Funding – Dakidarts® Hub

Voyage AI Tackles AI Hallucinations with Advanced RAG Tools

Dakidarts — Fri, 25 Oct 2024 12:55:53 +0000

AI tends to make things up. That’s unappealing to just about anyone who uses it on a regular basis, but especially to businesses, for which fallacious results could hurt the bottom line. Half of workers responding to a recent survey from Salesforce say they worry answers from their company’s generative AI-powered systems are inaccurate.

While no technique can solve these “hallucinations,” some can help. For example, retrieval-augmented generation, or RAG, pairs an AI model with a knowledge base to provide the model supplemental info before it answers, serving as a sort of fact-checking mechanism.

Entire businesses have been built on RAG, thanks to the sky-high demand for more reliable AI. Voyage AI is one of these. Founded by Stanford professor Tengyu Ma in 2023, Voyage powers RAG systems for companies including Harvey, Vanta, Replit, and SK Telecom.

“Voyage is on a mission to enhance search and retrieval accuracy and efficiency in enterprise AI,” Ma said in an interview. “Voyage solutions [are] tailored to specific domains, such as coding, finance, legal, and multilingual applications, and tailored to a company’s data.”

To spin up RAG systems, Voyage trains AI models to convert text, documents, PDFs, and other forms of data into numerical representations called vector embeddings. Embeddings capture the meaning and relationships between different data points in a compact format, making them useful for search-related applications, like RAG.

Image Credits:Voyage AI

Voyage uses a particular type of embedding called contextual embedding, which captures not only the semantic meaning of data but the context in which the data appears. For example, given the word “bank” in the sentences “I sat on the bank of the river” and “I deposited money in the bank,” Voyage’s embedding models would generate different vectors for each instance of “bank” — reflecting the different meanings implied by the context.

Voyage hosts and licenses its models for on-premises, private cloud, or public cloud use, and fine-tunes its models for clients that opt to pay for this service. The company isn’t unique in that regard — OpenAI, too, has a tailorable embedding service — but Ma claims that Voyage’s models deliver better performance at lower costs.

“In RAG, given a question or query, we first retrieve relevant info from an unstructured knowledge base — like a librarian searching books from a library,” he explained. “Conventional RAG methods often struggle with context loss during information encoding, leading to failures in retrieving relevant information. Voyage’s embedding models have best-in-class retrieval accuracy, which translates to the end-to-end response quality of RAG systems.”

Lending weight to those bold claims is an endorsement from OpenAI chief rival Anthropic; an Anthropic support doc describes Voyage’s models as “state of the art.”

“Voyage’s approach uses vector embeddings trained on the company’s data to provide context-aware retrievals,” Ma said, “which significantly improves retrieval accuracy.”

Ma says that Palo Alto-based Voyage has just over 250 customers. He declined to answer questions about revenue.

In September, Voyage, which has around a dozen employees, closed a $20 million Series A round led by CRV with participation from Wing VC, Conviction, Snowflake, and Databricks. Ma says that the cash infusion, which brings Voyage’s total raised to $28 million, will support the launch of new embedding models and will let the company double its size.

Prepared Raises $27M to Revolutionize Emergency 911 Calls

Dakidarts — Wed, 02 Oct 2024 08:14:12 +0000

A company that claims its tech can “revolutionize” emergency calls has raised $27 million in a Series B round led by Andreessen Horowitz.

The company, Prepared, enables 911 dispatchers to get a caller’s real-time GPS location if their phone supports it. Via Prepared, dispatchers can also receive and respond to texts and images, and — on iPhones with Apple’s Emergency SOS Live Video feature — answer a video call.

Prepared co-founder and CEO Michael Chime claims that the platform can give operators valuable context they wouldn’t otherwise have.

“The goal of our technology is to reduce the burden of each individual call so that emergency response can move faster,” Chime said. “If we can save even a few seconds on a given 911 call, we want to do that.”

Nationwide, a number of 911 centers are landline-bound, struggle to locate callers, and can’t process SMS or photos. That’s despite a two-decades-old effort, Next Generation 911 (NG911), to modernize the more than 5,500 emergency call centers in the U.S.

NG911 is internet-based, and capable of receiving multimedia and more accurate caller info. However, deployments have only reached about 56.2% of the U.S., according to consulting firm Frost & Sullivan.

Launched by Chime, Dylan Gleicher, and Neal Soni in 2019, Prepared initially focused on a single type of emergency response: school shootings. The trio, who grew up near the sites of devastating school shootings, including Sandy Hook Elementary, dropped out of Yale together to build a public safety app for school administrators.

A year in, Chime, Gleicher, and Soni realized there was a larger customer segment — 911 call centers — that could benefit from Prepared’s tech. So they pivoted the company.

Today, Prepared offers a web-based platform that shows dispatchers a running transcript of calls. It uses AI to pull out potential items of importance, like addresses and descriptions of emergencies, even translating texts for dispatchers where necessary.

Prepared recently launched a tool that lets dispatchers chat with a Spanish speaker using an AI-generated voice. Prepared transcribes and translates the dispatcher’s speech, and then reads the translation aloud over the phone; Chime claims that this can reduce the need to conference with a third-party translator, which is the typical procedure with non-English callers.

“With a growing non-English speaking population, especially in larger cities, this has been a high-priority request from agencies,” he added, “which otherwise depend on language translators that can sometimes take several minutes to join a call after a request.”

Prepared’s dashboard for 911 dispatchers.

Image Credits: Prepared

Minutes shaved off an emergency response could make a difference. According to U.S. regulators, thousands of lives could be saved each year by reducing 911 response times by just a minute.

But AI translation and Prepared’s other AI-powered features also come with risks. AI often gets summaries wrong. And it’s been found to transcribe speech from some speakers more accurately than others. One recent study showed that speech recognition systems from leading tech companies were twice as likely to incorrectly transcribe audio from Black speakers compared to white speakers.

Chime notes that Prepared’s AI features are optional — the company’s video, GPS location, and texting capabilities are free for 911 centers. But he also argues that, on the whole, AI can help process dispatcher calls faster and more accurately.

“We have pioneered the use of AI in public safety to synthesize data and make it actionable,” he said. “Prepared’s summarizer allows dispatchers to read short AI-generated summaries of incidents rather than listening to minutes of call audio or reading lengthy notes. And we believe that our translation feature will prove crucial in enhancing accessibility for Spanish speakers while simultaneously improving response times for Spanish-speaking calls.”

Prepared, which has deals with close to 1,000 public safety agencies across 49 states, plans to put the cash from the Series B toward product R&D and go-to-market efforts. Prepared will also step up hiring, with the goal of adding 20 staffers to its 50-person, New York-based workforce by the end of the year.

“We’ve only just tapped the surface when it comes to the potential of unlocking critical citizen data,” Chime said. “We’re moving toward a world where Prepared as a platform connects and optimizes the end-to-end workflow from the second a call comes in to when a field responder is on scene.”

First Round Capital, M13, and undisclosed angel investors also participated in Prepared’s Series B. It brings the company’s total raised to $57 million.

OpenAI’s $6.5B Funding Round Nears Completion

Dakidarts — Sun, 29 Sep 2024 05:21:22 +0000

ChatGPT maker OpenAI’s next funding round, said to be around $6.5 billion, could close as soon as the end of next week.

That’s according to a new report in The Wall Street Journal, which also reveals that OpenAI’s rumored conversion from a nonprofit org into a for-profit company is now a virtual certainty. Investors in the new round will be able to pull back their money if OpenAI doesn’t complete the transition within two years, per the WSJ.

The WSJ piece paints a picture of a startup that, while having recently hit $4 billion in revenue, is still losing billions — and is beset by internal clashes, a burnout-prone work culture, and technical delays. Researchers at the company reportedly had just nine days to safety-test GPT-4o, OpenAI’s recent flagship model, as OpenAI raced to debut the model to take attention away from Google’s annual I/O developer conference back in May.

AI Coding Assistant Supermaven Raises Funds from OpenAI and Perplexity Co-Founders

Dakidarts — Fri, 27 Sep 2024 16:50:39 +0000

Jacob Jackson was all-in on AI early in his career.

Jackson co-founded Tabnine, the AI coding assistant that went on to raise close to $60 million in venture backing, while still a computer science student at the University of Waterloo. After selling Tabnine to Codata in 2019 (during his final exams), Jackson joined OpenAI as an intern, where he worked until 2022.

It was at that juncture that Jackson had the urge to start a company again, one focused on supporting common developer workflows.

“In the years since I built Tabnine, tools like ChatGPT and GitHub Copilot have changed the way developers work,” Jackson said. “It’s a really exciting time to be working on developer tools because the underlying technology has improved so much since I started Tabnine — which has led to many more developers becoming interested in using AI tools to accelerate their workflow.”

So Jackson started Supermaven, an AI coding platform along the lines of Tabnine but with a few quality of life and technical upgrades.

Supermaven’s in-house generative AI model, Babble, can understand a lot of code at once, Jackson says, thanks to a 1 million-token context window. (In data science, tokens are subdivided bits of raw data — like the syllables “fan,” “tas” and “tic” in the word “fantastic.”)

A model’s context, or context window, refers to input data (e.g. code) that the model considers before generating output (e.g. additional code). Long context can prevent models from “forgetting” the content of recent docs and data, and from veering off topic and extrapolating wrongly.

“Our large context window helps reduce the frequency of hallucinations because it lets the model draw answers from the context in situations where it would otherwise have to guess,” Jackson said.

One million tokens is a big context window, indeed. But it’s not bigger than AI coding startup Magic’s, which is 100 million tokens. Meanwhile, Google’s recently introduced Code Assist tool matches Supermaven’s context at 1 million tokens.

So what are Supermaven’s advantages over rivals? Well, Jackson claims that Babble is lower-latency thanks to a “new neural architecture.” He wouldn’t elaborate beyond saying that the architecture was developed “from scratch.”

“Supermaven spends 10 to 20 seconds processing a developer’s code repository to become familiar with its APIs and the unique conventions of its codebase,” Jackson said. “With lower latency because of our in-house model serving infrastructure, our tool remains responsive while working with the long prompts that come with large codebases.”

The market for AI coding tools is a large and growing one, with Polaris Research projecting that it’ll be worth $27.17 billion by 2032. The vast majority of respondents in GitHub’s latest dev poll say that they’ve adopted AI tools in some form, and over 1.8 million people — and ~50,000 businesses — are paying for GitHub Copilot.

But Supermaven — along with startup competitors like Cognition, Anysphere, Poolside, Codeium, and Augment — have ethical and legal challenges to overcome.

Businesses are often wary of exposing proprietary code to a third party; for instance, Apple reportedly banned staff from using Copilot last year, citing concerns about confidential data leakage. Some code-generating tools trained using restrictively licensed or copyrighted code have been shown to regurgitate that code when prompted in a certain way, posing a liability risk (i.e. developers that incorporate the code could be sued). And, because AI makes mistakes, assistive coding tools can result in more mistaken and insecure code being pushed to codebases.

Jackson said that Supermaven doesn’t use customer data to train its models. He did admit, however, that the company retains data for a week to “make the system quick and responsive,” he said. On the subject of copyright, Jackson didn’t explicitly deny that Babble was trained on IP-protected code — only that it was “trained almost exclusively on publicly available code rather than a scrape of the public internet” to “reduce exposure to toxic content during training.”

Customers don’t appear to be dissuaded. More than 35,000 developers are using Supermaven, Jackson says, and a sizeable chunk are paying for the premium Pro ($10 per month) and Team ($10 per month per use) plans. Supermaven’s annual recurring revenue reached $1 million this year on the back of a user base that’s grown 3x since the platform’s February launch.

That momentum got the attention of VCs.

Supermaven this week announced its first outside funding: a $12 million round led by Bessemer Venture Partners and high-profile angel investors including OpenAI co-founder John Schulman and Perplexity co-founder Denis Yarats. Jackson says that the plan is to spend the money on hiring developers (Supermaven has a five-person team presently) and developing Supermaven’s text editor, which is currently in beta.

“We plan to grow significantly through the end of the year,” he added. “Despite headwinds for tech overall, the market for coding copilots has been growing quickly. Our growth since our launch in February — as well as our most recent funding round — position us well as we head into next year.”

Typeface Expands AI Portfolio with Treat and Narrato Acquisitions

Dakidarts — Fri, 27 Sep 2024 16:38:34 +0000

Typeface, a generative AI startup focused on enterprise use cases, has acquired a pair of companies just over a year after raising $165 million at a $1 billion valuation.

Typeface revealed on Monday that it has purchased Treat, a company using AI to create personalized photo products, and Narrato, an AI-powered content creation and management platform.

Treat and Narrato will “enrich [Typeface’s] multimodal capabilities,” the company said in a press release, while “propelling [its] vision of end-to-end content lifecycle transformation.”

“Building on our foundation of multimodal AI workflows, these acquisitions’ top-tier AI technology and talent further enrich our visual and textual capabilities,” Typeface wrote in the release. “By integrating these technologies, we’re supercharging the entire Typeface portfolio.”

Typeface, founded in 2022 by former Adobe CTO Abhay Parasnis, offers tools for text and image generation, a fine-tuning engine to personalize AI to a brand’s style, and integrations with third-party apps, software, and services. Typeface claims to place a greater emphasis on brand governance and privacy than its generative AI rivals; for example, Typeface trains dedicated AI models for each customer to ensure their assets and activity remain private.

So how do Treat and Narrato fit into this vision? Well, both were started by founders well-acquainted with the enterprise landscape. And — not for nothing — the startups offer products appealing to the sorts of corporate clients with whom Typeface does business.

NYC-based Treat, the brainchild of Matt Osman and ex-Drizly CTO Hugh Hunter, uses a company’s data on customers to generate product images that incorporate elements known to perform well with certain target demographics. For example, if a fruit vendor’s data suggested that younger men prefer seeing food ads that show a person eating the product, Treat may create an ad that depicts someone biting into fruit.

An image generated by Treat.

Image Credits: Treat

An Australian venture, Narrato — which coincidentally also launched in 2022 — sells access to an “AI content assistant” designed to help orgs achieve their internal content creation and planning goals. As founder Sophia Solanki explained in an interview last March, Narrato customers also get collaboration and workflow tools including templates for articles, video scripts, blogs, emails, social media content, art, and more.

Image Credits: Narrato

Treat raised at least $8.5 million from investors including Greylock prior to the acquisition, while Narrato manage to raise more than $1 million from Airtree Ventures, OfBusiness, and serial entrepreneur Shreesha Ramdas.

Typeface wouldn’t disclose the terms of either acquisition.

Treat and Narrato mark the third and fourth acquisitions for Typeface, which purchased AI photo and video editing suite TensorTour in January and chatbot app Cypher in May. It’s unclear how much of a dent those deals have made in Typeface’s $165 million war chest.

DryMerge: Bridging the Gap Between Incompatible Apps

Dakidarts — Mon, 16 Sep 2024 10:28:43 +0000

DryMerge: Bridging the Gap Between Incompatible Apps

Platforms to connect apps that wouldn’t normally talk to each other have been around for a minute (see: Zapier). But they have not gotten dramatically simpler to use if you’re nontechnical. Generative AI has lowered the barrier to entry somewhat. However, getting the most out of these platforms — and fixing things when they break — still requires a bit of programming know-how.

Software developers Sam Brashears and Edward Frazer perceived this to be the case as well. During internships at tech giants like Meta and Stripe, they struggled to get automations working using some of the more popular app-linking tools.

“I’d been dealing with the pain of designing integrations and automations from scratch,” Frazer said in an interview. “And Sam believed that generative AI models would solve the biggest problem in integrations — transforming data between APIs.”

So Brashears and Frazer, longtime friends who’d been building software together since elementary school, decided to try their hands at a streamlined, easy-to-use app-to-app integration platform.

DryMerge is the fruit of their work. A chatbot for building workflows, DryMerge lets you describe an automation you want between apps — for instance, “Whenever I get an email from a new prospect, ping the team on Slack and add them to HubSpot” — and handles the necessary technical scaffolding.

“Currently, IT departments use complicated no-code tools to automate workflows on behalf of non-IT teams,” Frazer said. “A natural language interface opens up automation to nontechnical people.”

It sounded like a neat idea, a chatbot that can string apps together for you — particularly if you, like me, have spent countless hours wrestling with IFTTT. So, I decided to give DryMerge a go, hoping to replace my old and rickety automations once and for all.

DryMerge’s UI is quite clean and minimalist. It reminds me a bit of ChatGPT; there’s not much to look at besides a text bot. Each new request (e.g., “Text me a summary of my calendar meetings every morning”) starts a new chat session, and these sessions can be revisited at any time from a list on the left-side panel.

DryMerge’s automations management screen.

Image Credits: DryMerge

DryMerge hooks into an expanding library of apps, including Gmail, Microsoft Outlook, Salesforce, storage services like Dropbox and OneDrive, social media platforms (e.g., X), and messaging clients (e.g., Discord). Once the platform creates an automation with these, it plops that automation into a dedicated window showing when the automation last run and whether DryMerge encountered any errors.

I tried setting up a few automations I thought might be useful for a reporter with an overfull schedule, like one to throw Gmail contacts into a spreadsheet and add dates from recent email invitations to a Google Calendar. Things started out promising — DryMerge had me log into the relevant apps and asked whether I’d like to test the automations to ensure everything was working properly.

But then, problems started to crop up.

Several times, DryMerge’s chatbot stopped responding altogether. Other times, it missed key details in a request. I tried repeatedly to get DryMerge to understand that I wanted to copy Gmail contacts to my Google Calendar, but every attempt, it thought I wanted to manually enter contacts into a spreadsheet.

The setbacks didn’t completely ruin my DryMerge experience. Giving credit where it’s due, the platform’s nifty when it works. For example, I successfully got DryMerge to set up an automation that copies posts from my X account to the personal Discord server I use to aggregate various notifications. A niche use case? Perhaps. But it’s going to save this reporter a lot of task switching.

Chatting with DryMerge’s bot.

Image Credits: DryMerge

The bugs, Frazer assures me, will be addressed in time. He and Brashears are DryMerge’s only employees, so there’s lots on the to-do list.

“We think we’re well-positioned to iterate quickly and nimbly,” Frazer said.

Assuming Frazer and Brashears can get DryMerge’s platform in good working condition, the bigger challenge the duo will have to face is staying relevant in the fiercely competitive integration-platform-as-a-service (iPaaS) space. According to recent poll released by IDG and TeamDynamix, iPaaS is one of the fastest-growing software markets, projected to reach $2.7 billion this year.

AWS has its own iPaaS called AppFabric. IBM recently acquired iPaaS tech from Software AG. A growing number of startups aside from DryMerge are attempting to break into the segment, while incumbents like Zapier and IFTTT are aggressively deploying generative AI capabilities.

Frazer makes the case that DryMerge’s differentiator is — and will remain — “being 10x easier to use” than drag-and-drop integration builders.

“Our users include online fashion retailers, school administrators, and asset managers — the vast majority of which have never touched a line of code,” he said. “They use us to save hours a day on tasks ranging from customer support automation to customer relationship management data entry.”

Frazer’s not wrong about the opportunity. Per the IDG and TeamDynamix poll, 66% percent of companies said that they’ll invest in iPaaS to address internal automation and data integration challenges.

“We think a gigantic enterprise opportunity is in increasing the simplicity of automation and delivering easy-to-use tooling that empowers nontechnical folks,” Frazer said.

It’s very early days for DryMerge, which only has around 2,000 users at present. But the company was accepted into Y Combinator’s Winter 2024 batch, and DryMerge this past summer closed a $2.2 million seed round led by Garage Capital with participation from Goodwater Capital, Ritual Capital, and angels whose names Frazer wouldn’t reveal.

Frazer says that the funds are being put toward adding new app integrations and doubling the size of DryMerge’s team in the next few months.

Codeium Raises $150M, Challenges GitHub Copilot in AI Coding Market

Dakidarts — Sat, 07 Sep 2024 12:41:34 +0000

Codeium Raises $150M, Challenges GitHub Copilot in AI Coding Market

A startup whose product competes with GitHub Copilot and other AI-powered coding assistants has achieved unicorn status.

On Thursday, Codeium said it closed a $150 million Series C round led by General Catalyst that values the company at $1.25 billion post-money. The round, which also saw participation from existing investors Kleiner Perkins and Greenoaks, brings the company’s total funding raised to nearly a quarter-billion dollars ($243 million) a mere three years since its launch.

Codeium’s co-founder and CEO, Varun Mohan, said that Codeium hasn’t even touched the $65 million Series B tranche it raised in January yet. Back then, just eight months ago, Codeium was valued at half a billion dollars.

“Even though we’ve barely made a dent in our existing funding, we believe that this injection of capital will allow us to significantly ramp up R&D and growth while making even larger strategic bets,” he said.

Codeium was founded in 2021 by Mohan and his childhood friend and fellow MIT grad, Douglas Chen. Prior to Codeium, Chen was at Meta, where he helped to build software tools for VR headsets like the Oculus Quest. Mohan was a tech lead at Nuro, the autonomous delivery startup, responsible for managing the autonomy infrastructure team.

The startup began as a radically different company called Exafunction, focused on GPU optimization and virtualization for AI workloads. But in 2022, Mohan and Chen sensed a bigger opportunity in generative coding and decided to rebrand — and pivot.

“Despite the influx of generative AI tools, developers are still struggling with time-consuming coding tasks,” Mohan said. “Many of the AI-driven solutions provide generic code snippets that require significant manual work to integrate and secure within existing codebases. That’s where our AI coding assistance comes in.”

Codeium’s platform, powered by generative AI models trained on public code, serves up suggestions in the context of an app’s entire codebase. It supports around 70 programming languages and integrates with a number of popular development environments, including Microsoft Visual Studio and JetBrains.

Codeium’s AI-generated coding suggestions.

Image Credits: Codeium

To attract devs away from Copilot and other rivals, Codeium has released a generous free tier to start. The strategy seems to have worked: Today, the startup has more than 700,000 users and over 1,000 enterprise customers, including Anduril, Zillow and Dell.

Quentin Clark, managing director at General Catalyst, implied that Codeium won some of its larger contracts by embracing a steadfastly client-centric approach to product research.

“The team’s approach has always been to follow its customers, leading the company to build solutions on their terms — deployable in any environment and supporting more languages than anyone else,” Clark said in a statement. “What Codeium has created isn’t just a demo, an announcement, or an idea — this is a fully scaling business, with large enterprises adopting the product across their entire organizations.”

Businesses are often wary of exposing proprietary code to a third party — for instance, Apple reportedly banned staff from using Copilot last year, citing concerns about confidential data leakage. To attempt to allay such fears, Codeium began offering a self-hosted installation option alongside its standard software-as-a-service plan.

Image Credits: Codeium

Companies can now deploy Codeium’s service on their own hardware if they wish. Or they can adopt a hybrid setup, keeping their data on their own devices while using Codeium’s servers for computing needs.

There’s always some risk involved in data transfers to the cloud, but Mohan claimed that Codeium leverages strong encryption. “We never train our proprietary generative autocomplete model on user data, never sell data and ensure all data transmission is encrypted,” he added.

Codeium has also taken steps to remove “non-permissively” licensed code (e.g., code under copyright) from the datasets it used to train its AI models. Some code-generating tools trained using restrictively licensed or copyrighted code have been shown to regurgitate that code when prompted in a certain way, posing a liability risk (i.e., developers that incorporate the code could be sued). Mohan said that’s not the case with Codeium, thanks to its training data prep and filtering approach.

“We also remove any remaining data that looks similar to code that is explicitly non-permissively licensed just in case other people copied code without providing the proper attribution and licensing,” he added. “On top of this, we have state-of-the-art, post-generation attribution filtering and logging in the case that these large probabilistic models produce code that is similar to public code, whether permissively or non-permissively licensed.”

But what about hallucinations? Most AI coding tools are notorious for making stuff up, which can be quite destructive in an enterprise environment.

An analysis by developer tooling startup GitClear found that generative AI tools have resulted in more mistaken code being pushed to codebases over the past few years. And a Purdue study found that over half the answers that OpenAI’s ChatGPT gives to programming questions are incorrect. Security researchers have warned of the potential for such tools to amplify existing bugs in software.

A recent survey from cybersecurity firm Snyk found that nine in ten developers worry about the broader security implications of using AI coding platforms. But Mohan claimed that Codeium’s supposedly superior, deep context-rich tech yields more trustworthy results than most.

“Our context awareness engine is able to ground results in what is already existing in a user’s codebase, leading to suggestions with fewer hallucinations and more adherence to existing syntax, semantics and standards,” he said.

Whether benchmarks back that up or not, Codeium’s sales pitch seems to be resonating with the right execs: Revenue hit eight figures this year. Mohan said the 80-person, Mountain View-based startup plans to expand headcount to 120 by 2025 as it aims to make a bigger dent in a market with formidable competitors like Tabnine, Anysphere and Poolside.

Catching up to Copilot, which had over 1.8 million paying users as of April, probably isn’t in the cards for Codeium — at least not imminently. It doesn’t have to be. As Mohan rightly noted, given the widespread adoption of AI coding tools among developers (despite their reservations), even a small slice of the nascent segment is bound to be lucrative.

Polaris Research projects that the AI code tools market will be worth $27.17 billion by 2032.

“An overabundance of hype is a challenge the industry faces,” Mohan said. “This will make it harder for every company to truly convince end users that they are at the forefront of possibility. But we believe that truth-seeking and realistic AI companies like Codeium will eventually cut through this noise.”

Magic Raises $320M: AI Coding Startup Gets Major Boost

Dakidarts — Fri, 06 Sep 2024 07:54:27 +0000

Magic Raises $320M: AI Coding Startup Gets Major Boost

Magic, an AI startup creating models to generate code and automate a range of software development tasks, has raised a large tranche of cash from investors, including ex-Google CEO Eric Schmidt.

In a blog post on Thursday, Magic said that it closed a $320 million fundraising round with contributions from Schmidt, as well as Alphabet’s CapitalG, Atlassian, Elad Gil, Jane Street, Nat Friedman and Daniel Gross, Sequoia and others. The funding brings the company’s total raised to nearly half a billion dollars ($465 million), catapulting it into a cohort of better-funded AI coding startups whose members include Codeium, Cognition, Poolside, Anysphere and Augment. (Interestingly, Schmidt is backing Augment, too.)

In July, Reuters reported that Magic was seeking to raise over $200 million at a $1.5 billion valuation. Evidently, the round came in above expectations, although the startup’s current valuation couldn’t be ascertained; Magic was valued at $500 million in February.

Magic also on Thursday announced a partnership with Google Cloud to build two “supercomputers” on Google Cloud Platform. The Magic-G4 will be made up of Nvidia H100 GPUs, and the Magic G5 will use Nvidia’s next-gen Blackwell chips, scheduled to come online next year. (GPUs, thanks to their ability to run many computations in parallel, are commonly used to train and serve generative AI models.)

Magic says it aims to scale the latter cluster to “tens of thousands” of GPUs over time, and that together, the clusters will be able to achieve 160 exaflops, where one exaflop is equal to one quintillion computer operations per second.

“We are excited to partner with Google and Nvidia to build our next-gen AI supercomputer on Google Cloud,” Magic co-founder and CEO Eric Steinberger said in a statement. “Nvidia’s [Blackwell] system will greatly improve inference and training efficiency for our models, and Google Cloud offers us the fastest timeline to scale, and a rich ecosystem of cloud services.”

Steinberger and Sebastian De Ro co-founded Magic in 2022. In a previous interview, Steinberger told Dakidarts that he was inspired by the potential of AI at a young age; in high school, he and his friends wired up the school’s computers for machine-learning algorithm training.

That experience planted the seeds for Steinberger’s computer science Bachelor’s program at Cambridge (he dropped out after a year) and, later, his job at Meta as an AI researcher. De Ro hailed from German business process management firm FireStart, where he worked his way up to the role of CTO. Steinberger and De Ro met at the environmental volunteer organization Steinberger co-created, ClimateScience.org.

Magic develops AI-driven tools (not yet for sale) designed to help software engineers write, review, debug and plan code changes. The tools operate like an automated pair programmer, attempting to understand and continuously learn more about the context of various coding projects.

Lots of platforms do the same, including the elephant in the room GitHub Copilot. But one of Magic’s innovations lies in its models’ ultra-long context windows. It calls the models’ architecture “Long-term Memory Network,” or “LTM” for short.

A model’s context, or context window, refers to input data (e.g. code) that the model considers before generating output (e.g. additional code). A simple question — “Who won the 2020 U.S. presidential election?” — can serve as context, as can a movie script, show or audio clip.

As context windows grow, so does the size of the documents — or codebases, as the case may be — being fit into them. Long context can prevent models from “forgetting” the content of recent docs and data, and from veering off topic and extrapolating wrongly.

Magic claims its latest model, LTM-2-mini, has a 100 million-token context window. (Tokens are subdivided bits of raw data, like the syllables “fan,” “tas” and “tic” in the word “fantastic.”) One hundred million tokens is equivalent to around 10 million lines of code, or 750 novels. And it’s by far the largest context window of any commercial model; the next-largest are Google’s Gemini flagship models at 2 million tokens.

Magic says that thanks to its long context, LTM-2-mini was able to implement a password strength meter for an open source project and create a calculator using a custom UI framework pretty much autonomously.

The company’s now in the process of training a larger version of that model.

Magic has a small team — around two dozen people — and no revenue to speak of. But it’s going after a market that could be worth $27.17 billion by 2032, according to an estimate by Polaris Research, and investors perceive that to be a worthwhile (and possibly quite lucrative) endeavor.

Despite the security, copyright and reliability concerns around AI-powered assistive coding tools, developers have shown enthusiasm for them, with the vast majority of respondents in GitHub’s latest poll saying that they’ve adopted AI tools in some form. Microsoft reported in April that Copilot had over 1.8 million paying users and more than 50,000 business customers.

And Magic’s ambitions are grander than automating routine software development tasks. On the company’s website, it speaks of a path to AGI — AI that can solve problems more reliably than humans can alone.

Toward such AI, San Francisco-based Magic recently hired Ben Chess, a former lead on OpenAI’s supercomputing team, and plans to expand its cybersecurity, engineering, research and system engineering teams.

Trunk Tools: From Carpenter to AI Construction Startup CEO

Dakidarts — Tue, 27 Aug 2024 10:41:04 +0000

Trunk Tools: From Carpenter to AI Construction Startup CEO

Construction companies deal with a lot of documents — so many that it can be difficult to process and manage them all. According to one recent survey, a third of construction professionals found accessing documents to be a challenge in completing a project, while a fourth said that inaccurate project paperwork has contributed to a construction delay.

Sarah Buchner knows this well. Originally a carpenter, she founded a startup, Trunk Tools, that provides automation tools to organize unstructured construction documentation.

“I grew up in a poor environment in a small village in Austria and started working as a carpenter at age 12,” Buchner said. “After many years in carpentry, I switched over to the general contractor side and worked my way up from superintendent to project manager to group leader. My PhD research made me realize that I could have a greater impact on my field by developing disruptive construction technology, and this inspired me to move across the world to Silicon Valley to attend Stanford and get my MBA.”

Trunk Tools’ platform can take in files like PDFs, spreadsheets, drawings, blueprints and tables and answer questions about them in a chatbot-like interface (e.g. “What type of power outlets are in the art studio?”). Trunk Tools can also “link” scheduled construction activities with supporting documentation, attempting to spot potential project issues and surface insights.

Image Credits: Trunk Tools

“Traditional construction software, like Procore, is centered around documenting workflows and storing data within a predefined system,” Buchner said. “In contrast, we’re introducing a paradigm shift where Q&A and AI enable construction teams to interact with information using natural language.”

Buchner says that for one customer’s $500 million high-rise condo in NYC, there were 3.6 million pages of documentation. Given the amount of time it takes to sort through file folders that massive, it’s not exactly surprising that construction industry workers loathe paperwork.

A poll by Dodge Data and Viewpoint, a construction accounting software vendor, found that only 28% of contractors were okay using paper processes, while just 47% said they were satisfied with spreadsheets. Seventy-nine percent of respondents to the poll expressed a willingness to adopt construction management tooling.

“If printed and stacked, the 3.6 million pages would be 3x the height of the building itself,” Buchner said. “It would take a human 50 years to read — it takes Trunk Tools seconds to structure and give insights.”

Image Credits: Trunk Tools

Occupying a construction software market that could be worth $7.5 billion by 2032, Trunk Tools competes with vendors like Briq (which uses AI to automate construction financial processes), Join (a “decision-making” platform for construction) and PlanRadar (which digitizes construction and real estate docs).

Trunk Tools appears to be holding its own, however, with a “double digit” number of construction industry customers and thousands of users. Buchner says that the company is targeting a 4x revenue to burn rate ratio.

To help get it there, Trunk Tools this month closed a $20 million Series A funding round led by Redpoint. Bringing the company’s total raised to $30 million, the new cash will be put toward growing Trunk Tools’ 30-person, New York-based team as well as developing new services like Trunk’s recently launched construction worker incentive program, Buchner says.

“Construction technology so far has focused mainly on digitizing — taking what we used to do on paper and doing it on computers,” Buchner said. “Slipped timelines and rework can completely crush the razor-thin margins of construction projects, and Trunk Tools can alleviate both.”