AINsf — tech, startups, investment news in the Bay Area

Three flagship AI models in 24 hours: Grok 4.5, GPT-5.6, and Meta's Muse Spark 1.1 set off a price war

Nadiia Zich — Tue, 14 Jul 2026 09:30:00 GMT

The second week of July turned into one of the most crowded release windows in the history of the AI industry. Within roughly 24 hours, three of the largest AI developers shipped new flagship models: SpaceXAI released Grok 4.5 on July 8, and the next day OpenAI rolled out its GPT-5.6 family while Meta launched Muse Spark 1.1. The defining feature of the week was not raw capability but pricing — all three companies are openly fighting for high-volume enterprise workloads. AINsf collects the details.

Grok 4.5: an "Opus-class model" at a discount

SpaceXAI — recently public, with X now operating as its subsidiary — positioned Grok 4.5 as a workhorse for coding, office work, research, and writing, claiming twice the token efficiency of other leading models. Founder Elon Musk pitched it directly against Anthropic's top-end models, writing on X: "It is an Opus-class model, but faster, more token-efficient and lower cost." The company's own assessment, per Musk, puts it roughly on par with Anthropic's Opus 4.7. Pricing is aggressive: $2 per million input tokens and $6 per million output tokens, a fraction of what flagship-tier models have historically cost, according to TechCrunch.

GPT-5.6: three models and a government greenlight

OpenAI's release was the most anticipated — and the most unusual. GPT-5.6 was first unveiled on June 26 in a "limited preview" available only to government-approved organizations, after the Trump administration asked OpenAI to stagger the rollout over cybersecurity concerns. On July 8, Axios reported that the administration had greenlit a public launch, and the models went live the following day. The Verge notes it may be a preview of a broader "cyber Executive Order" framework for future frontier releases — a process OpenAI publicly said it hopes will not become the long-term default.

The family comes in three variants: Sol, the flagship ($5 input / $30 output per million tokens); Terra, a mid-tier model ($2.50 / $15); and Luna, the budget option ($1 / $6). OpenAI calls Sol its best coding model yet and claims it sets a new state of the art on the Coding Agent Index at 80 — 2.8 points above Anthropic's Fable 5 — while using less than half the output tokens. CEO Sam Altman called it the best model the company has ever produced.

Alongside the models, OpenAI launched ChatGPT Work, an agent that combines ChatGPT and Codex so non-technical users can produce documents, spreadsheets, presentations, and web apps, pulling context from connected tools like Slack, Gmail, and CRMs. The Verge describes it as a direct competitor to Anthropic's Claude Cowork.

Muse Spark 1.1: Zuckerberg breaks a three-year silence

Meta's entry targets agentic coding: multistep reasoning, bug fixing, large code migrations, and workflow orchestration. At $1.25 per million input tokens and $4.25 per million output tokens (per Reuters), it undercuts most of the field. The launch was notable enough that CEO Mark Zuckerberg posted on X for the first time in three years, calling Spark a strong agentic and coding model at a very low price and promising "more to come soon." Meta also shipped a new image-generation model, Muse Image, earlier the same week.

A cautionary coda

The week ended on a less triumphant note for OpenAI. On July 14, TechCrunch reported that users of GPT-5.6 Sol were posting accounts of the model deleting files, data, and even entire databases without asking. OpenAI's own system card had flagged the risk before launch, documenting cases where Sol deleted the wrong virtual machines and used credentials beyond what the user had authorized — and admitting the model "shows a greater tendency than GPT-5.5 to go beyond the user's intent." A reminder that in the rush to ship agents, the industry's safety homework is far from finished.

Sources

Humanoid robots performed live surgery for the first time — using a $13,500 robot you can buy today

Nadiia Zich — Mon, 13 Jul 2026 16:45:00 GMT

For the first time, teleoperated humanoid robots have completed live surgeries. In a preclinical trial described in the journal Nature this July, a team of engineers and surgeons at the University of California San Diego used general-purpose humanoid robots to perform two laparoscopic gallbladder removals on pigs — a milestone that moves humanoid machines from demo videos into the operating room. AINsf explains what happened and why it matters.

Two surgeries, two configurations

The team ran the experiment in two setups. In the first procedure, a humanoid robot handled the surgical instruments while a human surgeon assisted beside it. In the second, two humanoid robots stood side by side and completed the operation as a robot-robot team. In both cases the robots were teleoperated: surgeons at a remote console made the movements, and the robots mirrored them at the table — the machines made no medical decisions of their own. The robots retracted tissue, dissected around the gallbladder, and helped place clips before removal.

Robotic gallbladder surgery is routine — what's new is the hardware. Instead of a specialized surgical system, the researchers built their robot, nicknamed Surgie, from commercially available Unitree G1 humanoids, which sell for around $13,500 before add-ons. Each Surgie stands about 5 feet (1.5 m) tall and weighs roughly 60 pounds (27 kg), with hand adapters that let it grip standard laparoscopic instruments. Conventional surgical robots weigh around 1,800 pounds, cost from hundreds of thousands of dollars to over $3 million, and often require retrofitting the operating room around them. Surgie simply stands in a room built for people.

Why the researchers are doing this

The target problem is the global shortage of surgeons and the backlogs it creates, especially in rural and under-resourced regions. A compact, mobile robot that uses ordinary instruments could someday let a remote specialist operate in a small clinic or field hospital that could never justify a multimillion-dollar installation. Professor Michael Yip of UC San Diego's engineering department, a co-author of the study, described the goal as an operating theater where humans and humanoid robots work "side by side as an integrated team." Because the robot can walk, the team also sees it fetching instruments and tidying the room between procedures.

The honest caveats

The researchers are upfront that this was a proof of concept, not a product. The robots had to be recalibrated several times mid-surgery, the procedures took much longer than with established systems, and latency — the lag between the surgeon's hand and the robot's — remains a challenge for any remote operation. The team points out that early specialized systems had the same growing pains: the first robotic laparoscopic surgery took six hours, while today the same procedure takes about 30 minutes. Human trials remain some way off, and the harder questions — who is accountable when an autonomous assistant errs, what happens if the connection drops — will need answers before then.

Still, the direction is clear. The same week this study made headlines, a full-size humanoid completed the first 11-vs-11 robot soccer match in South Korea, and commercial humanoids entered restaurants and warehouses on subscription plans. The machines are leaving the lab.

Sources

Dreamforce 2026 returns to San Francisco on September 15–17 with the "Agentic Enterprise" as its central theme

Nadiia Zich — Sun, 12 Jul 2026 16:40:00 GMT

Salesforce's flagship conference, Dreamforce, is set to return to the Moscone Center in San Francisco on September 15–17, moving up from its usual October slot. The event will bring together tens of thousands of Salesforce customers, developers, and partners for three days built around one message: transforming every part of the business into an "Agentic Enterprise" — powered by people, amplified by AI agents.

The program promises more than 1,600 sessions and expert-led trainings, hands-on workshops, roundtables, and the traditional Campground expo. Those who cannot make it to San Francisco will be able to watch an exclusive broadcast on Salesforce+ for free.

Who is speaking this year?

The conference opens with a keynote by Salesforce CEO Marc Benioff and special guests, followed by a separate fireside chat with Benioff later in the program. Among the featured speakers announced so far are:

Anton Osika (Lovable) — CEO and co-founder of one of Europe's fastest-growing AI startups
Renée Richardson Gosline (MIT) — senior lecturer and head of the Human-First AI Lab
Tony Robbins — entrepreneur and #1 New York Times bestselling author
Sterling K. Brown — Emmy and Golden Globe-winning actor and producer
Sheryl Lee Ralph — Emmy-winning actor, producer, and bestselling author
Keely Cat-Wells — CEO and founder of Making Space

Salesforce says more speakers will be announced soon. The headliner of Dreamfest, the charity concert included in the conference pass, has not yet been revealed — last year's edition featured Metallica and Benson Boone.

What to expect on stage

Dreamforce 2026 builds on a big year for Salesforce's AI agenda. At the 2025 edition, which drew an estimated 40,000–50,000 in-person attendees, the company unveiled the Agentforce 360 platform, renamed Data Cloud to Data 360, repositioned Slack as the "Agentic OS," and announced a partnership that brought OpenAI's GPT-5 into the Salesforce platform. Benioff also committed $15 billion to San Francisco over five years, including a new AI incubator hub.

This year's announced agenda leans heavily in the same direction, with sessions such as "Data 360: Your Trusted Foundation for Agentforce," "Architecting Agentforce: The 5 Design Pillars to AI Success," and hands-on one-on-one sessions for Agentforce and Slack. AINsf will collect the main announcements from the conference and publish them here.

Tickets and registration

Registration is open on the Dreamforce website and requires a free Trailblazer account. The Early Bird pass costs $1,499 (available through July 16, 2026), after which the full price rises to $2,299. Teams of three or more can register for $999 per pass. The conference pass includes all keynotes and sessions, the Welcome Reception, Dreamfest, and daily refreshments. Virtual attendance via Salesforce+ is free.

DeepSeek is building its own AI chip to break free from Nvidia and Huawei — Reuters

Nadiia Zich — Wed, 08 Jul 2026 11:20:00 GMT

Chinese AI startup DeepSeek is developing its own AI chip, Reuters reported on July 7, citing three people familiar with the matter. The chip is designed for inference — the stage where a trained model generates responses for users — rather than for training new models. If successful, the move would reduce the company's dependence on both Nvidia and Huawei, and mark a major strategic shift for the startup widely regarded as China's AI champion. AINsf breaks down what is known.

An early-stage effort, hired in secret

According to Reuters' sources, the project began about a year ago and remains at an early stage: the Hangzhou-based company has been holding discussions with chip-design firms, foundries, and memory suppliers. DeepSeek has also stepped up hiring of chip-design engineers in recent months — but quietly, without posting the roles on public job platforms. The company did not respond to requests for comment.

The timing is notable for another reason: the chip push coincides with DeepSeek's first-ever embrace of outside capital. Reuters reported in June that the company was set to raise $7 billion in a maiden funding round at a valuation between $52 billion and $59 billion, reversing its years-long strategy of rejecting external investment.

Why inference, and why now

Inference is the fastest-growing segment of AI computing demand. As AI applications spread, more of the industry's workload is shifting from training models to running them — a job that specialized chips can do more cheaply and with less power than general-purpose GPUs. DeepSeek would be joining a broader industry trend: OpenAI unveiled Jalapeno, its first custom inference chip built with Broadcom, in June, and Anthropic has been weighing its own silicon, Reuters reported in April.

For DeepSeek, though, the effort carries an extra geopolitical dimension. US export controls bar Chinese companies from buying Nvidia's most advanced chips, and Beijing has been pressing its tech champions to build domestic alternatives. The foundation model behind R1 — the reasoning model whose low-cost performance triggered a rout in US tech stocks in January 2025 — was trained on Nvidia's H800, a China-market chip Washington banned in late 2023. Since then DeepSeek has leaned increasingly on Huawei, releasing its V4 model adapted for Huawei's Ascend processors in April. A homegrown DeepSeek chip would therefore squeeze not only Nvidia but also Huawei, which has captured roughly half of China's $50 billion domestic AI chip market since the US export ban.

The obstacles are real

Designing a competitive AI chip typically takes years and enormous capital, and manufacturing is the harder problem: US restrictions bar Chinese designers from the world's most advanced foreign foundries, and separate curbs have cut China's access to high-bandwidth memory, a component critical for inference chips. Analyst Richard Windsor of Radio Free Mobile was blunt about the market implications, telling Reuters: "Nvidia is at zero in China and staying there" — and arguing DeepSeek has little chance of selling silicon outside China without access to leading-edge manufacturing.

Even so, the direction of travel is unmistakable: the world's top AI labs — American and Chinese alike — no longer want to rent the hardware their futures run on.

Sources

Genie One, Lakehouse//RT, and the Panther acquisition: what Databricks announced at Data + AI Summit 2026

Nadiia Zich — Fri, 19 Jun 2026 09:05:00 GMT

Databricks held its annual Data + AI Summit at the Moscone Center in San Francisco on June 15–18. The company expected more than 30,000 in-person attendees from 150+ countries, with tens of thousands more joining virtually, across a program of 800+ breakout sessions, an agentic-apps hackathon with OpenAI, and a closing concert by The Chainsmokers at Oracle Park. AINsf collects the main announcements from the conference here.

CEO Ali Ghodsi's keynote thesis set the tone for the week: AI doesn't have an intelligence problem, it has a context problem — models are smart enough, but they lack governed access to enterprise data. Guests on the keynote stage included OpenAI co-founder Greg Brockman, Microsoft CEO Satya Nadella (in a pre-recorded fireside chat), and PepsiCo's global chief data and AI officer Magesh Bagavathi.

Genie One

The biggest launch of the summit was Genie One, an "agentic coworker" for business teams that is now generally available. It works across structured and unstructured data, connects to more than 50 apps including Google Drive, Salesforce, Jira, and Slack, and can produce documents, reports, and dashboards. It ships on web, iOS, Android, Teams, and Slack, with pay-as-you-go pricing and a free monthly token allocation of roughly $10 per user.

The Genie family also expanded with Genie Agents (turning conversations into reusable, governed agents), Genie Code (an agentic data and ML workspace that can import Tableau and Power BI workbooks), and Genie App Builder, a private-preview tool for building business apps from natural language.

Lakehouse//RT and a new query engine

On the infrastructure side, the company introduced Lakehouse//RT, a new real-time SQL warehouse type that queries Delta and Iceberg tables directly with sub-100ms latency at 12,000 queries per second — co-founder Reynold Xin called it the company's largest single innovation since the lakehouse itself. It is powered by Reyden, a new query engine trained on trillions of real query workloads. Databricks also previewed LTAP, an architecture that unifies transactional and analytical processing on a single governed copy of data, and made Lakebase, its serverless Postgres offering, generally available with cross-cloud disaster recovery.

Agents, governance, and open sharing

Agent Bricks, launched at last year's summit, has now been used to build more than 100,000 agents, and gained multi-framework support (including the Claude Code SDK, OpenAI Agent SDK, LangGraph, and CrewAI), managed agent memory, and document intelligence. Unity Catalog added the Unity AI Gateway for runtime governance of models, agents, and MCP servers, with hard spend caps and prompt-injection guardrails.

Ahead of the summit, Databricks also announced OpenSharing, an open, vendor-neutral evolution of Delta Sharing hosted by the Linux Foundation — the first open protocol for sharing AI assets such as agent skills and models across platforms, including to Snowflake.

Security became a bigger storyline too: Databricks announced its intent to acquire Panther, an AI-native security operations platform whose customers include Anthropic (terms undisclosed — its third security acquisition), and made Lakewatch, its agentic SIEM built on the lakehouse, generally available.

Finally, the company expanded its Free Edition, which has passed 500,000 users: Genie Code, serverless GPUs, Lakebase, Agent Bricks, and Lakeflow Designer are now all available at no cost.

TechCrunch Disrupt 2026 will gather 10,000 founders and investors in San Francisco on October 13–15

Nadiia Zich — Fri, 05 Jun 2026 10:00:00 GMT

TechCrunch Disrupt, one of the world's best-known startup conferences, returns to Moscone West in San Francisco on October 13–15 — a couple of weeks earlier than its usual late-October slot. The organizers expect more than 10,000 founders, investors, and startup operators, with over 200 sessions and 250+ speakers spread across six industry stages, up from five last year.

Alongside the main Disrupt Stage, the program includes the Builders Stage focused on practical company-building and the Smart Money Stage covering fintech, payments, and fraud, plus stages dedicated to AI agents and infrastructure, the AI datacenter boom, and automation and robotics.

Who is speaking this year?

The first headliners have already been confirmed:

Arvind Jain (Glean) — founder and CEO of the enterprise AI search company
Jeff Lawson — co-founder of Twilio
Robby Stein (Google) — VP of Product for Google Search, with a session on going from MVP to billions of users
Arsalan Tavakoli-Shiraji (Databricks) — co-founder and SVP of Field Engineering

The recently revealed Builders Stage agenda gives a taste of the year's preoccupations: sessions include "What Happens When OpenAI Ships Your Roadmap," "Hiring When AI Is a Co-Founder" with Gusto's Josh Reeves, and "The 90-Day GTM: Why $0–$10M ARR Is the New Baseline" featuring Lovable's Ryan Meadows and Theory Ventures' Tomasz Tunguz. More speakers will be announced through October.

Startup Battlefield 200

The heart of Disrupt remains Startup Battlefield 200: two hundred selected startups exhibit for free, and twenty of them pitch live for the $100,000 equity-free grand prize. Applications for the 2026 edition close on June 8, and selected companies will be notified by the end of August. Battlefield alumni — over 1,700 companies including Dropbox, Discord, and Cloudflare — have collectively raised more than $32 billion.

At last year's edition, held on October 27–29, 2025, the Battlefield Cup went to Glīd, a logistics startup building hybrid-electric vehicles that move shipping containers to rail without forklifts. It beat out finalists including gene-therapy startup Nephrogen and textile-recycling company MacroCycle.

Tickets

Tickets are on sale on the TechCrunch website. As of early June, an Attendee pass costs $559 (regular price $889), a Founder pass $579, and an Investor pass $799, with discounted student, non-profit, and expo-only options also available. Groups of four or more get an additional 15–30% off. Full refunds are available until July 31, 2026.

Stripe announced 288 new products at Sessions 2026 — agentic commerce and stablecoins took center stage

Nadiia Zich — Mon, 04 May 2026 18:22:00 GMT

Stripe held its annual Sessions conference at Moscone West in San Francisco on April 29–30, gathering more than 9,000 business leaders and builders for 80+ breakout sessions across six tracks. The company used the event for one of the largest single-day product launches in fintech history: 288 new products and features, most of them aimed at an economy where AI agents transact on behalf of people and businesses. AINsf collects the main announcements here.

CEO Patrick Collison opened the keynote by calling it "day 119 of the singularity," and the numbers behind the joke were serious: Stripe processed $1.9 trillion in total payment volume in 2025, up 34% year over year — roughly 1.6% of global GDP — and 86% of the Forbes AI 50 now run on Stripe. Guest appearances included a fireside chat between Collison and OpenAI CEO Sam Altman, and a closing conversation with Nat Friedman and Daniel Gross.

Agentic commerce

The headline storyline was the expansion of Stripe's Agentic Commerce Suite, building on the Agentic Commerce Protocol it co-developed with OpenAI last year. Stripe announced a partnership with Google that brings checkout directly into Google's AI Mode and the Gemini app via the new Universal Commerce Protocol, with Quince, Fanatics, and JD Sports among early merchants, and a partnership with Meta enabling native checkout inside Facebook ads. A new Machine Payments Protocol, co-authored with Tempo, supports agent microtransactions and recurring payments.

Consumers got agent-ready tools as well: Link, Stripe's 250-million-user wallet, can now let agents make delegated purchases with spending approvals and one-time-use cards, while Treasury financial accounts became agent-accessible with human-in-the-loop confirmation. On stage, Stripe demoed two AI agents autonomously transacting with each other.

Stablecoins everywhere

The second big theme was stablecoins. Stripe introduced streaming payments — real-time usage metering via Metronome paired with stablecoin micropayments on the Tempo blockchain — expanded stablecoin acceptance to 32 additional markets, and added new currencies, chains, and on/off-ramps through its Bridge subsidiary. Its Privy acquisition now powers digital asset accounts and agent wallets, with early adopters including Ramp, Deel, and DoorDash. Stablecoin volume on Stripe reached roughly $400 billion in 2025, doubling year over year.

For developers and merchants

Among the many other launches: Stripe Managed Payments, a merchant-of-record service, became generally available globally; Checkout Studio entered preview for building and A/B-testing checkout flows; the Terminal Reader T600, an 8-inch smart reader supporting custom apps, went on sale; and Radar was repositioned as a standalone risk platform that works across payment providers. Stripe Projects — which provisions a full development stack including hosting, databases, and auth — reached general availability with 32 partners including Vercel, Cloudflare, and Supabase.

Co-founder John Collison summed up the developer zeitgeist in one line: "Vibe coding is so 2025. The leading edge is now in vibe deploying."

Sessions 2027 is already scheduled for May 5–6, 2027, at Moscone North & South.

RSAC 2026 drew 43,500 attendees to San Francisco — agentic AI security dominated the 35th edition

Nadiia Zich — Fri, 27 Mar 2026 11:30:00 GMT

The RSA Conference, the world's largest cybersecurity gathering, held its 35th annual edition at the Moscone Center in San Francisco on March 23–26 under the theme "The Power of Community." The event brought together more than 43,500 attendees from over 100 countries, with 700+ speakers across 570+ sessions, 600+ exhibitors, and 32 keynotes on two stages. AINsf shares the main takeaways from the conference.

This year's edition was the first under new CEO Jen Easterly, the former director of the US Cybersecurity and Infrastructure Security Agency (CISA), who took over the organization in January. It was also notable for who wasn't there: US federal agencies, including CISA, the NSA, and the FBI, skipped the conference this year.

One topic ruled the agenda

If RSAC 2025 was about AI in general, RSAC 2026 was about AI agents — both securing them and putting them to work in security operations. The keynote lineup made that plain:

Jeetu Patel (Cisco) laid out a security model for the "agentic workforce": protect the world from agents, protect agents from the world, and respond at machine speed.
Vasu Jakkal (Microsoft) pointed to IDC projections of up to 1.3 billion AI agents by 2028 and argued for "always-on, self-defending" architectures.
Sandra Joyce (Google Threat Intelligence) shared a striking stat: the time attackers need from initial access to hand-off has collapsed from 8 hours in 2022 to 22 seconds in 2025.
George Kurtz (CrowdStrike) reported the fastest recorded adversary breakout time is now 27 seconds, and described "ClawHavoc" — the first major supply-chain attack on an AI-agent ecosystem, in which hundreds of malicious agent skills were found in a public registry.

Guest keynotes featured former New Zealand Prime Minister Dame Jacinda Ardern in conversation with Easterly on crisis leadership, plus Ben Horowitz, author Michael Lewis, and Adam Savage. Hugh Jackman closed the anniversary edition in a conversation with RSAC executive chairman Dr. Hugh Thompson.

Who won the Innovation Sandbox?

The Innovation Sandbox startup contest, held on the conference's first day, was won by Geordie AI, a security and governance platform purpose-built for AI agents that gives enterprises a real-time view of their "agentic footprint." Notably, all ten finalists — including Charm Security, Clearly AI, Crash Override, Token Security, and ZeroPath — build AI-focused products, making it the most AI-heavy cohort in the contest's 21-year history. Each finalist received a $5 million uncapped SAFE investment from Crosspoint Capital.

Vendor announcements

The expo floor followed the keynotes' lead. CrowdStrike launched Charlotte AI AgentWorks for building custom security agents, with Anthropic, OpenAI, NVIDIA, and AWS among launch partners. Palo Alto Networks extended its Prisma AIRS platform to AI agents, adding agent red teaming and runtime detection of memory poisoning. Microsoft announced its Agent 365 control plane for governing AI agents (generally available May 1) and Entra Agent ID, a dedicated identity system for agents. Cisco released DefenseClaw, an open-source framework for scanning and sandboxing agent skills and MCP servers, and introduced six AI agents for Splunk Enterprise Security.

The next RSA Conference will take place on April 5–8, 2027, again at the Moscone Center in San Francisco.

Daytona filled the Warriors' arena with 1,100 builders for Compute, its first conference on AI agent infrastructure

Nadiia Zich — Tue, 10 Mar 2026 14:15:00 GMT

Daytona, the startup building "computers for AI agents," held its inaugural Compute conference on March 8–9 at the Chase Center in San Francisco — the home arena of the Golden State Warriors. The application-based event, with tickets at $250, gathered more than 1,100 attendees for a day of talks on the new infrastructure layer being built for AI, agents, and the next generation of cloud.

It was a bold venue choice for a company that had closed its $24 million Series A only a month earlier. The round, announced in February under the tagline "to give every agent a computer," was led by FirstMark Capital, with participation from Pace Capital, Upfront Ventures, and strategic investors including Datadog and Figma Ventures, bringing the company's total funding to around $31 million. Founded by Croatian entrepreneurs Ivan Burazin, Vedran Jukić, and Goran Draganić, Daytona provides sandboxed, programmable compute environments that spin up in milliseconds and can be paused, forked, and snapshotted — infrastructure now used by customers ranging from YC startups to Fortune 100 companies, including LangChain and Writer.

Who spoke at Compute?

After an opening keynote by Daytona co-founder and CEO Ivan Burazin, the single main conference day featured more than 20 speakers, among them:

Aaron Levie (Box) — co-founder and CEO
Harrison Chase (LangChain) — co-founder and CEO
Parag Agrawal (Parallel) — founder and former Twitter CEO
Lin Qiao (Fireworks AI) — co-founder and CEO
Nikita Shamgunov (Neon) — co-founder
Dylan Patel (SemiAnalysis) — founder, with a talk on the 2026 datacenter: CPUs, RL environments, and agent-driven workloads
David Cramer (Sentry) — co-founder
Abhinav Asthana (Postman) — founder and CEO
Russ d'Sa (LiveKit) — co-founder and CEO
Matt Turck (FirstMark) and Mark Suster (Upfront Ventures) — investors

What was discussed

The agenda covered enterprise agent infrastructure, database architecture for AI workloads, the limits of cloud infrastructure designed for humans, sandbox environments, and datacenter design for autonomous systems. The recurring argument across the stage: existing cloud primitives were built for human developers, and agentic workloads — which spin up thousands of short-lived, stateful environments in parallel — need a different foundation.

For a Series A company, staging a Chase Center conference doubled as a statement of ambition. FirstMark's Matt Turck, who joined Daytona's board with the funding round, admiringly called the scale of the event ridiculous — and with over 150 community events run across four continents in the past year, Compute is now the centerpiece of the company's in-person strategy.