Acta

Paste notes, get a prioritized action list.

The Problem

Productivity tools are bloated. Most task managers demand that you create projects, assign categories, set up boards, and learn a workflow before you can track a single task. The irony: the overhead of organizing your tasks becomes a task itself.

The real friction happens at the point of capture. You leave a meeting with scattered notes, a brainstorm session produces a wall of text, a voice memo captures ten ideas but no structure. The gap between raw input and actionable output is where things fall through the cracks. People default to sticky notes, plain text files, or simply trying to remember -- and tasks get lost.

What is needed is not another project management tool with 50 features. It is a single-purpose tool that does one thing well: turn unstructured text into a prioritized list of concrete actions, instantly.

Our Solution

Acta is a zero-overhead productivity tool. Paste any unstructured text -- meeting notes, brainstorms, voice memo transcripts, email threads -- and Acta uses Claude AI to extract concrete tasks, assign priority levels, and suggest deadlines. The output is a clean, actionable checklist ready to work through.

There are no projects to create, no categories to configure, no learning curve. The interface is deliberately minimal: a text input and a generated checklist. Acta respects the user's time by eliminating every unnecessary step between "I have notes" and "I know what to do next."

A free tier offers 20 transformations per month. Users can bring their own Anthropic API key for additional usage. A Pro tier with unlimited transformations and cloud sync is on the roadmap.

Architecture

Acta is a Next.js application deployed on Vercel with Supabase as the backend. The architecture is intentionally lean -- no microservices, no message queues, no infrastructure complexity. The entire value proposition depends on speed and simplicity, and the architecture reflects that.

The core flow is straightforward: the user submits text through the frontend, which sends it to a Next.js API route. The API route constructs a prompt with the user's input and sends it to the Anthropic Claude API. Claude processes the text and returns structured task data -- each task with a description, priority level, and suggested deadline. The frontend renders the result as an interactive checklist.

Supabase handles user authentication and persists transformation history, allowing users to revisit past checklists. The database schema is minimal: users, transformations, and tasks.

Key Design Decisions

• Single-purpose over feature-rich: Every feature request is evaluated against the core promise of zero overhead. If it adds a step, it does not ship.
• Claude for extraction: LLM-based extraction outperforms rule-based parsing for unstructured text. Claude understands context, infers deadlines from phrases like "by end of week," and distinguishes tasks from observations.
• Streaming responses: Task extraction streams to the frontend via the Vercel AI SDK, so users see results appearing in real time rather than waiting for a full response.
• Bring-your-own-key model: Lets power users bypass tier limits while keeping the free tier sustainable.

Tech Stack

Code Patterns

// Notes-to-actions extraction via Claude
interface ExtractedTask {
  description: string;
  priority: "high" | "medium" | "low";
  suggestedDeadline: string | null;
}

interface ExtractionResult {
  tasks: ExtractedTask[];
  summary: string;
}

async function extractActions(rawNotes: string): Promise<ExtractionResult> {
  const response = await anthropic.messages.create({
    model: "claude-sonnet-4-20250514",
    max_tokens: 1024,
    system: `You are a task extraction assistant. Given unstructured notes,
extract concrete, actionable tasks. For each task:
- Write a clear, specific description
- Assign priority: high (urgent/blocking), medium (important), low (nice-to-have)
- Suggest a deadline if one is implied or can be reasonably inferred
Return structured JSON only.`,
    messages: [{ role: "user", content: rawNotes }],
  });

  return parseExtractionResponse(response);
}

// Streaming extraction to the frontend
import { streamText } from "ai";
import { anthropic } from "@ai-sdk/anthropic";

export async function POST(request: Request) {
  const { notes } = await request.json();

  const result = streamText({
    model: anthropic("claude-sonnet-4-20250514"),
    system: EXTRACTION_PROMPT,
    messages: [{ role: "user", content: notes }],
  });

  return result.toDataStreamResponse();
}

Outcomes

• Sub-3-second extraction from paste to prioritized checklist, with streaming results appearing in real time.
• Zero onboarding friction: No signup required for first use. Users paste notes and get results immediately.
• 20 free transformations per month on the free tier, with BYOK (bring your own key) for unlimited usage.
• Minimal infrastructure cost: The lean architecture keeps operational costs near zero, with Vercel and Supabase free tiers covering the base load.
• High extraction accuracy: Claude consistently identifies actionable items, distinguishes tasks from context, and infers reasonable deadlines from natural language cues.