Next cohort

Intro to AI Evals with Reforge · starts Oct 12, 2026

Sign Up →
About Us

Work with expert product practitioners.

Calibre Labs is an independent AI advisory firm. We are AI builders and product practitioners — not influencers.

The firm

Two operators who have run this playbook themselves.

Calibre Labs was started by Sandhya Hegde and Justin Bauer, former executives at Amplitude, where they led the company from Series A to IPO. Both are Stanford GSB graduates with backgrounds in engineering and data science. Both lead every engagement directly.

Founders
Sandhya Hegde

Sandhya Hegde

Co-founder

Founder, investor and operator. Started her first company while studying engineering at IIT Bombay. After its acquisition she was a VC at Sequoia Capital and Khosla Ventures, then an executive at Amplitude from Series A to IPO. Board member at several AI-native startups. Writes on AI product quality and agent design.

Justin Bauer

Justin Bauer

Co-founder

Former Chief Product Officer at Amplitude, where he oversaw 100X growth in ARR and helped pioneer the discipline of product analytics. Previously a founder and an engagement manager at McKinsey.

Who we work with

Public software companies and high-growth AI startups.

Enterprise
  • Salesforce
  • Atlassian
  • Amplitude
High growth
  • Vapi
  • Monarch

We are also the official AI Product course partners on Reforge, where more than 20,000 PMs, designers and growth professionals learn AI-native practices.

What we believe

An AI-native product org is a system, not an L&D exercise.

Individual re-skilling has plenty of options. The harder work is building the system inside the company.

  • Prototypes over specs

    Working demos instead of documents. The PRD gets written after the prototype.

  • Shared context

    Team knowledge lives where an agent can read it, not in someone's head.

  • Skills get promoted

    One person's workflow becomes a team skill, then an org-wide standard.

  • Phased rollout

    A few pilot teams first, with explicit leadership expectations.

  • Measure or it decays

    Every workflow has a named owner and a review cycle.

  • Quality is defined first

    If nobody said what good means, no eval tooling will rescue it.

Tell us what's top of mind for you.

We reply to everything. If we aren't the right fit we'll say so.

Contact us →