--- url: https://pencel.ai/docs/getting-started/why-pencel.md description: >- Why Pencel — build AI agents and governed workflows with no code, curate your own knowledge and self-learning memory, choose from multiple models, and act on what matters, all on a local-first desktop app. --- # Why Pencel? Pencel is a **local-first desktop app** that organizes your recurring business operations — research, reporting, briefings, monitoring, and follow-through — into Programs, each with its own room to work in. It is built for the person who knows the work but doesn't write code. Here's what makes it different. ## Work happens in a Program room Open a **[Program](/concepts/programs)** — RFP tracking, inbox triage, or one you set up yourself — and you get a room: a Slack-style conversation with the AI teammate that owns that work. Talk to it, drop in a file, tell it a standing rule. It drafts, finds, and reports back right there, the way a colleague would. ## No agent or workflow editor required — but it's there if you want it Behind every room is an agent (its role, instructions, and knowledge) and, for anything that should run the same way every time, a **[Workflow](/concepts/playbooks)** Pencel compiles from your conversation into concrete, ready-to-run steps — with approval gates before anything is sent, read-only boundaries, and cost ceilings. Most people never open these directly; a room can save a good reply as a recurring workflow with one click. If you want to configure any of it by hand, it's all there in **Studio**. See [Agents](/concepts/agents), [Approval workflows](/privacy-and-security/approval-workflows), and [Creating your first agent](/guides/creating-your-first-agent). ## Curate your own knowledge — from a local folder Point Pencel at a folder on your machine and it imports your documents into a searchable **[Knowledge](/concepts/knowledge)** base. From there it builds a **knowledge graph** of the people, projects, and facts it finds, and links them to the source files. Your agents draw on this when they work — grounded in *your* material, not the open web. Files stay on your computer. ## Self-learning memory Agents accumulate **[Memory](/concepts/memory)** as they work — preferences, facts, and decisions worth remembering — so they get more useful over time instead of starting from scratch each run. You can review, edit, and prune what they remember. ## Multiple models — including an open-weight option Pencel isn't locked to one provider. Choose **Claude** (Opus 5, Sonnet 5), **GPT** (5.6 Sol, 5.6 Terra), or **Gemini** (2.5 Pro, 2.5 Flash). The managed **Auto** and **Private** plans can also route to an **open-weight model** (GLM-5.2) when it's the right fit. Match the model to the job — a cheap fast model for routine scans, a frontier model for hard reasoning. See [Supported models](/reference/supported-models). ## Quick-start industry solutions **[Solution Packs](/concepts/solutions)** are curated bundles — a crew, connections, and guidelines for a specific job (federal RFP scouting, competitive intel, and more). Install one and you have a working Program in minutes instead of building from a blank room. ## Manage to-dos as agent context **[To-dos](/concepts/todos)** aren't just a checklist — they're working context for your agents. Capture a task (type it, or use the `/todo` quick form), and an agent can pick it up, act on it, and report back. See [Using to-dos](/guides/using-todos). ## Create and edit real documents Agents produce editable **[Artifacts](/concepts/artifacts)** you can keep working on: documents and reports, notes and drafts, **spreadsheets**, **slide decks**, **HTML pages**, and email drafts. Edit them in-app, or ask an agent to revise them in place. ## See what matters on Home Results post first in the room that produced them, and also roll up to **[Home](/concepts/operations)** — your surface for decisions and things waiting on you, plus a running feed of briefings that group related items so the important topics rise to the top. Opt into **[Radar](/concepts/radar)** and Pencel quietly scans in the background and posts insight cards about things worth your attention (read-only — it never acts without you). ## Local-first and private Your workspace — documents, agents, runs, and knowledge — lives in a local SQLite database on your machine and is **never synced to Pencel's servers**. Only model calls leave your computer, and on **Private** mode they route through a zero-retention proxy. See [Data handling](/privacy-and-security/data-handling). ## Where Pencel is headed Today Pencel is a **single-user** desktop app. Team features — sharing agents and knowledge in a collaborative workspace — are on the roadmap. For now, curated Solution Packs are the way to share a working setup across people. ## Start here * [Programs & Rooms](/concepts/programs) — what a Program is and what working in a room feels like * [Quickstart](/getting-started/quickstart) — get to a first useful result in a few minutes * [What is Pencel?](/getting-started/what-is-pencel) — the concepts, in one page * [Installation](/getting-started/installation) — download, sign in, and choose how to power your models