Notorious: Unlocking the Future of Intelligent Agents (Highlight)
Notorious: the best founders, startups, strategies, metrics and community. Every week.
Current subscribers: 10,844, +71 since last week
Share the love: please forward to colleagues and friends! 🙏
Notorious Startup of the Week: Highlight
In the rapidly evolving world of AI and large language models (LLMs), one major challenge stands out: how do we move beyond simple question-and-answer interfaces into a realm where agents can truly work for us, long-term and in the background, across all the tools and environments we rely on every day?
Enter Highlight. Built from the ground up to empower agents, Highlight leverages underlying browser technology—specifically Chromium via Electron—to bridge the gap between the intelligence of LLMs and the practical, complex tasks that span web apps, operating systems, and beyond. Think of it as the “iPhone for LLMs”—an easy-to-use platform that consolidates all the best interfaces and agents into a single, seamless ecosystem. For this edition of Notorious, Highlight CEO and Founder Pim de Witte shares his approach and strategy to building Highlight with us:
Designed for Agents from Day One
From its inception, Highlight has been designed to run agents. Unlike traditional browser-based interfaces that are limited by security models and URL-based permissions, Highlight provides a native runtime environment. This environment empowers agents to interact directly with the operating system in ways a standard browser can’t. Here’s how:
Seamless Agent Integration: You can publish web interfaces as Highlight agents in under a minute. Because Highlight shares the same foundation as Chrome, any web-based experience can become a native Highlight agent.
Integrated Authentication & Runtime API: Agents can hook into Highlight’s own authentication system and use a special Runtime API to access OS-level features securely. This is a permissions-based approach that maintains safety while granting more powerful capabilities.
App Store for Agents: Highlight has its own app store with full deep-linking support. It can transform popular websites—like Goodreads or LinkedIn—into Highlight-compatible apps in minutes. As a result, nearly any web experience can quickly become part of your agent’s toolkit.
These elements come together to give agents everything they need to accept tasks, convert them into prompts, find the right tools, and execute them as long-running background processes.
How Long-Running Tasks Work
With Highlight, the dream scenario becomes a reality: you delegate a complex, time-consuming task, and your agent runs with it—quietly, in the background, until it’s done. Here’s what that looks like:
Prompt Assembly: Highlight identifies or constructs the right prompts for the agent. For many consumer tasks, a robust prompt repository will supply the perfect instructions.
Enterprise-Ready Guardrails: In more controlled environments, deterministic guardrails can ensure tasks run safely within corporate policies and compliance requirements.
Task Detection: Highlight eventually will be able to recognize when you want to initiate a long-running task—something that would traditionally require you to babysit a browser tab or script (coming soon).
Over time, as model-based systems like RunnerH improve at performing computer-level tasks, these agents will become more adept at handling intricate, behind-the-scenes operations. Highlight consolidates this complexity, choosing the best intelligence and context for you automatically, just as an iPhone chooses the best networks and configurations without user intervention.
The User Experience: Invisible, Yet Powerful
The ideal user experience is effortless. With Highlight:
Background Execution: Assign a task and let the agent work silently. No invasive pop-ups or interruptions. Check in from anywhere—phone, computer, or tablet—and see how your tasks are progressing.
Sandboxed Browser Agents: Because these agents run within Highlight’s controlled environment, they’re more secure and intuitive. They understand web interfaces by default, which covers a huge range of use cases immediately. Over time, Highlight’s runtime will evolve to let agents control entire desktops and local applications with proper permissions.
Scalability: Spawn agents will eventually work on tasks simultaneously (coming soon). Each agent knows when to finish and when to request more input, enabling a level of productivity unheard of in today’s point-and-click workflows.
Winning on the Task Interface
The first step to mainstream adoption is to significantly improve how users interact with and benefit from LLMs. Highlight focuses on two key accelerations:
10x Faster Inputs: By automatically gathering context and writing prompts for you, Highlight reduces the friction of talking to an LLM. Screen-based prompting—where the system anticipates what you need next—lets you move from idea to execution almost instantly.
10x Faster Outputs: Highlight delivers results where you need them most. Whether writing directly into a Notion doc, updating tasks in Linear, or scheduling events in your Calendar, Highlight’s output integration ensures the agent’s work lands exactly where it’s useful.
As power users begin to automate their own input and output workflows—essentially building personal prompt repositories—Highlight’s underlying runtime can leverage these same prompts to control browser instances with natural language. This creates a cycle of continuous improvement, personalization, and capability expansion.
Memory, Prompts, and Enterprise Scalability
Once Highlight nails the core task interface, the platform turns its attention to deeper capabilities:
Memory: Agents should know you—your preferences, habits, and past tasks. Highlight’s memory layer sits above all agents, ensuring continuity and personalized results.
Prompt Repository: By accumulating a library of high-quality prompts, Highlight matches context-rich tasks with the right solutions, accelerating productivity even further.
Long-Running Tasks & Enterprise-Ready Workflows: Eventually, organizations want these capabilities in secure, audited environments. Highlight’s architecture supports that transition seamlessly.
Building a Moat Through Integration and Context
Highlight’s advantage compounds as it integrates into every aspect of the user’s digital life:
Capture Interface: Highlight becomes the single pane of glass for delegating tasks. Instead of juggling multiple apps, users rely on Highlight’s interface. One platform, one workflow.
Cross-Platform Reach: With native support across devices, it doesn’t matter if you’re on Windows at work or an iPhone on the go—Highlight ties everything together.
Growing Ecosystem: Agents and apps published in Highlight’s store will proliferate, making it harder for big tech players to replicate the ecosystem quickly. Coupled with Highlight’s unique memory and prompt optimization, this creates a powerful defensible advantage.
The Path to Magic
Winning in the consumer world means delivering an experience that feels like magic—where the agent already knows what you want and acts on it without exhaustive setup. Achieving this requires continuous refinement, focus, and user-centric design. Highlight’s long-running tasks and dynamic prompt ecosystem won’t appear as “infrastructure” to users; they’ll simply know it as the capability to do more, faster, with less hassle.
The trajectory is clear:
Perfect the task interface.
Accelerate input and output.
Leverage prompts, memory, and runtime integration.
Personalize at scale.
Let users carry these capabilities across every device and platform.
By following these steps, Highlight aims to usher in an era where LLM-powered agents are not just tools you consult occasionally—but trusted, ever-present assistants that quietly shape and streamline your digital life, no matter what device you’re on or what task you need done.
Thanks for reading. By way of background, I am an early-stage investor at Wing and a former founder. Please reach out to me on X @zacharydewitt or at zach@wing.vc. Some of the early-stage PLG + AI companies that I have the privilege to work with and learn from are: AirOps, Copy.ai, Deepgram, Hireguide, Slang.ai, Tango, Tome and Workmate.
Operating Benchmarks (from PLG Startups):
I will continue to update these metrics and add new metrics. Let me know what metrics you want me to add (zach@wing.vc)
Organic Traffic (as % of all website traffic):
Great: 70%
Good: 50%
Conversion rate (website → free user):
Great: 10%
Good: 5%
Activation rate (free user → activated user):
Great: 50%
Good: 30%
Paid conversion rate (free user → paid user):
Great: 10%
Good: 5%
Enterprise conversion rate (free user → enterprise plan):
Great: 4%
Good: 2%
3-month user retention (% of all users still using product after 3 months):
Great: 30%
Good: 15%
Conversion from waitlist to free user:
<1 month on waitlist: ~50%
>3 months on waitlist: 20%
For more detail on acqusition rates by channel (Organic, SEM, Social etc), please refer to this prior Notorious episode.
Financial Benchmarks (from PLG Public Companies):
Financial data as of previous business day market close.
Best-in-Class Benchmarking:
15 Highest EV/ NTM Revenue Multiples:
15 Biggest Stock Gainers (1 month):
Complete Dataset (click to zoom):
Note: TTM = Trailing Twelve Months; NTM = Next Twelve Months. Rule of 40 = TTM Revenue Growth % + FCF Margin %. GM-Adjusted CAC Payback = Change in Quarterly Revenue / (Gross Margin % * Prior Quarter Sales & Marketing Expense) * 12. Recent IPOs will have temporary “N/A”s as Wall Street Research has to wait to initiate converge.
Recent PLG + AI Financings:
Seed:
Across AI, a startup developing Agentic AI for enterprise workflows, has raised $5.75M. The round was led by Village Global and Cota Capital.
AIMon, a full-cycle LLM App accuracy platform, has raised $2.3M. The round was led by Tidal Ventures and Bessemer Venture Partners.
Argmax, a startup dedicated to empowering developers and enterprises in deploying commercial-scale inference workloads on user devices, has raised $8M. The round was led by Salesforce Ventures, with participation from General Catalyst.
Briefcase, an AI-powered bookkeeping assistant designed to automate manual bookkeeping tasks, has raised $3M. The round was led by Earlybird Venture Capital, with participation from Entrepreneur First and Tiny Supercomputer Investment Company.
Cake, a managed open-source AI infrastructure platform, has raised $10M. The round was led by Gradient Ventures, with participation from Firestreak Ventures, Primary Venture Partners, Alumni Ventures, Friends & Family Capital and Correlation Ventures.
Cartesia, a multimodal intelligence tool designed to enable real-time, on-device AI capabilities on every device, has raised $22M. The round was led by Index Ventures, with participation from Lightspeed India Partners, FactoryHQ, SV Angel, General Catalyst and A* Capital.
Elvex, an AI-based tool intended to reduce complexity and mitigate risk in enterprise workflows, has raised $6.4M. The round was led by m]x[v Capital and FundersClub, with participation from Remarkable Ventures, Industry Ventures and Accelerator Ventures.
eSelf AI, a platform for businesses to create face-to-face conversational AI agents, has raised $4.5M. The round was led by Explorer Investments, with participation from Ridge Ventures.
NeuBird, a company using generative AI to automate IT site reliability operations, has raised $44.5M at a $100M valuation. The round was led by M12, with participation from Prosperity7 Ventures, Mayfield Fund, AIspace Ventures and StepStone Group.
Pathway, a startup building live AI systems that think and learn in real-time as humans do, has raised $10M. The round was led by TQ Ventures, with participation from Inovo VC, Market One Capital, Id4 Ventures and Kadmos Capital.
Pin, an AI-powered recruitment platform designed to streamline the hiring process from sourcing to interview scheduling, has raised $3M. The round was led by Expa.
Pixeltable, an open-source AI data infrastructure company, has raised $5.5M at a $25M valuation. The round was led by The General Partnership, with participation from Liquid 2 Ventures, South Park Commons, Exceptional Capital and Serena Ventures.
Sightglass, an AI-powered investor relations platform for the private markets, has raised $3M. The round was led by Base10 Partners.
WaveForms AI, a startup creating AI assistants that pick up on emotional cues from verbal interactions, has raised $40M at a $200M valuation. The round was funded by Andreessen Horowitz.
Zero, a startup developing an AI-powered “zero-click” CRM, has raised $2.7M. The round was led by 20VC, with participation from 20Growth, 20Product, Discovery Ventures, Illusian and Greens.
Early Stage:
xAI, Elon Musk’s artificial intelligence venture, has raised $6B. The round was funded by Valor Equity Partners, Sequoia Capital, Andreessen Horowitz, AJI Capital and Qatar Investment Authority.
Series A:
Aampe, a provider of mobile app engagement and personalization infrastructure for short-form notifications and messaging, has raised $18M. The round was led by Theory Ventures, with participation from Z47.
Brellium, a clinician-validated AI audit system designed to help automate medical chart review, has raised $11.14M. The round was funded by First Round Capital.
BRM, an AI-powered vendor management and procurement platform, has raised $15M. The round was led by Caffeinated Capital, with participation from Definition and Original Capital.
Chargezoom, a platform for AI-powered billing and integrated payments, has raised $11.5M. The round was led by Kickstart Fund, with participation from Adly, Frazier Group, Early Light Ventures, Motley Fool Ventures, SaaS Venture Capital, Stout Street Capital and Okapi Venture Capital.
Enterpret, an AI-integrated customer feedback intelligence platform for product development and CX teams, has raised $20.8M. The round was led by Canaan Partners, with participation from Wing VC, Kleiner Perkins, Peak XV Partners and Recall Capital.
Happyrobot, a startup that specializes in AI-enabled voice solutions for the logistics sector, has raised $15.6M. The round was led by Andreessen Horowitz, with participation from Ryder Ventures and Y Combinator.
Liquid AI, a foundation model company building general-purpose AI systems at every scale, has raised $250M at over a $2B valuation. The round was led by AMD, with participation from Itochu Techno-Solutions, Nuclea, OSS Capital, Duke Capital Partners and PagsGroup.
RapidCanvas, a startup developing AI agents that automate complex tasks traditionally performed by data scientists and engineers, has raised $16M. The round was led by Peak XV Partners, with participation from Titanium Ventures, Accel and Valley Capital Partners.
Stainless, a startup creating an SDK generator powered by AI, has raised $25M at a $152M valuation. The round was led by Sequoia Capital and Andreessen Horowitz, with participation from Zapier, MongoDB Ventures, The General Partnership and Felicis.
Stigg, a monetization platform for engineers, has raised $17.5M. The round was led by Red Dot Capital Partners, with participation from Unusual Ventures, Emerge Ventures, Cerca Partners and Redseed.
Twelve Labs, a developer of hyperscale AI models that understand various types of information in video, has raised $30M. The round was led by NVentures and New Enterprise Associates, with participation from Intel Capital, Databricks, HubSpot Ventures, SK Telecom, Snowflake, In-Q-Tel, Operator Collective, InnoWhale Ventures, Index Ventures, Radical Ventures, WndrCo, Samsung NEXT Ventures and Korea Investment Partners.
Vapi, a developer platform for deploying Voice AI agents, has raised $20M. The round was led by Bessemer Venture Partners, with participation from Abstract Ventures, AI Grant, Y Combinator and Saga Ventures.
Vooma, an AI platform designed for freight brokers and carriers, has raised $16.3M in Seed and Series A funding. The $3.6M Seed Round was led by Index Ventures and the $13M Series A was led by Craft Ventures.
Series B:
Connectly, a conversational AI platform that automates how businesses communicate with their customers, has raised $20M at a $90M valuation. The round was led by Alibaba Group, with participation from RX Ventures, Globant Ventures, Falabella Ventures, Unusual Ventures and Volpe Capital.
Luma AI, a video generator platform intended to offer multimodal AI to expand human imagination and capabilities, has raised $90M at a $267.71M valuation. The round was led by Andreessen Horowitz, with participation from Amazon Web Services, Advanced Material Development, South Park Commons, Prototype Capital, Factorial Funds, Bossa Invest, AJI Capital and Silicon Valley Future Capital.
Sublime Security, an AI-powered email security platform, has raised $60M. The round was led by IVP, with participation from Citigroup Alternative Investments, Decibel Partners, Index Ventures and Slow Ventures.
Yurts, a company providing generative AI integration at scale for defense, government, and enterprise customers, has raised $40M. The round was led by XYZ Ventures, Glynn Capital and Nava Ventures.
Series C:
Evinced, a software company focused on accessible web and mobile development, has raised $55M. The round was led by Insight Partners, with participation from M12, Capital One Ventures, Engineering Capital, Vertex Ventures US and BGV Group Management.