Quilt

Generative AI has opened up many opportunities for AI-powered workflow automation, and RFP automation is one of the top use cases. Since we launched the RFP automation product a few months ago, we have had the chance to work with many large enterprise companies to understand what they look for in an AI RFP automation solution. There are a number of options - from incumbent vendors offering new gen AI features to new startups offering LLM-first options, how do you sort through the noise? Our team put together a comprehensive guide and a checklist for any solutions engineering, proposal, and sales teams looking for an RFP automation solution, covering considerations ranging from security, integration, to the ability to audit AI generated responses. Check it out here! #RFPAutomation #AI #RFPBuyersGuide

As a repeat founder, I’ve made a ton of GTM mistakes in both the distant and recent past. Sam Blond gave a great SaaStr presentation recently that covered some of the most impactful ones. I shared these with the team and thought I’d share them here in case it’s helpful for anyone going through the same situation: Mistake 1: Worrying too much about how I sold instead of how many at-bats I got As a technical founder, I was unsure of my ability to sell, so I would hyper-focus on making sure I said the right things. This led to us dragging our feet early on and going through fewer iterative cycles. Most startups are demand-constrained, and many other problems seem to magically go away when you solve that problem. Mistake 2: Doing what everyone else does and not being creative It is easy to follow a “playbook” — set up your outbound sequences, build lookalike audiences, enrich your CRM, build a targeted leads list, etc. But in an LLM-first world where AI outbound is automated, the ROI on standard outbound channels is decreasing. You need to find creative ways to meet your customer where they are. For Salesforce, it was an infamous conference protest outside of Siebel HQ. For us, a couple of sneaky Reddit posts have worked surprisingly well. Mistake 3: Not being prescriptive on the sales process As a tiny startup selling into 800 pound gorillas, it is very easy to fall into the trap of deferring to the customer's timeline, requirements, etc. because you don't want to lose the deal. However, this will often lead to confusion: if you don't know how to buy your product, neither will your customer. Always dictate a process, even if you are still learning and need to adjust it from customer-to-customer. (Side note: this is an area where good demand gen does wonders.) Mistake 4: Treating implementation and onboarding as an afterthought For AI-first companies whose products are dependent on things that are not entirely in your control, expectation setting is even more important. You only have one chance to make a good impression when you start onboarding users, and changing a bad initial impression is orders of magnitude harder than taking the time to start things off with a good one. We learned this the hard way with a few early customers. Follow Quilt to stay updated on we are building! #FounderLedSales #GTM

Grateful to be featured for our work in building the future of GTM! We are just getting started 🚀

Low-latency models are the future. While it is unclear whether or not models are getting better (or hitting an asymptote), it is very clear that models are getting faster and cheaper. In the next 6-12 months, low-latency models will be ubiquitous. Why does this matter? Low-latency models unlock a new universe of use cases, largely centered around real-time conversational AI: a virtual tutor, psychologist, friend, celebrity avatar, SDR, customer support agent, etc. But there are still challenging problems to solve: 1. Latency is still an issue. A startup recently showed us a really impressive demo with ~1s turn latency (“laggy Zoom”). However, the lag was still noticeable. 2. There is still a bit of an uncanny valley in both speech and video, though companies like Tavus and HeyGen are doing amazing work improving this. 3. Similar to the lag in enterprise chatbot adoption vs. consumers, we see a strong aversion in companies to deploying Zoom bots in any active opportunity or deal cycle, whereas tech-forward consumers are more accepting of interacting directly with AI. 4. For enterprise use cases, it is important to ground all responses on proprietary knowledge and eliminate hallucinations entirely (vs. consumers may not care). This requires entirely different infrastructure than a transcription + Grok/Llama completion setup for conversational use cases. Follow Quilt for more! #LowLatencyModels #ConversationalAI #EnterpriseAI

⭐ Introducing Tags ⭐ We are excited to launch tags 🏷 based on customer feedback. When you have multiple products or question categories, tagging helps to keep your knowledge organized. You can filter for questions in a category and bulk assign questions to the relevant stakeholder seamlessly. Check it out at quilt.app!

Some of my most fulfilling moments as a founder are hearing from our customers about how our product makes their day a little better: “Love the tool - it’s a huge time saver / sanity saver.” “This system is amazing - SO much faster than me writing something myself.” “This tool is insanely helpful!!! So thank you!!” “Another example of Quilt saving the day. They demanded that we answer 100+ questions in the 11th hour. Quilt allowed us to complete this request.” “We’re going to revamp our security review to take some load of our security team.” Reach out to me directly if you want to learn more about how we’re helping these customers 10x their solutions teams, and give us a follow at Quilt! #AI #GTM #CustomerLove

🌟 Thanks to our 1,000 Linkedin followers 🌟 Thanks for following our journey. We look forward to sharing more about AI, solutions engineering, and GTM with you. We are just getting started! 🚀

RAG (retrieval-augmented generation) for dummies! 🤓 If I were to explain RAG in one sentence: RAG is an open-book exam while ChatGPT is a closed-book exam. Here’s why👇 Imagine you are studying for an open-note exam and have unlimited time and memory. RAG follows the following steps: 1. You read and take notes over all the course material - these notes are categorized and stored as "embeddings" (numeric representations of those documents in a high-dimensional vector space whose components tend to correspond to facts about the world) 2. When exam time comes around and you get a new question, you do a quick comparison against all of the passages you have jotted down and pick the top passages - this is "ranking" (picking the k most semantically similar passages to the input question) 3. For the top passages, you might do one more pass reading and comparing the questions and candidate passages more closely - this is "re-ranking" (feeding the results to a cross-encoder to re-sort the results accounting for additional context) 4. Finally, you combine all of the passages together with your knowledge of how the world works to come up with the final answer - this is "inference" and is the step that leverages LLMs. While steps 1-3 focus on "searching" for information to answer the question (“retrieval”), step 4 focuses on actually answering the question (“generation”). ChatGPT (or, more broadly, pre-trained foundation models) on the other hand does the following: 1. Read everything there is to know on the internet 2. To fine-tune, you re-read the subset of the documents you care about so you remember them better 3. When someone asks you a question, try to answer it the best you can based on all of your prior knowledge. But do not look up any references or sources beforehand At the risk of over-simplifying, RAG does a bit of research - or just search - on each question whereas using ChatGPT and fine-tuned models draws from prior knowledge of the world. RAG is an open-book exam while ChatGPT is a closed-book exam. There are no sources or references to "ground" your answers. Sometimes this is helpful, such as for creative or expository tasks, but in general, RAG is better for information retrieval (e.g. answering questions, taking a test). For more AI content, give us a follow Quilt! #RAG #LLMs

This has been a long time coming... excited to finally announce Rubrik as a customer! We have been working with Alok Agrawal, Evan Shelley, and team for the past months, and are thrilled to continue the partnership to help them scale their solutions org with AI. 🚀 If you're curious how Quilt is helping companies like Rubrik, reach out to me directly and give us a follow at Quilt!

🔨 How to assign tasks with Quilt 🔨 Half of the battle in RFPs is project management. In Quilt, once you tag a team member, a to-do list is automatically generated for them, making it easy to track unanswered questions. Check it out at quilt.app! #AI #RFP #KnowledgeManagement