Workplace chaos is the norm. Our days are filled with distractions across multiple tools and channels. That Works is an AI-powered solution that interconnects your workplace tools and extracts meaningful insights from all that data. With That Works, you get contextual summaries of key information, automatic tracking and visualization of critical data, insights into what you should focus on and when.
External link for That Works
San Francisco Bay Area, US
That Works reposted this
If you spend hours buried in Slack channels (or actively avoiding them when they get too noisy) - then this video is for you. That Works can now summarize Slack channels and automatically highlight key discussions, updates and conversations that you need to stay on top of without forcing you to constantly switch context. Sign up today for a free trial at https://thatworks.ai
That Works reposted this
Scaling Laws Be Damned The history of transformative software products teaches us that the big business opportunities are in the unglamorous problems that silently grind down productivity, plague daily workflows, and eat into profitability. We've watched LLMs swing like massive hammers at all kinds of problems. They've changed how we discover information and generate content. But we've hardly scratched the surface in building actual products – not demos, not magical prototypes that fail to stick. I'm talking about the hard work of constant iteration. Of finding not just strengths but the critical weaknesses of these technologies. Anyone who's spent time prompt engineering knows how quickly you hit diminishing returns. And no, making RAG your universal solution isn't the answer when there's a whole universe of approaches worth exploring. Nor is dumping lots of data into an LLM and expecting consistent magic. Are LLMs hitting limits? Are we discovering scaling laws? Who cares. We've hardly begun to exhaust what the current generation can do. Stop chasing technology. Start building products that people love. Everything else be damned.
That Works reposted this
Looking forward to, very shortly, chatting about all things M&A with some very fine company!
Have you signed up to our upcoming free M&A event yet? Let us introduce our panel of leading entrepreneurial minds: 👂 Varun Nair – CTO of That Works and former CTO of TwoBigEars, sold to Facebook 📺 Calum Smeaton – Former CEO of TV Squared, sold to Innovid 🧬 Chris Wright – Founder and former CTO of DeltaDNA, sold to Unity Software All experts in their own right, we can’t wait to hear their secrets to start-up success and a triumphant exit to US buyers in our session led by Stephen Trombala. Our friends and co-sponsors of the Turing Fest Founders Dinners, RBC Brewin Dolphin, have kindly supported this event by providing wine. We look forward to seeing you there for an evening of drinks, networking, and learning. Click here to register for this free event: https://lnkd.in/ed5-jCSa 📍 Shepherd and Wedderburn's Edinburgh Office 📅 Tuesday, 29 October ⏰ 5:30pm – 9pm
Seamless updates on work across different teams? ✅. Automatic visualization of key data points to call out critical work? Also ✅.
Data dashboards are great. Except when they are not. At That Works, we use "Indicators" to automatically generate visual insights from everyday work tools like JIRA and Notion. This saves time on reporting while keeping different teams from engineering to sales and marketing aligned through contextualized data summaries.
While we use LLMs to do new amazing things, they are also great at making small product user interactions magical.
Calling products "LLM-wrappers" is like calling everything out there a "database-wrapper". It doesn't make sense. The internet is full of threads like "10 🤯 things a LLM can do 👇" but I think one of the more interesting areas is where you can use LLMs to reduce friction in different parts of your product. A real-world example: we noticed a friction point while onboarding users onto That Works where after creating an automation, they'd need to name it. Coming up with names slows people down, so we used a LLM to suggest a name automatically based on the data they have selected. It took 10 minutes to add this feature and saves our users hundreds of collective minutes. The cool part is you don't even need a super large language model. Something like Mistral 7B can do a great job at a negligible inference cost. Other ideas: Have a list of preset sentences you use to welcome a user into the product? Use those sentences as prompts to get a LLM to create something more personalized and unique. Have a UI flow that needs lots of basic information? Use a LLM to create placeholders that a user can easily accept and edit. So much low hanging fruit for every kind of product out there—even if you are busy fine-tuning or training your own foundation models for something 🤯 #LLM #AI #BuildInPublic #product #WhyAmIUsingAHashTag
That Works reposted this
How can we use LLMs for complex tasks—reliably? Using LLMs for complex, multi-step problems is challenging. Errors add up at each step and the models can "reason" differently each time. This can make them unpredictable for advanced tasks. The fix? Combine rule-based systems with smaller, more accurate models that work with LLMs and enforce guardrails. The result: better accuracy, control, and transparency that builds trust in your systems and your product. Here's a real-world example: That Works is used for a variety of use cases and one of which is where engineering leads automatically generate pre-reads for their meetings. Blindly getting LLMs to summarize information from various APIs is insufficient and noisy. People care about specific things: What work is at risk and why? What code have we shipped this week and how long did it take? What tasks are blocked and what's their impact? By using specific models and algorithms to define concepts that map to people's expectations, our LLMs become significantly more effective. They ensure consistent, reliable, and meaningful information every time. This approach also makes it much easier to detect and debug issues, rather than struggling with the black box of prompt engineering. We're tackling some pretty novel problems and solutions while we build out That Works. I'll continue to share more every week!
Getting LLMs to work reliably for multi-step problems across a range of use cases and data types isn't trivial. Our co-founder and CTO Varun Nair digs into some of the details of how we do things differently at That Works.
How can we use LLMs for complex tasks—reliably? Using LLMs for complex, multi-step problems is challenging. Errors add up at each step and the models can "reason" differently each time. This can make them unpredictable for advanced tasks. The fix? Combine rule-based systems with smaller, more accurate models that work with LLMs and enforce guardrails. The result: better accuracy, control, and transparency that builds trust in your systems and your product. Here's a real-world example: That Works is used for a variety of use cases and one of which is where engineering leads automatically generate pre-reads for their meetings. Blindly getting LLMs to summarize information from various APIs is insufficient and noisy. People care about specific things: What work is at risk and why? What code have we shipped this week and how long did it take? What tasks are blocked and what's their impact? By using specific models and algorithms to define concepts that map to people's expectations, our LLMs become significantly more effective. They ensure consistent, reliable, and meaningful information every time. This approach also makes it much easier to detect and debug issues, rather than struggling with the black box of prompt engineering. We're tackling some pretty novel problems and solutions while we build out That Works. I'll continue to share more every week!
The right messaging, for the right people, at the right time. Watch Abesh Thakur automatically create release notes and share with relevant stakeholders effortlessly.
Not only is it important to communicate with clarity at work, it's also essential for messaging to be tailored to the audience. To enable this, That Works now allows you to format work summaries as Release Notes. Broadcast your team's achievements with precision and purpose. Capture work in an easy-to-parse, actionable format designed for your audience. Sign up at thatworks.ai
Start your day fully informed, effortlessly. Watch how Abesh Thakur sets up a daily automation for the remote team at That Works. Streamline your routines and focus on what truly matters