Privacy, Security, and Trust Drive AI Devtools Innovation

At the heart of it all this week: privacy, security, and trust in AI and devtools solutions. With innovations like zero-knowledge networks and deepfake shields showing developers how to build verifiable systems that keep data locked down. Other insights touch on AI agents streamlining workflows, fresh capital accelerating infrastructure growth, YC fueling founder momentum, and targeted fixes for legacy dev hurdles.

These developments point to a more resilient ecosystem…read on for the details.

🏆 Funding Wins

Heizen Bids Farewell to Legacy IT with $500K for AI-Native Teams

Agents aren’t reliable. Humans aren’t efficient enough. At Heizen, Agents help Humans move fast, and Humans help Agents execute reliably!

Heizen, founded in April 2024, has secured $500k in pre-seed funding from Titan Capital, Varun Alagh (Mamaearth), Abhishek Goyal (Tracxn), and Chandigarh Angels Network. Founders Aman Arora, Abhilasha S., and Nijansh V. want to move India’s IT services beyond bloated, legacy models. Their platform uses small, AI-native teams where agents and engineers work together to deliver software faster and more reliably. Heizen already supports 50 customers and runs over 120 active sprints each month. The new funding will help grow their engineering team in India, expand in the U.S., and build out their multi-agent delivery platform.

💡 Meet the team and learn more at Heizen.

Runloop Secures $7M to Bridge the AI Coding Agent Gap

Runloop AI has landed $7M in seed funding led by The General Partnership with Blank Ventures joining the round. Founded by Jonathan Wall and Cristian Strat the San Francisco startup provides enterprise infrastructure for AI coding agents. Their platform offers secure sandboxes (Runloop devboxes), GitHub integration, and public benchmarks that help companies deploy agents up to six months faster. The 12-person team will use the funding to accelerate hiring and product development as demand grows for tools that bridge the gap between AI coding prototypes and production.

➰ Explore Runloop and get $25 in credits!

Retab Lands $3.5M to Make Document Processing Painless for Devs

French AI startup Retab has secured $3.5 million in pre-seed funding from VentureFriends, Kima Ventures, K5 Global, and tech leaders including Eric Schmidt (via StemAI), Olivier Pomel (Datadog CEO), and Florian Douetteau (Dataiku CEO). The developer-first platform automates structured data extraction from documents through a simple SDK. Developers define their schema in natural language while Retab handles dataset labeling, prompt engineering, model selection, and evaluations. The platform features self-optimizing schemas, model-agnostic routing, guided reasoning, and integrations with Zapier, n8n, and Dify.

Founder, Louis de Benoist speaks to the incentive behind Retab:

“People keep building demos that look like magic, but break the moment you put them into production…We lived that pain ourselves. Wiring up fragile pipelines just to extract a few fields from a PDF. We built Retab because it’s the developer-first platform we always wished we had.”

🔍 If you haven’t already, give retab a go.

🦄 And from the Unicorns….

Anaconda Secures $150M+ as Valuation Hits $1.5 Billion

Anaconda, Inc. has secured over $150 million in Series C funding led by Insight Partners with participation from Mubadala Capital. The Python distribution platform now boasts $150M in annual recurring revenue, 21 billion downloads, and 50 million users worldwide. Founded in 2012, Anaconda serves 95% of Fortune 500 companies and 10,000+ enterprises building AI systems. The company has strengthened its leadership team with Laura Sellers (CPTO), Jane Kim (CCO), and Barry Russell (SVP Partnerships) while launching its comprehensive AI Platform to help organizations manage open source components securely.

🐍 Explore Anaconda’s developer tools here.

🚀 3, 2, 1… Launches

Modelence Launches All-in-One TypeScript Platform for Startup Speed

Modelence (YC S25) is a platform that eliminates infrastructure headaches for startup developers. The team has created a unified solution that handles authentication, databases, real-time events, AI observability, cron jobs, email services, and monitoring in one package. Founders Aram Shatakhtsyan (formerly at Intuit) and Eduard Piliposyan, PhD (from CodeSignal) built Modelence after spending years creating similar infrastructure repeatedly at scale. Their open-source framework lets TypeScript developers skip straight to building product features, combining functionality that would normally require Next.js, Vercel, and Supabase separately.

Ⓜ️ Create a free cloud environment here

Reality Defender Brings Two-Line Deepfake Detection to Developers

Reality Defender spots AI-generated content instantly using just two lines of code. The platform functions as a Stripe-equivalent for deepfake protection, working seamlessly across video, audio, and text formats. Founded by Ben Colman and Ali Shahriyari, Reality Defender offers enterprise-grade security already trusted by major banks, governments, and media organizations. Their new developer tools include SDKs for five programming languages and protect against synthetic CEO video calls, cloned voices, and manipulated media. Early adopters receive 50 free monthly scans, with additional bonuses for the first five integrators.

🛡️ Find out more about Reality Defender and check out their YC Launch Offers here.

Okibi Builds AI Agents from Conversational Prompts

Mahyad Ghassemibouyaghchi and Saurav Mitra return to YC with Okibi, a web app that builds AI agents using simple natural language prompts instead of complex code. The platform automatically handles tool calls, browser interactions, human-in-loop workflows, and initial evaluations from conversational descriptions. Already working with 15 YC companies, Okibi helps automate tasks like lead qualification, invoice generation, and meeting preparation. The tool positions itself as “Lovable but for Agents,” letting users start with templates for common workflows.

Mahyad’s journey to this second YC startup includes a remarkable story of getting kicked out of YC S21 with SigmaOS, due to export laws, securing Paraguay residency in weeks, and getting reinstated, proving the resilience that now drives Okibi’s mission – we love this story!

🤖 Checkout Okibi and build your first agent in minutes here

Cyberdesk Automates Legacy Windows Apps Through Plain English

Mahmoud Al-Madi and Alan Duong have built Cyberdesk (YC S25), a computer use agent that automates legacy Windows applications through simple API calls. The YC S25 startup targets developers in healthcare, supply chain, and regulated industries who struggle with poor APIs in customer EHRs, ERPs, and other legacy systems. Cyberdesk combines memorized step sequences for reliability with computer use models for handling unexpected popups, creating a hybrid approach that works consistently.

The platform already serves customers automating dental EHRs, accounting software, and other legacy applications. Mahmoud previously scaled RPA solutions to 20,000 employees at a Fortune 100 company, while Alan brings NASA experience and computer use agent expertise from his previous startup.

🖥️ Book a demo and automate your legacy apps

Decipher AI Spots User Frustrations Before Support Tickets Arrive

Decipher AI (YC W24) watches every customer click through vision-powered LLM agents that catch bugs and frustrations in real time. Built by Michael Rosenfield and Rohan D., the platform identifies silent drop-offs, UI issues, and errors the moment they happen, then automatically generates detailed reports with reproduction steps and drafts pull requests.

While dev velocity increases across the industry, most teams still spend days investigating user problems after the damage is done. Decipher flips this reactive approach by monitoring the entire user experience continuously and surfacing issues before they impact metrics. Setup takes under five minutes with integrations for Linear, Slack, and popular development tools.

🧩 Find out more and sign up for free

Kernel Delivers 300ms Browser Spin-Up for AI Agents

Catherine Jue and Rafael Garcia have created Kernel, a browser-as-a-service platform that launches cloud browsers in just 300 milliseconds for web automations and AI agents. The infrastructure handles the complex parts of running browsers at scale including isolated virtual machines, live views, video replays, residential proxies, and auto-CAPTCHA solving. Kernel works with any browser framework like Playwright, Puppeteer, and Browser Use, while offering straightforward pay-per-use pricing. The platform also includes an agent runtime that lets developers host their web agents directly on Kernel’s infrastructure and invoke them on-demand with simple prompts.

🌐 Launch your first cloud browser here

Kombai Builds Domain-Specific AI Agent for Frontend Development

Dipanjan Dey and the Kombai team have evolved their Figma-to-code tool (Product Hunt’s most upvoted devtool in 2023) into a specialized AI agent focused entirely on frontend development tasks. Unlike general-purpose coding agents that struggle with real-world frontend work, Kombai understands codebases like a human developer and consistently outperforms frontier models on frontend benchmarks. The agent handles complex tasks like building entire app UIs, refactoring existing code, and creating components from Figma designs, text, images, or code snippets. Key features include advanced Figma interpretation, built-in knowledge of 30+ frontend libraries, visual previews in sandboxes, and smart model routing that automatically selects the best AI model for each task.

🐕 Try Kombai’s frontend AI agent and be sure to give them your feedback!

 

🔦 DevTools of the Week: TACEO

TACEO caught our eye this week as a standout in the privacy infrastructure space, especially with their recent production launch of TACEO:Proof AND a fresh $5.5 million seed round.

Founded in 2022 out of Graz University of Technology’s cryptography research group, the company emerged from a team passionate about bridging the gap between privacy and verifiable trust in decentralized systems.

Lukas Helminger (CEO), a former cryptography researcher and lecturer with roots in the Austrian Defense Department, leads alongside Lukas Götz (CPO), who has shaped DeFi ventures since 2014, and Aisling Connolly (CSO), a PhD holder in threshold cryptography from ENS Paris with experience in traditional finance and web3. Their technical backbone also includes Daniel Kales and Stefan Plank, who helped pioneer tools like the Poseidon hash function, now a ZK standard.

The why behind TACEO stems from a core problem: public blockchains expose too much data, while private solutions often sacrifice decentralization. They started to build coSNARKs (a blend of multiparty computation (MPC) and zero-knowledge proofs) for collaborative, private computations without leaking sensitive info. This vision powers their private shared state network, enabling apps to process encrypted data verifiably.

Last week, TACEO:Proof went live after a beta that generated over 1 million proofs, offering developers a decentralized MPC network for delegating ZK proof generation. It outperforms local proving on mobiles and laptops (e.g., 2.2 seconds for a 2¹⁸ constraint circuit versus 15.4 seconds on mobile), with features like batch proving (30 RSA proofs in under 30 seconds), a developer dashboard for job monitoring, and voucher-based delegation without user auth.

This, along with their seed funding raise, positions TACEO to scale rapidly, unlocking privacy-first apps in DeFi, AI training on sensitive datasets, decentralized identity, and healthcare. For technical founders and developers in regulated spaces, it means faster, reliable ZK without heavy libraries or trust compromises, like seamless Ethereum integration via coNoir tooling that already aids Aztec builders and secures World ID’s 14 million users through MPC for biometric checks.

Look into their docs and start integrating at docs.taceo.io, or read the full launch post from the team. If running a node or collaborating interests you, reach out directly.

🗞️ Other News

Today marks the final day to apply on-time for Y Combinator’s Fall 2025 batch.

The program runs from October to December in San Francisco, focusing on in-person kickoffs, weekly meetups, and guidance from experienced partners. Founders get $500,000 in seed funding, access to a vast alumni network, and introductions to investors at Demo Day. Applications submitted by 8pm PT today receive decisions by September 5; late submissions are considered but without a guaranteed timeline.

Technical founders building devtools or AI startups should apply if ready to accelerate. Submit your application here.

Got a launch, raise, or tool we should feature next week?

Hit reply and send it our way.

Louise Ogilvy (Recruitment Director) – UK/USA

Becca Combe MIRP CertRP (Lead Recruiter) – USA

Ben Chipchase (Lead Recruiter) – UK

Natalie Harper (Administrator)

Looking for your next DevTools role?

Visit devtoolsjobs.com