devrev

Quarter-end has a way of piling everything on at once. The decks, the pipeline reviews, the reporting you swore you'd stay on top of. So we're going to make a bit of a bet on ourselves: let Computer carry some of that load this quarter, and if it doesn't earn its keep, ask for your money back. Subscribe to Pro by July 31, give it a real 30 days, and there's a $30 gift card in it for you. Still not sold two months in? We'll refund you, no fuss. Feels like a fair dare for quarter-end: dvrv.ai/3RZhO2f

It's Monday morning. A customer writes in with a billing problem – a problem that your talented support team has already solved 14 times this quarter. But your AI agent has no idea. It pulls the record, gives a generic answer, and routes the rest to a human – who re-reads the same history the agent couldn't see. Every AI agent on the market can respond to a query. That used to be the hard part. It isn't anymore. The real difference shows up somewhere else: what the agent truly understands – and what it remembers. Computer’s Customer Agent runs on, and inside, Native Shared Memory: the living digital twin of your whole company. It doesn't just know the customer's record – it knows everything your team knows, everything your entire business knows. How this issue got solved last time. What was promised on last week's call. The bug engineering is shipping a fix for right now. The proof? BILL ran it on 200,000 real queries. These results speak for themselves. Learn more: bit.ly/44M4LUR

Advisor Spotlight 🔦 When someone who helped build India's GST and income tax platforms tells you the biggest AI-first enterprise on earth isn't a company, you listen. Meet Raghu Cavale, Chairman at Fracktal Works - Industrial 3D Printers, DevRev advisor, and one of the engineers behind the digital rails now running a country of 1.4 billion. His claim? The world's largest AI-first enterprise isn't a company at all. It's India. And he built a big chunk of it back when large language models weren't even a thing. His point is refreshingly blunt. The magic wasn't predicting AI. It was getting the data right first. Unified rails. Digital from day one. Fraud analytics baked in, not bolted on. The results still surprise people. UPI runs 49% of all real-time digital transactions on the planet. A tax return that used to take 63 days now takes one. There's one line that stuck with us. You cannot bolt intelligence onto chaos. If your data lives in silos, your AI will too. It's the same bet we made with Computer. We built it on one unified knowledge graph instead of a pile of tools that can't talk to each other. So what's Raghu's advice to leaders? Don't buy more AI. Build the foundation first. Know how: bit.ly/4wkjbHI

Who's allowed to run this? Permissions and blast-radius control for agent actions x.com/i/broadcasts/1…

Most voice AI can remember individual conversations with a customer. But they don’t remember – or understand – your entire business. That’s why all those voice AIs know nothing about the fix that went out last night. That four accounts were hit by update this before lunch. That the caller's problem is already solved – it just hasn't shipped yet. The voice was never the problem. What it can reach, understand, and remember: that’s what really matters Voice AI in Computer runs on your org’s Native Shared Memory – the connected, always-current understanding of your whole organization. So the same agent that’s already resolving issues in chat and email – is now answering out loud. It doesn't just hand off. It resolves. On a live call, Computer’s voice AI reads the error logs, pulls the order state, and acts across your systems – so a single conversation goes from spotting the bug to issuing the refund. No handoff. And it gets better, every single week. Not because you're tuning it, but because every call, every resolution, feeds back into that Native Shared Memory. Read more: dvrv.ai/4wpzF1y

We think most "AI for work" is quietly useless – it forgets you the moment you close the tab. Computer doesn't. And we're so sure of that, we're doing something a little unusual: We're paying people to try our product. On purpose. Subscribe to Computer Pro, get a $30 gift card, and if it doesn't earn its place in 60 days – full refund, no questions. We're this confident because Computer actually remembers what your whole team knows, so it only gets better the more you use it. Your move 👉 dvrv.ai/4px5Auh

We welcome Malcolm Koh, Head of ASEAN & North Asia, @devrev, as an esteemed speaker at The Experience Show Asia, Malaysia 2026 to share his valuable insights. Register as a group customer-experience.live/cx-live-show-a… #CXLiveTweets #CXLive #TheExperienceShow #CXLiveShow #DevRev #WorkSofter

You start a conversation on your laptop. Two hours later, you're on your phone and that conversation is gone. So you screenshot it. Or you re-explain. Or you just forget about it entirely. However with Computer by DevRev, you don't have to. Sessions now sync across every surface – desktop, web, and mobile. Start a conversation on the desktop app, open the same chat on your browser later, and it's all there. Same messages, same context, same progress. But it's not just the conversation that follows you. The files do too. Every chat also has a Canvas – a single workspace that stays the same no matter where you open it. Computer created a report on desktop? It's already visible on web. You uploaded a spreadsheet on mobile? It's there when Computer picks the task back up on desktop. No exporting. No re-uploading. No "wait let me send that to myself." Kick off deep work on the desktop app. Step away. Pick it up on your phone during a commute. Continue on the web when you're back. Nothing lost, nothing fragmented. Your work should always follow you. Now it does: bit.ly/4ff6KH9

You're being told to do more with AI. But most AI still gets things wrong often enough that you end up double-checking everything – which defeats the point. So we tested it properly. We ran Computer and the raw Opus 4.8 model on the same tasks, scored by an independent judge. Computer solved 94.3% of them correctly. The base model managed 63.6%. Same underlying model. The 30-point difference comes from everything Computer wraps around it: 1. Shared Memory across your org, so context is never lost and you never repeat yourself. 2. Real connection to your accounts, your team, and your systems – not a chatbot working in isolation. 3. Built-in skills like Account Deep Dive, Deal Health, and Call Prep, ready to run just by asking. That accuracy is the whole game. The difference between an answer you trust and one you have to verify before you use it When the work actually matters, an AI that's right 6 times out of 10 is slowing you down. Computer is built to get it right – so you can get back to the work only you can do. See how it works, free, at Computer.io: dvrv.ai/4gN6OiG

One agent. Same settings for everyone. Your preferences, your teammate's preferences, the whole org's – all the same. That's where most AI tools stop. It's where Computer starts. With Computer, every person gets the personalization they prefer - how it works, what it knows, what it can do. Same foundation, tailored to you. 1. Set instructions – your role, your defaults, how you like things done 2. Add skills – in plain language, by uploading a file, or just say "save this as a skill" 3. Connect your systems – Google Calendar, Notion, Slack, whatever you work in 4. Control memory – let Computer recall past chats or build from your work. Both on by default, and off just pauses - nothing's deleted. That's your side. Your admin has theirs. They set the org-wide foundation – mandatory skills, shared tools, company-level instructions. Users can't remove what the admin pushes. But everything you add on top is fully yours, visible only to you, and doesn't touch anyone else's setup. The best of both: personal enough to feel like yours, aligned enough to stay consistent across the org. Know more: bit.ly/4f7xKbI

Initial reactions: Building and working with an agentic PM x.com/i/broadcasts/1…

DevRev has earned the AWS AI Software Competency – and with it, has become a ‘Differentiated Partner’ on the AWS Software Partner Path. What does this mean in plain terms? @awscloud reviewed our architecture, our production deployments, our security practices, and whether real customers are seeing real outcomes. It's not a self-certification. It's a rigorous, independent technical validation – and they don't make it easy. For us, this felt like the natural next step. We built computer by DevRev with AI as the core, not the coating. It reasons across your data, connects teams, and acts with shared memory and real context. That depth is exactly what AWS validates when they look under the hood. If you're evaluating AI partners in the support and product development space – this is one signal worth paying attention to. Not because of the badge, but because of what it took to earn it. Great things get built when the foundation is solid. If you're a startup figuring it out on AWS, we'd love to be part of yours: dvrv.ai/4w6F8Ko

Here's what nobody says out loud at AI conferences. Most of what's on stage has never survived contact with production. At Leadership Circle Bengaluru last week, we skipped that script 💯 The day opened with a hands-on workshop and hackathon, teams building with AI before a single slide went up. Then the conversations began. Manoj Agarwal, our Co-founder & President, said the quiet part first. The model was never the hard part. It's everything around it. Scattered data. ROI you can't defend. Shadow AI. Governance always two steps behind. Neeraj Matiyani, our Country Manager for India, brought it down to earth with the real customers and the five problems every AI programme eventually hits. And Murali M., our Head of Solutions Engineering India, showed what it looks like when it actually works. The sharpest moments came from the people living it. Prabu Ram from @Razorpay , Neelu Shaikh from @Adobe , and Satvik Dudeja from Arvind Fashions Limited, on what it actually takes to run AI in production. The line that stuck? It's not about the LLM. It never was. It's about context, cost, and execution. And because the best rooms are smart and fun, @one_by_two closed the night out and reminded everyone how to laugh again after a full evening of AI. Best sign it worked? The room kept talking long after the agenda ended. To the folks who came, listened, and shared openly, thank you. You're the reason it worked.

Your pipeline isn't dying because of bad leads – it's dying in discovery. Most reps get one shot to prove they understand the buyer's problem. One call to earn the next. And yet, most teams still prep the same way they did five years ago. On July 21, we're bringing together GTM people who think differently about that moment. Inside the Call is an evening of drinks and honest conversation on qualifying smarter – with AI in the room, not replacing the room. We've got Laura Fu from our team, plus Amit Prakash and Rahul Balakavi from AmpUp.AI – two founders who've spent years building the tools that make discovery a system, not a guessing game. If you're in the Bay and this sounds like your kind of evening, come through. 📅 Tuesday, July 21 | 5:30-7:30 PM PDT 📍 DevRev HQ, Palo Alto ✅ Register now: luma.com/23h70tkf

You know that one workflow someone on your team just nails every time? The sales rep whose CRM updates are always clean. The support agent who triages faster than anyone. The CSM who is always on top of their QBRs. That expertise used to stay locked in their head. Now there's a place for it. Skills Marketplace is where you browse pre-built skills, find what fits, and install it to Computer in one click. CRM sync, competitive positioning, outstanding action items from meetings - it's all there, tested and ready. You can add a skill just for yourself or extend it to your entire org. And if you've built something that works, publish it to the marketplace so others can use it too. The idea is simple: one person figures something out, and the entire org now benefits from it. That's how Computer gets better - by extending what your people already know how to do. Our CTO Ahmed Bashir and Product Manager Shashank Sharan talk more about this here: bit.ly/4f0NOvS

We just launched “Enterprise-Bench” – the AI industry’s first open, vendor-neutral benchmark built specifically for enterprise use cases. And we’re making a challenge to other AIs: come test yourself against Computer by DevRev. AI companies throw around lots of benchmarks – but they mostly test single-user tasks that don’t run inside the organizational complexity of a real enterprise. We built Enterprise-Bench to fill that gap. We ran Computer against Claude Code. Same Opus 4.8 model. Same dataset. Same independent judge. The only variable was architecture. – ACCURACY. Computer: 94.3% vs. Claude: 63.6%. – TOKEN EFFICIENCY. Computer: 4.4x fewer tokens per correct response. – COST AT SCALE. As the dataset grew, Computer’s token usage: stayed roughly flat vs. Claude Code: climbed by 29%. Independently validated by Professor Alexandros Dimakis (University of California, Berkeley). Built on Terminal Bench. Full dataset on Harbor Hub. Full GitHub access. The leaderboard is open. Any vendor, researcher, or enterprise can take the same test and submit their results: dvrv.ai/4pvkwZX So, who’s going to step up?

The world's first benchmark for enterprise AI agents x.com/i/broadcasts/1…

Here's something nobody talks about in AI buying cycles: Your vendor's demo looked incredible. But run that same question again next week - did you get a different answer? Most platforms will. And nobody notices until it's in front of a customer. Our Head of EMEA, Patrick van de Werken, has been sitting in these evaluations and boiled it down to three questions that separate platforms built for scale from ones built for demos: ❓ Can you trace the answer back to the exact record - or is it "based on patterns" (fancy way of saying vibes)? ❓ What happens to your cost per query when your data doubles next year? Flat or vertical? ❓ If your AI agent breaks something at 2 AM Saturday, can you hit undo before your coffee gets cold? If your platform can't answer all three, it might be time to ask what you're actually paying for. Patrick breaks the whole thing down here: dvrv.ai/4p92EDC

Monday is the loneliest day in most orgs. Everyone's online. Nobody's aligned. You're all in the same Slack workspace, technically – but mentally, you're in five different versions of last week. So you do what everyone does. Re-read threads. Search for that doc someone "definitely shared." Ping a teammate who pings another teammate who says "let me check." Forty minutes later, you have half the answer and a meeting invite. By 11 AM, you've been very productive at figuring out what happened before today. Congrats! 🥲 This is the problem we kept coming back to when building Computer – your data lives in dozens of tools, none of them talk to each other, and the person manually keeping it all in sync is losing their weekends to it. So we built a system that connects your tools, keeps everything current and consistent, and organizes it all so AI can actually search, reason, and act on it. Permissions intact. Context preserved. Always live. The result? Steve in Ops gets his weekends back. Angela in Product gets her charts in seconds. Janet in Support clears more tickets. And you stop rebuilding context every Monday morning. Here's how the whole thing works, layer by layer: devrev.ai/how-computer-w…

Not many events promise you AI strategy and belly laughs in the same evening. But we're doing both. Here's how the night goes. Leadership Circle Bengaluru, July 9. You walk into a room full of enterprise leaders who are done pretending their AI deployments are going great. Instead, they're talking about what's actually working, what's quietly failing, and what nobody wants to say out loud in a boardroom. And then, once your brain is properly stuffed with knowledge graphs and memory architectures? We let @one_by_two ( Atul Khatri ) take the stage. Best way to end an evening that made you think? One that makes you laugh just as hard. Smart evening. Funny ending. When's the last time your calendar offered you that. Register now → bit.ly/4wCqbjd 📍JW Marriott Bengaluru · 5 PM onwards