a16z speedrun SR006 companies

Acceler8 builds an agentic AI platform that gives executives continuous, real-time answers to workforce planning questions: which roles to automate, where to redeploy people, and how to structure an organization as AI agents take on more work. The decisions it replaces were previously handled through spreadsheets, outside consultants, and static HR systems, a process that took months and produced plans already obsolete on delivery.

Advocate builds an AI voice assistant for seniors and their families, handling everyday requests, such as rides, groceries, home repairs, and appointments, end to end over a single phone number. Family members receive text updates when tasks are completed, and Advocate proactively calls seniors to catch problems early. A shared record keeps the full picture visible to everyone involved, replacing the informal coordination burden that typically falls on one family member.

Alike builds a shared memory layer for AI coding agents, letting tools like Claude Code, Cursor, and Codex share context across sessions. Without it, each agent starts from scratch: redoing rejected work, colliding on the same files, and losing decisions the moment a terminal closes. Alike replaces the developer acting as a manual context router between agents.

Amdahl builds a context layer for B2B go-to-market teams, sitting between raw sales, support, and customer success data and the AI agents that act on it. It structures messy GTM data into cited, deal-grounded context so agents writing outreach, researching prospects, or diagnosing pipeline can produce outputs checked against what real buyers said and did. It replaces ad-hoc data wrangling, manual win/loss review, and ungrounded AI drafts that teams currently can't trust before they send.

Antihero Studios builds mobile games designed around social play, starting with MISFITZ, a free-to-play extraction game for iOS and Android. In MISFITZ, 16 players drop into a shared map, form ad-hoc alliances to loot, and choose whether to extract together or betray each other at the gate. The studio targets the mass-market mobile audience that currently has no native extraction game, replacing the solo or algorithmically matched experience with one built on emergent social tension.

August builds AI agents that execute commercial banking work, covering underwriting, servicing, and risk, replacing the manual processes that bank staff run today. The platform accumulates a proprietary context layer as usage grows, making each subsequent task faster and more informed. It targets commercial banks, with deployment already in production at a major institution.

Auto turns photos into Frames, small single-purpose AI apps that act on what the camera sees: counting calories, rating an outfit, identifying a plant, splitting a bill, or logging water. Consumers point their phone at the world and Auto selects the right Frame and runs it, replacing the manual effort of opening separate apps or searching for information. Users can also build their own Frames when an existing one does not cover their need.

Belong builds artist-owned membership programs that turn fan loyalty into a financial relationship. Artists issue memberships tied to real financial infrastructure, so fans who have spent years buying records, attending shows, and streaming releases are recognized and rewarded for that history. The product replaces informal, unmonetized fan devotion with a structured, recurring revenue stream driven by spending fans already do.

Bilrost builds AI-native infrastructure for commercial lenders, automating the full credit workflow from document ingestion through underwriting and portfolio monitoring. Its agents pull deal files across email, portals, and cloud storage, reconcile rent rolls, leases, and financials, and populate lender-specific underwriting models, replacing weeks of manual spreading, re-keying, and document chasing. Every output traces back to source documents, giving risk and compliance teams a defensible audit trail.

Bota gives AI agent developers a programmatic way to capture and process real-world, offline conversations, covering in-person meetings and phone calls that cloud-based tools never reach. Developers integrate Bota's API to add offline recording to their products; Bota supplies white-labeled hardware, handles manufacturing, logistics, and support, and delivers transcripts and metadata ready for any AI workflow. It replaces the manual work of wiring together physical recording hardware, operations, and data pipelines for teams building AI agents in sales, healthcare, law, finance, and productivity.

Cascade builds an AI-powered opportunity and proposal platform for architecture, engineering, and construction firms. It continuously scans government and private-market sources to surface matched RFPs, scores each for fit, and then drafts the proposal, replacing weeks of manual portal-watching and proposal writing. The target customer is AEC firms that currently rely on word of mouth and fragmented portals to find and bid on projects.

Cedar builds an AI-native professional network that scales employee referrals for fast-growing companies. Its agents conduct short calls with employees to map their trusted relationships, surface high-signal candidates, and build a proprietary graph that grows with every hire. The product replaces the manual, people-heavy referral programs that large companies have historically run at significant cost and headcount.

Clair Health builds the first continuous, noninvasive hormone monitor: a wrist wearable that tracks estrogen, progesterone, LH, and PdG in real time without blood draws. It uses a custom multi-modal biosensor array and an in-house female biology model to capture over 130 biomarkers, replacing periodic lab tests and manual cycle tracking. Women use it to manage fertility planning, athletic performance, skin health, mood, and perimenopause transitions.

Coalition Systems builds a secure operating system for national security coordination, connecting military units, intelligence services, government agencies, and critical infrastructure operators on a single purpose-built platform. It replaces the ad hoc use of Signal, WhatsApp, Telegram, and Google Sheets that defense and security organizations have adopted by default, providing communication, collaboration, and decision-support tools designed for field conditions and classified environments.

Concorda builds an operating system for litigation teams, handling case research, discovery drafting, and document generation end to end. Litigators use it to run interrogatories, requests for admission, and legal research across every U.S. court, replacing the manual drafting and research work that consumes associate and partner time. It is built by founders who practiced law and wrote the code themselves.

Crebit solves a core problem in stablecoin-based foreign exchange: blockchain rate quotes expire in ten minutes, which prevents businesses from committing to payments in advance. Crebit built a rate-locking layer that holds stablecoin exchange rates for up to 180 days, giving enterprises the certainty they need to plan and execute cross-border payments. The product replaces the manual workarounds and exposure to rate drift that companies currently absorb when moving money across borders via stablecoins.

Emanate builds AI agents that run revenue operations for industrial materials companies, handling the work that sales and operations teams do manually: fielding inbound inquiries, generating quotes, tracking account expansion signals, prospecting new accounts, and informing pricing decisions. The product operates continuously across every shift, replacing the patchwork of human-driven workflows that govern how industrial suppliers capture and grow revenue.

Grove Tax builds AI agents that automate the full tax return workflow for tax preparation firms, covering client intake, document collection, data extraction, return preparation, and final delivery. The product targets a profession where preparers spend the majority of their time on administrative work rather than actual tax work, replacing manual document chasing, data entry, and client back-and-forth with autonomous agents. The result is a single platform that lets a preparer move from onboarding a client to delivering a signed return without leaving Grove.

Hammock builds an AI agent that helps consumers capture HSA and FSA savings on healthcare and wellness spending. The agent scans past purchases to surface eligible expenses users have missed, then generates Letters of Medical Necessity through an integrated physician network to make gym memberships, massage, and recovery products HSA/FSA reimbursable. It replaces manual paperwork and guesswork that cause eligible spending to go unclaimed.

Heavi builds AI agents that source and order parts for heavy-duty vehicle mechanics and fleet teams, replacing the manual work of calling suppliers, checking fitment across OE and aftermarket catalogs, and chasing backorders. A technician drops in a VIN and a part description, and Heavi searches catalogs, contacts vendors, places the order, and coordinates delivery. The product targets a repair and service market where retiring technicians take hard-won diagnostic and parts knowledge with them, leaving shops slower and vehicles sidelined.

Hotbox converts Instagram engagement into direct sales for sellers of high-demand drops: jerseys, trading cards, collectibles, tickets, and limited-edition merchandise. It scans comments, stories, and DMs to identify the highest-intent buyers, enriches them against a seller's ideal customer profile, and completes the transaction inside a DM checkout, replacing manual follow-up and lost post-drop revenue.

Idilio builds a mobile streaming platform for Spanish and Portuguese-speaking audiences, delivering original serialized microdramas in a vertical, 1-3 minute episode format. The content targets Latin American viewers raised on telenovela storytelling who now consume video on their phones, replacing longer-form TV drama with fast, bingeable episodes written by telenovela veterans. Beyond the consumer app, Idilio is developing AI-driven production tools and proprietary data infrastructure to create and distribute serialized stories at scale.

Kaaro builds AI agents for railroad operations, replacing the phone calls, emails, and manual paperwork that still drive day-to-day rail workflows. The product handles tasks like incident reporting, billing, data entry, and shipment tracking, work that today runs on tribal knowledge and fragmented systems. Freight railroads are the customer, and the agents act as a shared operational layer that surfaces real-time status and drives action without human chasing.

Loops AI builds an intelligence layer for e-commerce brands that structures and enriches product catalog data so AI agents can act on it across search, virtual try-on, and omnichannel sales. The platform converts shopper intent signals into purchases and cuts return rates by giving every downstream agent a consistent, AI-readable view of the catalog. It replaces manual data preparation and disconnected point tools that brands currently stitch together to run AI-driven commerce.

Miraka delivers clinician-led cardiac care to patients at home, combining continuous wearable and biometric monitoring with real-time clinical intervention to catch deterioration before it triggers an ER visit or readmission. The product targets post-surgical and high-risk cardiac patients who currently go unmonitored between hospital appointments. It replaces reactive, episodic follow-up care with a persistent care team that adjusts medications and therapies as patient data shifts.

Mirror Mirror AI operates a marketplace where fashion models license their likeness for AI-generated imagery, and brands book those models to produce lookbooks, e-commerce shots, and campaign visuals in hours rather than weeks. Each image generated under agreed terms earns the model a fee, replacing traditional photoshoots for brands while giving models a compensated, consent-based channel for AI use of their likeness.

Modaic builds verification and alignment infrastructure for AI decision-making. For teams running classification, moderation, or automated evaluation workflows, it scores model confidence on each decision, routes uncertain cases to human reviewers, and uses that feedback to refine the underlying instructions over time. This replaces the binary choice between costly full human review and brittle automated pipelines that fail on edge cases.

Modern Industrials builds custom AI systems for building materials distributors, automating the back-office work that sales and operations teams handle manually today. Its first product targets inside sales workflows: quoting, order entry, and analytics that currently run on slow, error-prone manual processes. The longer-term goal is a full operating system for physical-goods distribution.

Oasis builds a hybrid workspace where human teams and AI agents collaborate in shared rooms, documents, and codebases. It targets non-technical teams who want to run multi-agent workflows without writing the underlying infrastructure themselves. The platform maintains a persistent memory layer across every task and decision, so agents carry full context when swapped or upgraded, replacing the manual handoff work that otherwise falls on people each time a model or workflow changes.

Oasiz builds a social distribution platform for AI-native software, starting with games. It treats playable software as a content format, giving creators a feed-based arcade where users discover and play AI-built games the way they scroll short-form video. The platform replaces traditional app store distribution for a new generation of creators who generate games with AI tools rather than conventional development pipelines.

Omi Health builds a longevity platform for dogs, giving pet owners proactive health data instead of waiting for illness to appear. A blood draw and saliva test measure 80-plus biomarkers and detect more than 1,000 conditions; an AI layer then produces a personalized care protocol for each dog. The product replaces the standard reactive veterinary checkup with continuous, preventive monitoring.

Panorama builds AI systems for enterprises, ingesting workplace data from Slack, email, and documents to map how teams actually work and surface personalized recommendations for AI adoption. The product delivers those recommendations through a feed-style interface, driving adoption from individual contributors upward rather than through top-down mandates. It replaces manual discovery of workflow inefficiencies and the guesswork of AI rollout planning.

PartyHat builds AI-powered tools for creating and running interactive party and social experiences. The product targets hosts and event organizers, replacing manual game facilitation and activity planning with automated, on-demand entertainment.

PayPath builds an AI operating system for debt management, replacing spreadsheets, manual workflows, and legacy software across the full credit lifecycle. Enterprise lenders and servicers use its autonomous agents to handle origination, servicing, collections, and recovery. The platform automates the account-level work that previously required large manual operations teams.

PicPet builds a social messaging app that turns photo sharing between friends into a virtual pet care loop: every photo a user sends feeds their shared pet, making the friendship itself the gameplay. It targets consumer social and creator audiences, replacing passive group chats and static photo-sharing with an ongoing, pet-driven ritual that gives friends a shared object to tend together.

Piper-ai builds an AI operating system for preconstruction that general contractor estimating teams use to manage drawings, specs, addenda, subcontractor quotes, and schedules in a single connected project context. It generates scope documents, bid leveling sheets, estimates, and final bid reviews in the contractor's own formats, with every finding linked to its source document. The product replaces the manual work of assembling and reconciling preconstruction information across scattered files, freeing senior staff from reviewing every package by hand.

Pluvo builds an agentic operating system for corporate finance teams, replacing manual spreadsheet work and fragmented reporting with AI agents that run analysis, reconciliations, forecasting, and close tasks end to end. Finance teams connect their existing systems into a single governed context, then delegate recurring workflows to agents whose outputs trace every number back to its source data and calculations. The platform is model-agnostic, so teams are not tied to a single AI provider.

Prior Foundry builds a decision intelligence platform for government policy and program teams. It combines data connections, scenario modeling, and stakeholder input to track how programs are implemented, where performance falls short, and whether outcomes are being achieved. The platform replaces fragmented, periodic analyses with continuous, real-time evidence that feeds directly into policy decisions.

Quanto builds an AI-driven sales and operations layer for accounting firms, handling client prospecting, qualification, and calendar booking, then automating onboarding into QuickBooks and cleaning up new clients' books to the firm's standards. It replaces the patchwork of manual outreach, referral dependence, and ad-hoc onboarding that keeps most firms from growing without adding staff.

Quinn builds AI-powered training and operational readiness software for field service companies, covering trades like HVAC, pest control, roofing, and facilities maintenance. It converts a company's SOPs and top performers' knowledge into structured training that the entire workforce completes, replacing human-led training programs. New hires onboard faster, veterans stay consistent, and crews get instant answers by text while on the job.

Quo Labs builds Sam, a physical voice companion robot for older adults living at home. Sam conducts daily check-in conversations with seniors, then relays updates to their families, replacing the ad-hoc calls and in-person visits families rely on to monitor a loved one's wellbeing. The product targets the gap between full-time caregiving and no oversight at all.

SafeWorld builds a safety testing and evaluation platform for robots operating around people. It generates rare, high-risk scenarios in simulation to surface failure modes that physical testing misses, covering edge cases introduced by human behavior, environmental change, and software updates. Robotics teams use it to assess and document risk before and during deployment, replacing slow, expensive, and unscalable real-world test runs.

Sentra builds a managed organizational memory layer for enterprises, ingesting meetings, messages, emails, documents, code, and deal activity from over 200 tools into a single, permissioned context store. It replaces the manual effort of hunting across siloed systems by giving employees and AI agents a real-time, connected record of decisions, commitments, and context as work unfolds. The result is less coordination overhead and fewer hallucinations when agents reason across company knowledge.

Simula builds native ad infrastructure for AI apps, replacing standard banner and interstitial formats with interactive experiences, such as AI brand ambassadors and games played alongside AI companions. AI app developers integrate Simula to monetize their user bases without degrading the conversational or generative experience. The product targets the gap between how conventional ad formats disrupt AI products and what brands need to reach those audiences.

Sirius builds an AI retention platform for subscription businesses, targeting the early-months churn window that costs consumer companies the most. Its models process subscriber interactions at scale to intervene before cancellation, replacing manual lifecycle campaigns and rule-based retention playbooks. The primary customers are large consumer subscription brands.

Smart Bricks builds autonomous AI agents for real estate investors, covering the full investment cycle from deal discovery and underwriting through transaction execution and portfolio monitoring. The agents replace manual market scanning, spreadsheet-driven risk modeling, and fragmented diligence workflows that institutional and professional investors run today. Proprietary foundational models trained on real estate data feed multi-step reasoning across more than a thousand variables per opportunity.

snag builds an AI-native marketplace for short-term sublets, serving Gen Z renters who currently piece together housing through Facebook groups, Instagram, and Craigslist. Its AI model ingests raw social posts and converts them into structured listings, then matches renters to available sublets within seconds. snag sits between vacation rental platforms and traditional long-term rental sites, making flexible, month-to-month urban housing searchable and transactable in one place.

Snapp AI builds a sports AI assistant that consolidates live scores, news, highlights, analytics, and fantasy lineup decisions into a single app. Sports fans and fantasy players use it to get real-time answers about matchups, injury reports, and waiver wire moves without switching between tabs or apps. It replaces the manual work of aggregating sources, Googling pre-game questions, and guessing at lineup decisions.

Sparta builds agent testing environments that run AI agents against sealed, simulated replicas of the real systems they operate on, such as Stripe, Zendesk, and Salesforce. Instead of reading conversation transcripts, Sparta grades what actually changed in the system state after each run, giving teams exact, replayable proof of agent behavior. This replaces manual review and transcript-based evaluation with automated, outcome-level verification.

Straia builds an AI-native data platform for college and university administrators, connecting fragmented systems, including student information, learning management, and CRM platforms, into a single environment where staff can query institutional data in plain language. It targets campus leaders across enrollment, academic affairs, and the president's office, replacing manual reporting workflows and disconnected legacy software with AI agents that surface real-time insights on enrollment funnels, retention risk, course outcomes, and financial performance.

SUN generates on-demand podcasts, audiobooks, audio courses, and other spoken audio from a single text prompt, letting listeners customize voice, length, and language. It targets anyone who learns or entertains through audio and replaces the fixed, one-size-fits-all content of traditional podcasts and audiobooks. Users can ask questions mid-playback, clone their own voice, and schedule fresh episodes that arrive automatically.

Syncere builds Lume, a floor lamp that deploys a pair of robotic arms to handle household chores, starting with laundry folding, then retracts into a standard lamp when the task is done. The product targets homeowners who want chore automation without dedicated robot hardware sitting visibly in their space. It replaces the manual labor of routine domestic tasks and the clutter of purpose-built home robots.

Taxnova automates R&D tax credit and CapEx documentation for technology companies, replacing the manual process of engineer interviews and year-end reconstruction. It connects read-only to existing engineering tools, such as Jira, GitHub, Slack, and Confluence, and continuously builds source-linked, audit-ready evidence throughout the year. A tax advisor of the client's choice then reviews and submits the finished claim.

ThirdBrain Labs builds post-training infrastructure that lets domain teams convert their own expert knowledge into fine-tuned, proprietary AI models they own outright. It captures training signals directly from expert decisions in production, then continuously updates models without requiring large labeled datasets or manual data-preparation work. The product targets teams that currently depend on general-purpose, closed-model APIs and want task-specific models they control.

Variant builds an AI-powered conversion rate optimization engine for consumer brands. Three coordinated agents handle the full CRO workflow: one analyzes shopper behavior and ad context to identify friction points, one designs and codes on-brand A/B experiments for team approval, and one accumulates learnings across tests so each successive recommendation is informed by prior results. The product replaces the manual cycle of dashboard analysis, backlog-dependent development, and disconnected test-by-test iteration.

Vereda connects Brazilian farmers directly to agricultural input manufacturers, cutting out the distribution intermediaries that compress producer margins. A WhatsApp-native purchasing agent called Vera collects quotes from multiple suppliers simultaneously, structures the buying process from inquiry to payment, and is supported by on-the-ground field consultants who carry existing relationships in their regions. The product replaces a multi-layered reseller chain that adds cost at every step without adding supply.

Verra builds Meridian, an agentic operating system for management consulting firms. It organizes a firm's entire body of work, past engagements, decks, and models, into a single searchable knowledge base, then uses agents to draft proposals and deliverables in the firm's own templates. This replaces the manual work of hunting for precedents, rebuilding from scratch, and losing institutional knowledge when staff turn over.

ZeroDrift builds a compliance enforcement runtime that intercepts AI-generated communications before they are sent, validating, rewriting, or blocking each message against pre-loaded regulations and company-specific policies. It serves regulated enterprises in financial services, insurance, and healthcare, replacing manual compliance review and ad-hoc output filtering with a real-time, audit-ready enforcement layer. The runtime operates across AI agents, copilots, and any LLM via a single API call.