CHAPTERS
- 0:00 – 2:06
Mission-driven career choices: impact over prestige (eBay → Facebook → Instacart)
Fidji Simo explains the throughline behind her career moves: working on technologies that solve real problems for real people. She connects each transition to a personal resonance with the mission—community commerce at eBay, staying connected as an immigrant via Facebook, and saving families time and stress through better access to food via Instacart.
- •Defines herself as a “pragmatic technologist” focused on real-world outcomes
- •eBay: commerce as passion/community exchange
- •Facebook: personal utility for maintaining family connections across borders
- •Instacart: food as a core life need; time back for families
- •Choosing roles based on where she can have the most impact
- 2:06 – 3:48
Authentic leadership and resisting Silicon Valley monoculture
Simo discusses how she learned to lead without conforming to Silicon Valley’s expected aesthetic or cultural norms. She describes an early attempt to “blend in” and the realization that performing an identity wastes energy better spent building products.
- •Pressure to conform (remove makeup, wear a hoodie) when moving into product roles
- •Anecdote: blending in for a day led to being unrecognized
- •Decision to embrace her identity (accent, style) as a leadership strength
- •Principle: allocate energy to product and team impact, not self-editing
- •Modeling authenticity makes it easier for others to adapt and accept differences
- 3:48 – 5:48
Earning followership: high standards paired with deep support
Asked how she builds loyalty while being tough, Simo outlines a leadership contract: she will push people to excel with candid feedback while backing them fully. She emphasizes fit—this style isn’t for everyone, and that’s okay.
- •Leadership as a partnership: she aims to accelerate leaders’ careers
- •Two-part contract: high standards + tough feedback
- •Second part: strong advocacy, compassion, and psychological safety
- •Not everyone opts in; selecting for people who want to stretch and excel
- •Outcome: teams doing the best work of their lives drive exceptional results
- 5:48 – 8:23
Taking over Instacart after COVID hypergrowth: resetting vision and responsibility
Simo recounts joining Instacart as it absorbed massive pandemic-driven growth and scrutiny about whether demand would persist. Her response was to re-anchor the company around mission, critical-infrastructure responsibility, and a broader vision as a retail-enablement technology provider.
- •Instacart had ~4x growth; rapid scaling of team and operations
- •External doubt: “pandemic darling” vs durable habit change
- •Focus on mission/vision to unify a rapidly grown organization
- •“Vision Summit” to align on responsibility as critical infrastructure
- •Instacart as retail enablement: powering retailer sites (e.g., grocery chains’ dot-com experiences)
- 8:23 – 11:10
Instacart’s AI roots and why generative AI changes the interface to commerce
Simo explains that AI has always been core to Instacart—search, substitutions, and shopper-store matching are ML-driven. The generative AI shift is about expressing intent in natural language, replacing awkward keyword-based shopping with conversation that mirrors how people actually plan meals and budgets.
- •Instacart as a “search engine” at massive scale (high query volume)
- •AI already powers substitutions and operational matching
- •Online commerce historically forced users into keyword thinking
- •Natural language enables intent-rich queries (diet, budget, preferences, constraints)
- •Generative AI aligns software responses with real human decision-making
- 11:10 – 12:32
Implementing genAI inside a large org: problem-first mindset + centralized and federated models
She shares how she guided the organization to adopt genAI by starting from user problems rather than an “AI roadmap.” Operationally, Instacart combines a centralized team that goes deep on capabilities with a federated approach that spreads practical usage across product teams.
- •Prompt to teams: identify longstanding user problems now solvable better with genAI
- •Critique of “AI roadmap” thinking; prioritize biggest problems and new solution paths
- •Centralized expert group to explore what’s newly possible
- •Federated rollout so every team incorporates AI where relevant
- •Balancing depth of expertise with broad product impact
- 12:32 – 15:27
Tactical adoption: hackathons, internal ‘Ava’ assistant, and becoming truly AI-native
Simo describes concrete mechanisms used to build intuition and adoption—company hackathons, an internal GPT-4-based assistant, and a cultural push to redesign workflows rather than bolt AI onto old paradigms. She uses Facebook’s mobile monetization shift as an analogy for how genAI requires rethinking UX fundamentals.
- •GenAI hackathon framed around improving existing problems, not novelty demos
- •Internal “Ava” assistant to make GPT-4 accessible with workflow-friendly shortcuts
- •Analogy: mobile ads weren’t “shrunken desktop ads” (newsfeed ads were a new paradigm)
- •AI-native means rethinking product from the ground up
- •Example: debate over placing “Ask Instacart” inside the sacred search box vs separate surface
- 15:27 – 17:33
Productivity beyond engineering: marketing, legal, and removing ‘soul-crushing’ busywork
Simo argues productivity gains will extend far past coding by automating repetitive, low-satisfaction tasks across functions. She offers examples where AI reduces cost and time while improving creative output and internal operations.
- •Early to quantify, but expectation of broad productivity impact
- •Marketing: generating a large library of animated characters/creative variants with AI
- •Legal: scanning and extracting custom contract terms across many documents
- •AI frees specialists from tedious review work to focus on judgment and strategy
- •Higher job satisfaction as routine tasks are automated and summarized
- 17:33 – 20:34
AI-driven grocery discovery: plugins, intent capture, and Instacart as the ‘bits-to-atoms’ conversion layer
Simo explains how integrations like the ChatGPT plugin revealed higher-funnel, occasion-driven intent (e.g., planning parties) that Instacart historically didn’t capture. She positions Instacart as the conversion layer that turns conversational intent into real-world fulfillment, powered by a unique store network and proprietary data.
- •ChatGPT plugin showed new behaviors: planning occasions, not just restocking items
- •Instacart today is “bottom of funnel”; genAI unlocks upstream discovery
- •Partnerships/experiments across assistant ecosystems (e.g., search/chat interfaces)
- •Differentiator: translating software intent into delivery via an 80k-store network
- •Moat: combining LLMs with unique catalog and behavioral/commerce data
- 20:34 – 22:31
Why robotics and fully automated fulfillment struggle: economics, proximity, and customer speed
Asked about drones and automation, Simo emphasizes that the limitation isn’t always technical—often it’s unit economics and delivery speed. Existing retailer stores represent fixed-cost assets close to customers, making in-store picking and rapid delivery economically compelling compared to distant automated warehouses.
- •If rebuilding grocery from scratch: more automated ‘dark’ stores might exist
- •Reality: current system has many stores with fixed costs already paid by retailers
- •Distant automated warehouses require aggregation but push delivery to next-day
- •Customers value speed; same-day/fast delivery is a key advantage
- •Pragmatic lens: tech must improve cost and/or customer experience to scale
- 22:31 – 24:31
Connected stores and Caper Carts: bringing AI into the physical grocery aisle
Simo details Instacart’s push to digitize in-store experiences through Connected Stores and the acquisition of Caper Carts. Smart carts use computer vision for item detection and enable checkout-from-cart, while also supporting guidance, savings, and personalized suggestions—bridging online and offline retail.
- •Grocery undergoing digital transformation both online and inside stores
- •Acquisition of Caper Carts: computer vision detects items; checkout directly from the cart
- •In-store screens enable recommendations, aisle guidance, and budget management
- •Omnichannel customers are more valuable than online-only or offline-only
- •Rollouts with major grocers; expectation this is the future of grocery shopping
- 24:31 – 26:21
AI’s impact on ads and commerce: hyper-personalization, creative explosion, and new placements
Drawing from Facebook and retail media experience, Simo predicts AI will make ads more relevant through personalization and dramatically more diverse creative generation. She expects new ad placements to emerge in conversational and intent-rich shopping surfaces like Q&A and assistant-driven discovery.
- •AI increases personalization: “perfect ad at the perfect moment”
- •Generative tools enable massive creative variation and improved matching
- •Advertising follows consumer experiences; new UX creates new inventory
- •Conversational commerce surfaces enable intent-based sponsored results
- •Examples: sponsored content within Ask Instacart and similar query flows
- 26:21 – 30:51
Agentic commerce: automating chores while making shopping more inspiring
Simo and the hosts discuss a future where AI agents handle routine purchasing while humans focus on enjoyable, expressive shopping moments. She argues Instacart should evolve from a purely transactional utility into a more inspiring food experience, and that richer conversational interfaces may enhance (not diminish) advertising and engagement.
- •Agents likely automate “chore” purchases while preserving fun browsing categories
- •Instacart today feels transactional; food should be inspiring and connective
- •Opportunity: recipe discovery, new snacks, and planning as engaging experiences
- •More immersive conversational interfaces could increase engagement vs keyword search
- •Advertising may thrive in richer, more inspiring environments
- 30:51 – 34:58
Metrodora and neuroimmune care: blending clinic + research and AI’s role in accelerating cures
Simo explains founding Metrodora after her own chronic illness diagnosis, motivated by slow translation of research into clinical care. She describes how AI can improve diagnostics and drug discovery while arguing strongly against slowing progress due to speculative fears, emphasizing the immediate benefits for patients.
- •Personal catalyst: diagnosis revealed gaps in research focus and patient care
- •Metrodora’s model: integrate clinical care and research; build patient data infrastructure
- •Problem: ~14-year delay from discovery to clinical impact
- •AI opportunities: genomics, pathology, immune-system analysis, better diagnostics
- •Strong stance: maximize benefits and mitigate harms—don’t halt progress needed for cures
- 34:58 – 39:37
Regulation, adoption barriers, and proof points for AI in healthcare/biotech
The conversation turns to what’s missing for AI to deliver on biotech and clinical promises: workflow integration, accuracy, and real-world adoption. Simo notes that while AI’s promise in drug discovery is significant, the industry still needs clear evidence—better clinical trial success rates and demonstrable patient outcomes.
- •Elad critiques premature regulation analogies (e.g., nuclear) that can stall progress
- •Key bottleneck in healthcare: embedding AI into doctors’ real workflows
- •AI can keep clinicians current by summarizing latest research amid admin overload
- •Need high accuracy and vetted sources; echoes existing issues (doctors already use Google)
- •Biotech needs measurable outcomes: more AI-driven drugs, higher trial success, better real-world efficacy
