Circleback CEO Ali Haghani: Why Your Company Should Be Recording More Meetings

Company & Product Overview

• Circleback is an AI note-taking platform in the YC Winter 24 batch that records, transcribes, and organizes conversations.
• The tool automatically generates notes, assigns action items, and enables custom automations to update CRMs, issue trackers, and Slack.
• It functions as a "company brain" allowing users to query across all team conversations and connect custom agents.

Hardware & Office Setup

• Founder Ali uses a Topre membrane keyboard (Tokyo Precision) for its tactile feel, quiet operation, and early key registration, avoiding loud mechanical switches in open offices.
• He utilizes a "Timu" fidget toy kept after initially considering a full purchase due to the product's engaging onboarding design.
• AirPods Max are worn primarily as a visual signal of being "locked in" rather than for active noise cancellation.

Circleback Usage Patterns

• Recruitment as ATS: The founder uses the action item view to track interview candidates, manage interviewer assignments, and review full communication timelines (emails and meetings).
• Partner Management: The tool maintains unified views of contacts at partner organizations (e.g., Y Combinator) to track action items, conversation history, and upcoming office hours.
• Meeting Preparation: Circleback was used to prep for interviews by synthesizing past emails and conversations with the interviewer to generate a briefing and identify key talking points.
• Quick Capture: A dedicated hardware button triggers Circleback recording; a watch app is currently in beta testing to allow start/stop functionality without a phone.

Workflow & Tool Usage

• Open/Claw: The founder spends 10–20% of his day in Open/Claw, utilizing specialized agents for Finance, Customers, and People Ops.
• Slack Migration: Teams are migrating automation workflows from Open/Claw to Slack channels to increase visibility and foster collective handling of customer and sales inquiries.
• Support Rotation: Engineers participate in a weekly support rotation to engage directly with customers and capture feedback in real-time.
• Coding Shift: The founder has reduced terminal usage to focus on code review and high-leverage product decisions rather than writing code manually.
• Coding Tools: Reliance has shifted from manual coding to using AI orchestrators to generate code, with the founder intervening only for small styling changes or copy edits.
• No-Meeting Day: Thursdays are dedicated to deep work and end-to-end shipping of projects without interruptions.

AI Governance & Evaluation

• Evals Strategy: The team employs evaluations for quantifiable tasks (e.g., identifying action items) and qualitative writing style to prevent filler words like "discussed."
• Incremental Improvements: Systems are in place to test incremental model changes to avoid "whack-a-mole" issues where improvements in one area degrade another.
• Token Usage: The company currently maintains an unlimited AI budget to prioritize output quality over cost, trusting team judgment.
• Safety Constraints:
  • Agents are prohibited from sending emails autonomously (drafts are permitted).
  • Agents are not allowed to architect systems; engineers must define the architecture.
  • Agents handle data access for review but require human oversight for security-sensitive actions.

Strategic Thesis & Industry Outlook

• Recording Everything: As LLMs and agents become central to operations, recording all conversations is becoming a necessity to provide context and prevent agents from working in isolation.
• Context Sharing: Future software will likely default to sharing conversation context across teams while maintaining strict privacy controls to prevent inadvertent data leaks.
• Software Engineering Future: The barrier to entry is shifting from technical bandwidth to strategic decision-making; lower coding costs make the "what" and "how" of building more critical than speed alone.
• Operational Automation: Founders are increasingly automating low-leverage operational tasks (e.g., payroll logging) to focus on high-impact activities.