ground before generate. User actions shape awareness. Framing a diagram, linking a Notion page, or tagging a decision card tells the retriever what belongs in scope. Without those cues, context-aware AI devolves into generic chat. - Selection and framing on visual workspaces - Connector retrieval with permission checks - Prompt policies that prioritize grounded snippets - Explicit uncertainty when context is incomplete ## Examples Ask: "List open risks for the launch frame." Context-aware AI reads risk cards on the canvas and linked Slack threads — not generic launch checklists from the web. Ask: "Does this sequence diagram match the repo?" With GitHub connected, AI compares diagram labels to module names and flags mismatches. Agent task: "Expand this partial C4 diagram with suggested boundaries." The agent uses existing boxes and connector metadata rather inventing a greenfield architecture. ## Common approaches and problems Fake context awareness — long system prompts claiming memory without retrieval — erodes trust quickly when users test specific facts. Over-broad retrieval pulls irrelevant chunks and produces verbose, hedged answers. Users then assume AI "does not work" when the context pipeline is misfiring. Privacy mistakes happen when awareness crosses project boundaries. Separate workspaces per client or initiative; never connect personal inboxes to shared boards without thought. - Marketing memory claims without retrieval backing - Whole-drive search with no project scoping - Shared boards with over-connected personal accounts ## How Drawsy relates Drawsy is explicitly not "just a chatbot." Q&A and agents read canvas structure plus authorized connector data. You control connectors; basic tiers include a subset, Pro unlocks all connectors per the pricing page. Try context-aware workflows at https://drawsy.tech/canvas: sketch a small system, link one doc or repo, frame the diagram, then ask a narrow question and check the answer against your source. Drawsy does not claim SOC 2 or infinite memory. Context is scoped, revocable, and subject to model limits. For help scoping connectors, see /#connectors or email hello@drawsy.tech. ## Related concepts Context-aware AI depends on context engineering and connected context. Compare with generic chat in the AI workspace vs chatbot article on the blog. --- # What Are AI Coding Agents? > AI coding agents are systems that plan and execute software tasks with tool access — they work best when they can see repositories, architecture, and decisions, not just isolated files. ## Definition AI coding agents go beyond autocomplete. They can search a codebase, edit multiple files, run commands, open pull requests, and iterate toward a goal you describe. They combine language models with tools, policies, and often human approval gates. Agents differ from chat assistants in autonomy and scope. An assistant answers a question; an agent may pursue a multi-step plan — refactor a module, add tests, or scaffold a feature — while checking its work against the repo. Effective coding agents still need context: which service owns a boundary, what the team decided about APIs, and what is out of scope. That is why agents increasingly appear inside workspaces that link code, diagrams, and docs. ## Why it matters Software work is more than typing. It is navigation, trade-offs, and alignment with existing patterns. Agents that only see the current file recreate wheels or violate conventions because they miss the map. When agents share context with humans — architecture sketches, ADRs on Notion, Slack decisions — they reduce rework. The agent's plan can be reviewed on the same board the team uses for planning. Autonomy also raises stakes. Unguarded agents can delete tests, leak secrets into logs, or ship changes nobody reviewed. Teams need boundaries: branch policies, sandbox environments, and clear tasks. - Code tasks require repository and architectural context - Shared workspace memory improves agent plans and human review - Autonomy demands guardrails — review, sandboxes, scoped tokens ## How it works A coding agent loop typically looks like: receive goal → retrieve relevant files and docs → propose plan → use tools (read, write, search, terminal, browser) → verify → report. Some loops pause for human approval before writes. Context comes from embeddings over the repo, explicit file picks, issue trackers, and increasingly visual architecture maps. Connectors to GitHub (or local git) are table stakes; connectors to specs and diagrams reduce wrong assumptions. Drawsy Agents focus on workspace-grounded tasks — expanding diagrams, analyzing options, summarizing decisions — rather than replacing your IDE. Pair an IDE agent for file edits with a workspace agent for planning and context-heavy questions. - Goal decomposition and tool calls - Retrieval over repositories and linked docs - Human-in-the-loop approval for risky operations - Verification via tests, linters, or diff review ## Examples Migration planning: an agent reads a service diagram on the canvas and a GitHub monorepo, then proposes an incremental extraction order — humans adjust on the board before any IDE agent cuts code. Onboarding: a new engineer asks an agent how authentication flows work; retrieval pulls the auth ADR from Notion, the sequence sketch, and the relevant package in the repo. Incident follow-up: linked Slack thread and architecture frame help an agent draft a postmortem outline with affected components — not a generic template. ## Common approaches and problems IDE-only agents excel at local edits but may ignore product context sitting in Notion or Figma. Workspace-only agents excel at planning but may not touch git. Many teams need both with a shared map. Failure modes include infinite loops, silent partial fixes, and hallucinated APIs. Always review diffs; run CI; never grant production credentials to experimental agents. Cost and rate limits matter. Long agent runs burn tokens and API quotas. Scope tasks narrowly and cache retrieved context where products allow it. - IDE silo — great edits, weak product context - Unreviewed autonomy — shipping risky changes - Vague goals — agents wander or over-edit - Missing diagrams — wrong service boundaries assumed ## How Drawsy relates Drawsy Agents operate on canvas plus connected context — GitHub, Notion, Slack, Drive, Gmail, Figma, PostgreSQL, HydraDB — so coding discussions stay tied to sources. They help research, structure, and analyze; they are not a replacement for your editor's agent or CI. Connect a repo, sketch boundaries on the board at https://drawsy.tech/canvas, and run agents to expand structure or surface open questions before hands-on coding. Agent runs vary by plan; there is no unlimited agent tier unless pricing explicitly says so. Drawsy does not claim agents always produce correct code. Use them to accelerate thinking and preparation, then verify in your toolchain. ## Related concepts Coding agents sit at the intersection of context engineering and connected context. See the blog posts on why agents need more than code and how agents understand a codebase. --- # What Is Visual Thinking? > Visual thinking uses diagrams, maps, and spatial layout to explore problems — making relationships, gaps, and trade-offs visible before committing to words or code. ## Definition Visual thinking is the practice of externalizing cognition on a surface: whiteboards, paper, sticky notes, or digital canvases. You draw boxes for concepts, arrows for flows, clusters for themes, and lanes for time or ownership. The surface becomes a shared thinking aid, not just decoration. It complements verbal and textual reasoning. Some problems — system boundaries, user journeys, option trade-offs — are easier to see than to narrate. Visual thinking makes those structures explicit so teams can critique them. In software and product work, visual thinking spans napkin sketches, C4 diagrams, story maps, and research synthesis boards. The medium changes; the goal is the same: make the invisible structure visible. ## Why it matters Alignment meetings fail when participants hold different mental models. A shared diagram forces concrete questions: Is this arrow synchronous? Does this box own data? Which user path is in scope? Visual artifacts also age better than meeting memory. A photo of a whiteboard fades; a live canvas with links to sources stays queryable — especially when AI can read the board. For remote and async teams, visual thinking reduces the need for synchronous re-explaining. New contributors start from the map. - Surfaces hidden assumptions in group discussions - Creates durable artifacts beyond meeting notes - Supports async collaboration across time zones ## How it works Effective visual thinking cycles between divergent and convergent modes. Diverge: sketch many boxes, explore alternatives, avoid premature neatness. Converge: frame the decision, label owners, link evidence, archive alternatives. Notation can be informal. Hand-drawn style often lowers perfectionism and speeds iteration. What matters is consistent semantics: solid vs dashed lines, color meaning, legend for abbreviations. Digital AI workspaces add a retrieval layer: diagrams stay linked to docs and code, and questions can reference frames. That closes the loop between thinking visually and acting in tools. Facilitation matters in group settings. Time-box divergent sketching, then force a converge step with explicit decision owners. Visual thinking without ownership becomes wallpaper. - Diverge — explore breadth on the canvas - Converge — frame decisions and scope - Annotate — legends, owners, and status - Link — connect visuals to sources and tasks ## Examples Story mapping: horizontal user steps, vertical tasks underneath, revealing MVP slices versus later releases. Architecture sketching: boxes for services, cylinders for data stores, numbered flows for critical paths before formal diagrams. Research synthesis: affinity clusters of interview quotes with severity tags, making patterns obvious to stakeholders who will not read fifty pages of notes. Decision records: two option columns with explicit trade-off arrows and a framed winner — easier to revisit than a paragraph in meeting notes. Learning maps: concept nodes with prerequisite edges for onboarding complex domains like payments compliance or ML pipelines. ## Common approaches and problems Workshop theater: beautiful boards that never connect to execution. Visual thinking fails when diagrams are not linked to tickets, repos, or decisions. Over-formalization too early: spending hours on perfect UML before understanding the problem. Start messy; formalize when stability appears. Tool fragmentation: Miro for workshops, FigJam for design, Excalidraw for eng, Notion for text — each holds a fragment. Without connected context, teams rebuild maps repeatedly. - Orphan diagrams — pretty, not operational - Premature polish — slows learning - Tool sprawl — maps duplicated across products ## How Drawsy relates Drawsy is an AI workspace built for visual thinking plus connected context. The infinite canvas supports fast sketches; connectors attach the docs and repos that give sketches teeth; AI helps summarize, expand, and question the map. Start visually at https://drawsy.tech/canvas without an account wall. When a diagram stabilizes, sign in to save, share, and collaborate per your plan. Drawsy is not claiming visual thinking replaces written specs or code. It keeps visuals, context, and AI in one loop: Draw → Connect → Understand → Create. ## Related concepts Visual thinking powers AI canvases and workspaces. Pair with the software architecture visualization blog post and use-case hubs for hands-on patterns. --- # What Is Connected Context? > Connected context is project information living in authorized tools — docs, code, chat, files, data — linked into the workspace so humans and AI share the same sources. ## Definition Connected context means your workspace can reach outward to tools where work already lives, under permissions you control, and bring relevant slices into the same place as your diagrams and decisions. It is the Connect step in Draw → Connect → Understand → Create. Connection is not bulk duplication. Good systems index or fetch on demand, respect OAuth scopes, and tie external objects to canvas elements so relationships stay explicit. Examples of connected context include a GitHub repository linked to a service box, a Notion PRD linked to a milestone frame, or a Slack thread linked to an open question card. ## Why it matters Teams waste hours rebuilding context: copying links into chats, re-uploading PDFs, or summarizing meetings that already happened in Slack. Connected context makes the workspace the hub instead of the chat thread. For AI, connected context is the difference between plausible generic answers and answers that reflect your repo, spec, or schema. Without it, models guess. For compliance-minded teams, explicit connectors beat shadow IT paste workflows. You can audit what is connected, disconnect sources, and scope access per project. - Eliminates repetitive copy-paste into AI chats - Improves factual grounding for Q&A and agents - Makes access auditable compared to ad-hoc uploads ## How it works Typical flow: user authorizes a connector → workspace stores tokens with least privilege → user links an external object to a canvas element or project → retrieval layer fetches snippets when questions reference that link. Drawsy supports HydraDB for structured project data plus GitHub, Notion, Slack, Google Drive, Gmail, Figma, and PostgreSQL. Basic connectors appear on Free; full connector access unlocks on Pro and above per the pricing page. Refresh strategies vary by connector. Code and chat change quickly; Drive files change moderately. Users should know when snapshots might be stale and re-sync when making decisions. - OAuth authorization per user or team - Explicit linking — context follows intent - Retrieval at query time with token budgets - Plan-gated connector availability ## Examples Engineering: link the payments repo to the payments service on an architecture map. Questions about retries pull code and the diagram together. Product: link a Notion discovery doc to interview clusters on the board. AI summarizes themes with pointers to original quotes in Notion. Design: connect a Figma file to a user journey frame so engineering questions about UI states reference the actual design, not verbal descriptions. Operations: attach a PostgreSQL schema to a data-flow diagram before a compliance review. Questions about retention reference columns and storage nodes you labeled — reducing generic policy language. Leadership: link board meeting notes from Drive to decision frames so quarterly planning AI answers cite the same documents executives already approved. ## Common approaches and problems Manual link lists in wikis rot quickly. Without live connectors, humans forget to update URLs after moves and renames. Over-connecting personal accounts to shared boards creates leakage risk. Use team accounts and project-scoped drives where possible. Assuming connectors imply perfect sync is dangerous. Retrieval can miss files, mis-parse formats, or truncate large PDFs. Always spot-check critical facts. Another pattern is treating connectors as backup storage. They should index and reference — not replace authoritative systems of record. If Notion owns the PRD, edit it there; the workspace should reflect changes, not fork a shadow copy teams stop trusting. - Static link rot in wikis - Personal vs team account mixing - Stale or truncated connector snapshots ## How Drawsy relates Connectors are a core product surface, not an afterthought. Drawsy keeps connected context beside the canvas so Ask Drawsy and Agents can use authorized sources alongside visual structure. Explore connectors on the home page at https://app.drawsy.tech/#connectors and try linking one source after opening https://drawsy.tech/canvas. Start narrow — one repo or doc — before connecting entire drives. Drawsy does not sell your data or claim certifications it has not published. You can disconnect sources; guest content may be ephemeral until sign-in. Questions: hello@drawsy.tech. ## Related concepts Connected context enables context-aware AI and effective context engineering. Read those definitions alongside the home Connectors section (/#connectors) and the pricing page for scope details. --- # Context Engineering for AI Agents: How to Give AI the Right Information > Learn how context engineering helps AI agents use the right instructions, knowledge, tools, memory, and project context to produce more reliable results. ## Why agent quality starts with context AI agents are getting better at reasoning, using tools, and completing multi-step tasks. But there is a problem that becomes more obvious as agents move beyond simple chat: **An agent can only work with the context it has.** Give it the wrong files, stale documentation, incomplete requirements, irrelevant conversation history, or poorly structured tool output, and a capable model can still make the wrong decision. That is where [context engineering](/context-engineering) comes in. Context engineering is the practice of designing the information an AI system receives so it has the right information, in the right form, at the right time. For an AI agent, that can include instructions, retrieved knowledge, conversation history, tools, tool results, memory, project state, and constraints. The goal isn't to give an agent **more** context. It's to give it **better** context. ## What is context engineering? Context engineering is the process of deciding **what information an AI agent should receive, how that information should be structured, and when it should be provided**. A simple agent might look like: *[Visual diagram: simple-agent]* ### What a real project agent needs A useful agent working on a real project often needs much more than a single prompt. The model is only one part of this system — the surrounding information determines what it can actually reason about. *[Visual diagram: agent-context-stack]* ## Context engineering vs prompt engineering Prompt engineering usually focuses on **how you phrase instructions to a model** — for example, “Review this function and explain the bug.” Context engineering asks a larger question: **What should the model know before it reviews the function?** Maybe it needs the relevant source files, the function's callers, database schema, API contracts, existing tests, recent changes, architecture decisions, error logs, project conventions, and the user's actual goal. Prompt engineering still matters. But for agents that operate across multiple steps, the context pipeline becomes part of the system design. | Prompt engineering | Context engineering | | --- | --- | | How should I instruct the model? | What should the model know? | | Focuses on instructions | Focuses on the information environment | | Usually one interaction | Can span an entire workflow | | Prompt wording | Retrieval, memory, tools, state and instructions | | "What should I ask?" | "What should the agent have access to?" | ## Why context matters more for AI agents A chatbot can sometimes answer a question with general knowledge. An agent is different. It may have to understand a goal, inspect information, choose a tool, perform an action, inspect the result, update its plan, act again, and validate the outcome. Every step creates or consumes context. Consider a coding agent asked: “Add authentication to the application.” That request is nowhere near enough information to safely implement the feature. The agent may need to understand the repository layout — frontend, backend, database, authentication, API routes, tests, environment configuration, and deployment — plus which auth provider is already used, how users and sessions work, which routes are protected, existing security conventions, deployment constraints, and previous architecture decisions. Without that context, the agent can produce code that looks reasonable while being wrong for the actual system. This is exactly why [AI coding agents](/ai-coding-agents) fail when they only see a truncated slice of the codebase. *[Visual diagram: repo-tree]* ## The five useful layers of agent context There isn't one universal context architecture for every agent. But a useful way to think about the problem is through several layers. ### 1. Instructions These establish what the agent is supposed to do: role, task, constraints, output format, safety requirements, and project rules. For a coding agent, instructions might look like: ```text Follow the existing TypeScript conventions. Do not introduce a new authentication library. Run the existing test suite after changes. Do not modify database migrations unless required. ``` - These instructions establish boundaries — not the full project truth. ### 2. Knowledge Knowledge gives the agent information about the world or project: documentation, databases, repositories, APIs, files, knowledge bases, search, and retrieved documents. The important part is **relevance**. Dumping an entire knowledge base into a context window doesn't automatically make an agent smarter. The system needs to identify what matters for the current task — the heart of [connected context](/connected-context). ### 3. State and memory Agents often need information that persists across interactions — for example, a prior decision to use PostgreSQL because analytics already depends on it. That decision can be much more valuable to a future agent than another generic explanation of PostgreSQL. Memory therefore isn't simply “remember everything.” Good memory preserves information that remains useful. *[Visual diagram: decision-memory]* ### 4. Tools An agent may have access to GitHub, databases, APIs, file systems, browsers, terminals, deployment systems, or observability tools. Tool access creates another context problem. The agent needs to understand what the tool does, when to use it, what inputs it accepts, what its output means, and what permissions it has. A powerful tool with poor context can still produce poor results. ### 5. Execution feedback An agent also learns from what happens after it acts. Test failures, error logs, and validation results become part of the working context for the next step. This is one reason long-running agents require more deliberate context management than simple chat interactions. *[Visual diagram: feedback-loop]* ## More context isn't always better This is one of the most important ideas in context engineering. Imagine asking an AI coding agent to fix a login bug and giving it the entire repository, every GitHub issue, every Slack message, all logs from the last year, every deployment, the entire documentation site, and hundreds of previous conversations. Technically, the agent has **more** context. Practically, it may have a harder problem. Some of that information is irrelevant, stale, contradictory, duplicated, noisy, or expensive to process. A better system might retrieve the login service, authentication middleware, relevant database model, recent auth-related commit, failing test, and current error log. That is a smaller context — and much more useful. This is why context engineering is not simply about increasing the context window. **It is about selecting and structuring the right context.** ## Context engineering for coding agents Coding agents make this problem especially visible because software systems contain a huge amount of implicit context. A developer might know: “We don't use that service here because it caused a race condition two years ago.” That knowledge may live only in an old pull request, a design document, a Slack conversation, a code comment, or a teammate's memory. The code itself may not explain the decision. An AI agent looking only at the current files can therefore miss an important constraint. A stronger coding-agent context might look like: *[Visual diagram: coding-agent-pipeline]* - The goal is not to expose the entire history of a project. - The goal is to expose the history that explains the current system. ## A practical context-engineering workflow If you're building an AI agent today, start with the task rather than the model. ### Step 1 — Define the task What exactly is the agent trying to accomplish? Bad: “Improve the application.” Better: “Add pagination to the `/api/projects` endpoint without changing the existing response shape.” ### Step 2 — Identify required context Ask: What would an experienced engineer need to know before making this change? ```text API implementation Database query Existing response type Frontend consumer Tests Pagination conventions ``` ### Step 3 — Retrieve only relevant information Don't automatically load everything. Use semantic retrieval, keyword search, metadata filters, repository structure, dependency relationships, and task-specific routing. ### Step 4 — Separate facts from instructions This distinction makes context easier for an agent to interpret. *[Visual diagram: facts-constraints]* ### Step 5 — Preserve important decisions When an architectural decision is important, record the decision, why it was made, what it affects, and what it replaces. This prevents future agents from repeatedly reconsidering already-settled decisions. ### Step 6 — Validate the result Context engineering doesn't end when the model generates an answer. For an engineering agent, plan → implement → test → inspect failures → update context → fix → validate. The validation result becomes part of the next decision. ## A simple context architecture For many AI applications, a useful starting point makes one thing obvious: **context is a pipeline, not a text box.** *[Visual diagram: context-builder]* ## How to know whether your context is good Before sending context to an agent, ask five questions: ### Is it relevant? Does this information help with the current task? ### Is it sufficient? Does the agent have enough information to make the decision? ### Is it current? Could some of the information be outdated? ### Is it consistent? Are there conflicting instructions or sources? ### Is it traceable? Can you understand where an important piece of information came from? These questions are especially important for agents operating on real systems. ## The future of AI agents may depend less on bigger prompts As AI models become more capable, the interesting engineering problem increasingly shifts outward. Not only “How intelligent is the model?” but “What can the system give the model to work with?” An agent with a strong model but poor context can still make bad decisions. An agent with strong context can often make much better use of the same underlying model. That's why context engineering is becoming an important part of building reliable AI agents. The model matters. But the information environment around the model matters too. ## Where AI workspaces fit This is also where the idea of an [AI workspace](/ai-workspace) becomes interesting. Instead of keeping ideas in one tool, documents in another, code in GitHub, decisions in Slack, diagrams on a whiteboard, and AI conversation in a chat window, a workspace can bring relevant pieces of that context together. The goal isn't simply to give AI access to more information. It's to make the relationship between the information visible and useful. For visual and software-development workflows, that can mean connecting idea → architecture → documentation → code → decisions → AI → implementation. This is the direction Drawsy is exploring: an AI workspace where visual thinking, connected context, and collaboration can work together rather than living in completely separate places. You can try the canvas at https://drawsy.tech/canvas, or read [how to start using Drawsy](/how-to). Learn more about [Drawsy agents](/agents) when you want AI that works with the board — not beside it. *[Visual diagram: workspace-flow]* ## Final takeaway Context engineering isn't about writing the perfect prompt. It's about designing the **information environment around an AI system**. For an agent, that environment can include instructions, project knowledge, retrieved information, memory, tools, tool results, constraints, and execution feedback. The goal is simple: **Give the agent the information it needs, when it needs it, in a form it can use.** As agents take on longer and more complex tasks, that may become just as important as choosing the model itself. ## Related reading - [/context-engineering](/context-engineering) - [/ai-coding-agents](/ai-coding-agents) - [/ai-workspace](/ai-workspace) - [/connected-context](/connected-context) - Start on the canvas: https://drawsy.tech/canvas --- # Drawsy Team > Product We build Drawsy — a canvas that thinks with you. Visual thinking, connected context, and AI agents in one workspace. Expertise: AI workspaces, visual thinking, connected context, developer workflows Articles: https://app.drawsy.tech/blog Profile: https://app.drawsy.tech/authors/drawsy-team --- # Optional # Drawsy Changelog > What is shipping. Release notes for canvas improvements, AI, connectors, auth and account features will be listed here as they ship.