Kim Burgaard

Claude Sonnet 4.6 dropped yesterday and it's legitimately competing with Opus. Sonnet 4.6 did better than Opus 4.6 in a few technical problem solving experiments at Checksum.ai yesterday. Super exciting! Sonnet costs a fraction of Opus pricing but now handles what used to require the premium model. Here's what changed: 1M token context window means entire codebases in a single API call. Computer use jumped from 14.9% to 72.5% on OSWorld benchmarks. That's near human performance. Anthropic's Opus is still better for "deep reasoning". But if you're paying for Opus to power your daily workflows, this might be the nudge to downgrade and pocket the savings. The intelligence gap basically disappeared overnight. The AI model wars are getting spicy 🚀

11

Scaling your startup on AWS without burning through your budget? Sounds like a dream, but AWS actually hands out credits and runs programs that can save you serious cash - if you know how to grab them. This guide breaks down how startups can tap into the AWS accelerator, snag free credits, and turbocharge growth without sweating over the cloud bill. Spoiler: it’s less complicated than you think, and skipping this means leaving money on the table. #AWS #Startups #CloudComputing #TechGrowth #Entrepreneurship

2

If you’re a CEO/founder or PE operator and your tech org is: ❌ Shipping late ❌ Stuck in reactive mode ❌ Struggling through the Series A/B jump, or post-acquisition integration or after a major release It’s not “a delivery problem.” It’s an operating model problem. At the inflection point, the old playbook breaks: what worked at 10 engineers fails at 40… and breaks again across multiple products/teams. That’s where I come in as a fractional CTO: - Spot platform and delivery risks before they hit revenue and retention - Rebuild execution rituals (prioritization, ownership, release confidence) without derailing the roadmap - Align architecture + team structure to your value-creation plan (growth, margin, or exit timeline) If you’re feeling the strain of growth, I can help you turn “busy and reactive” into predictable shipping, and a platform that scales. #CTO #SaaS #StartupScale #PrivateEquity #TechStrategy

8

3 Comments

We just launched CloudRig Estimating — a big step toward our vision of building a system where every job makes the next one smarter. We’ve always believed better data is the foundation of better operations. Estimating sets the standard — and when it’s grounded in real production, the entire feedback loop strengthens. For too long, estimating has lived separately from the field. You build a bid, then months later find out how close you were. CloudRig Estimating tightens that loop. It’s fast — teams are turning bids in hours instead of days — but more importantly, every estimate connects to real production history: time cards, daily reports, field observations. Smarter estimates → better targets → tighter tracking → sharper estimates The goal isn’t just to bid faster. It’s to get better with every job you run. Book a demo in the comments!

28

2 Comments

LLM state management for a text rpg (...not usually my jam, but I want to use it to help me learn classical languages better) and LLM state management for a personal assistant have a lot of overlaps, and working on them in parallel has repeatedly helped me make breakthroughs on the other side. JSON documents, as a whole - not specific fields, like Mongo Vector search - should be trivially handled by embeddings and vector search. Not quite 'done', but here's some progress.

2

𝙀𝙫𝙚𝙧𝙮𝙤𝙣𝙚’𝙨 𝙩𝙚𝙨𝙩𝙞𝙣𝙜 𝘼𝙄 𝙩𝙤𝙤𝙡𝙨. 𝘽𝙪𝙩 𝙝𝙚𝙧𝙚’𝙨 𝙬𝙝𝙖𝙩 𝙣𝙤 𝙤𝙣𝙚’𝙨 𝙩𝙖𝙡𝙠𝙞𝙣𝙜 𝙖𝙗𝙤𝙪𝙩: *** Adoption is the real bottleneck. At Aumet inc, we’ve been hands-on with tools like: - Cursor for pair programming - Windsurf for orchestrating tasks - Claude Code for code review - Tabnine for code autocompletion Yes, the tools work. But the real challenge? Transforming how people work. It’s not about model quality. It’s about: - Legacy workflows - Teams waiting for permission to adapt - Rigid organizational structures Adoption is not installation. Transformation takes: - Internal champions - Education across departments - Controlled pilots, not mass rollouts - Iteration, not plug-and-play Here’s what surprised us most at Aumet inc: While developers test AI tools - The finance team is automating workflows—no code. - Product ops is summarizing feedback loops instantly. - Business leads are building dashboards using prompts, not SQL. 𝙏𝙝𝙚 𝙦𝙪𝙚𝙨𝙩𝙞𝙤𝙣 𝙞𝙨𝙣’𝙩: “𝙒𝙝𝙞𝙘𝙝 𝘼𝙄 𝙩𝙤𝙤𝙡 𝙞𝙨 𝙗𝙚𝙨𝙩?” 𝙄𝙩’𝙨: “𝙃𝙤𝙬 𝙙𝙤 𝙬𝙚 𝙢𝙖𝙠𝙚 𝙚𝙫𝙚𝙧𝙮 𝙩𝙚𝙖𝙢 𝘼𝙄-𝙣𝙖𝙩𝙞𝙫𝙚?” The companies that win this race won’t be the ones with the flashiest tools— They’ll be the ones who: - Flatten access to intelligence - Empower non-technical teams - Build new habits and workflows around AI At Aumet inc, we’re not just experimenting. We’re evolving how we work. 𝙃𝙤𝙬 𝙞𝙨 𝙮𝙤𝙪𝙧 𝙤𝙧𝙜𝙖𝙣𝙞𝙯𝙖𝙩𝙞𝙤𝙣 𝙚𝙣𝙖𝙗𝙡𝙞𝙣𝙜 𝙣𝙤𝙣-𝙩𝙚𝙘𝙝𝙣𝙞𝙘𝙖𝙡 𝙩𝙚𝙖𝙢𝙨 𝙩𝙤 𝙬𝙤𝙧𝙠 𝙬𝙞𝙩𝙝 𝘼𝙄? 𝙒𝙤𝙪𝙡𝙙 𝙡𝙤𝙫𝙚 𝙩𝙤 𝙝𝙚𝙖𝙧 𝙝𝙤𝙬 𝙮𝙤𝙪’𝙧𝙚 𝙣𝙖𝙫𝙞𝙜𝙖𝙩𝙞𝙣𝙜 𝙩𝙝𝙞𝙨 𝙨𝙝𝙞𝙛𝙩. #aumet #AI #DigitalTransformation #NoCode #Claude #Cursor #Tabnine #Windsurf #FutureOfWork #Automation

37

2 Comments

We’re in the thick of it here at TalosTitle and I’d like to carve out a few moments to celebrate a recent joyful experience regarding Agentic Assisted Coding.  Particularly on how it opens the opportunity for more great people to contribute to the critical path on production software. Recently, Adrian Horner asked to contribute to our primary web product using Claude Code. I’ve been working with Adrian filling a quality and test capacity on my engineering teams on and off for about 20 years.  He now leads our Quality Engineering Team at Talos, where his primary focus is on the structured output of Large Language Models (LLMs) processing legal instruments. This work also includes developing Agentic capabilities and optimizing training outcomes. A little TLDR;  Until 3 weeks ago, Adrian hadn’t ever contributed features to the customer facing product.  Sure, he’s checked in test suites.  In my previous organizations he led the development of entire robotic based quality stacks.  With all of this, he’s never dropped a feature into prod on a product shipped to customers. After Adrian asked to try his hand, we selected some features from our “Table Stakes Backlog”, walked him through the process of how we would build the features, helped him get Claude Code up and running in the repos he had checked out… and then we got out of the way. The fact that he now has features flowing into prod every couple days isn’t exactly what I want to celebrate.  It’s his discoveries, his change in awareness that is so very important to note. A few weeks ago, Adrian got a feature into our pre-prod environment and started to hammer away at it.  He is very good at this as you might imagine.  He found a bug and reached out to me saying: “I found this bug in a corner case that we didn’t talk about in the Backlog Ticket, what do you think I should do about this?”  My response was pretty standard: “Well Adrian, you need to write the bug up and work with the engineer who wrote the code to figure out if the corner case should be addressed with this feature delivery, or as a follow-on to the feature.”  And he did. This next piece is what I’d like to celebrate: Later, Adrian said to me: “I really love building these features and shipping them, but dealing with the corner case I didn’t think about when building the feature has helped me have a greater sense of empathy for the different pressures software engineers juggle. What a great partnership we form with Product, Engineering, and Quality as a close knit team.” Wow.  Let’s get more people into the thick of it. Thank you Adrian Horner for daring to explore new ways to contribute.

24

5 Comments