Thoughts | Emergence Capital

Cutting Through the Noise to Build an Enduring AI-Enabled Business

April 4, 2023

February shared thoughts current limitations generative AI technology tuned B2B Despite believe early innings transformative era applications leveraging Founders face daunting number questions thinking capitalize opportunity recent conversations common build new tech AI-enabled business defensible SaaS mitigate risk bad trip initial learning alongside portfolio broader community eager reactions rejections additions perspectives excitement borderline hysteria caused lose sight fundamental principle company building something people desperately need startup market looks hoard hammers searching nail start looking neglected find lived experience Assembled Regal started former operators tried failed solve pain points off-the-shelf solutions Eric Yuan Zoom running WebEx's Engineering team deeply understanding incumbent's weaknesses outside via deep customer research Regardless drive unique insight deluged winners provide solution desperate read achieving product-market fit needs addressable without GenAI strong justification let fancy tail wag dog Problem first second leaders established companies existing product roadmap ideally derived similar point hunting approach Identify accelerated API format products GPT-4 means incumbents integrate relatively easily see Doximity's DocsGPT helping doctors draft letters insurers Ironclad's Assist coaching redline contracts businesses likely time Given relative ease integration think defensibility Certainly always true speed execution matters However underlying foundational models differentiating tough least near term know cool app top OpenAI vs enduring asking simple question job done JTBD focusing within Anthropic etc tackling complex jobs require meaningful software beyond element good old fashioned blocks workflows data integrations advanced permissioning prove scaffolding around built Ironclad example provides complete digital contracting allowing negotiate sign binding tremendous amount code required pull including native text editor robust tons adding help redlining making easier remain countless parts must addressed high-quality workflow Proprietary play role though valuable forms come outcomes key processes e.g quickly contract close core fine-tuning domain-specific use cases contexts mostly right enough wrong clause added lots money stake AI's tendency hallucinate best deployment co-pilot coaching-style machine human pairing individual contributor IC manager relationship thanks James Cham analogy produce ultimately responsible outcome work coach thought worker learn accelerate accountability application developing UX's encourage contribute actively rather passively ensure accuracy creativity mutations improve model everyone else operating network Networks diving deeper future newsletters page meantime love hear space Please reach gordon emcap.com jake

Generative vs. Genuine: Why Today’s Generative AI Isn’t Tuned for B2B

January 31, 2023

2022 year personified AI moments last playing ChatGPT DALL-E saw future applications practical students getting help homework software engineers de-bugging code fun asking model new mac cheese recipe chatting dead celebrities strong potential consumer audiences marketing techniques evidenced fast growth companies Jasper believe generative alone impact business today genuine consistently accurate outcome-tied context-specific limitations likely blockers B2C B2B higher hurdle clear said play critical role solving problems Paired models human influence technology create step change ability world works broad generic value risks foundational tool drafting whether words images reciprocal helping condense summarize save time generally improve communication personal work life rely heavily run risk sounding often wrong Since 2016 called crutch Coaching Networks thesis based premise humans mutation engine age test abilities willingness expand baseline rest recommendations context dive particular 1 LLMs OpenAI's GPT-3 trained sound writers write compelling well-worded essay history Napoleonic Wars ensure facts dataset specific point released unable incorporate current data e.g content 2021 included Thus war Ukraine yield poor results mimicking trillions consumed match distribution text entire internet everything trusted generated December piece Mashable titled chatbot OpenAI amazing creative totally detailed instances either answers basic-knowledge questions used imagination make completely examples low-stakes understand riskier use settings Atlantic argued toy mostly tuned copywriting cold emails advertising purposes cases occasional factual inaccuracies typically tolerable accuracy order businesses feel confident using assistance required 2 outcome-oriented output-oriented consumers spit taglines beverage brand tell performs better interaction one-way street lacks continuously learn outcomes comes need generator iterative driven industry Promising apps anchor ROI-based example portfolio company Ironclad draft edit contracts efficiently helps lawyers move quickly platform built coach drafters clause formulations drive faster deal close rates marrying LLM suggestions proprietary building defensible outcome-focused product 3 deep needle needs company-specific off-the-shelf language publicly available lack access IP needed effective without created within Ironclad's workflow ascertain contract fastest anticipate volume truly incorporated near ever enterprise organizations unwilling share valuable competitive advantage lies wary direct indirect Similar concerns originated year's class action lawsuit Microsoft Github regarding CoPilot Corporations large enterprises contributing types broadly distributed fear losing Instead find another approach marriage external internally developed sets start querying generate initial output fed fine-tuned refine case Small pardon leverage knowledge graph build generating chaining together open-source SSLMs end-to-end stack able deliver ROI Next-gen look leveraging coupling people intellectual property customers Every function transformed next years job extend protect core heart integrating new-found power think means unique attributes app powering solely provide portion lie combination intervention ingenuity leave general architecture diagram thinking want Stay details meantime love hear space Please reach gordon emcap.com jake

Want To Hit Your First $1M? Triple Down on Marketing

December 13, 2022

early-stage founder maniacally focused pipeline meeting revenue goals first 1M naturally think need hire salespeople since ones follow inbound leads However something neglecting going generate ultimately feed sales team press release recent funding bit buzz product-led growth motion translate thousands early users without marketing leader place likely run sellers faster piranha eats prey leaders responsible far building websites running campaigns customer correspondents experts conversion analysis driving force behind company's image reputation short customers door keep invest day setting company success line piece cover detail sooner problem sales-first strategy Outside product engineering positions manager CSM must able wear hats retainer trained art keeping paying happy engaged turn keeps churning team's best friend know features resonating decided buy feedback crucial comes understanding objections prospects buying process feel counterintuitive strong default assumption among startups equals Founders trip reps employees attempt increase base companies experience highest churn rates lowest satisfaction scores shortest lifetime value words risky proposition common fallacies Every struggles VCs expect close milestone time room spare hitting ARR puts immense pressure young flawed though feeds ton acknowledge takes quality rep plow space month less backups upon looking beyond thinking ways web traffic brand awareness engagement let grow stagnant due false sense security freemium Never assume product's status enough win conversions reality freemiums average 3 rate small nearly impossible hit speaks makes statement main competitor competitors hundreds statements competing attention online single compete publicity either forming buyer personas understand motivates target market craft message emotional level account execs unfortunately pipe dream AEs want hired business head generating nurturing step right deal realities covered basic hypothesis talk figures moment use simple multiplication figure example 10 10M Others go simpler multiply three five times 3M minimum 5M 15 25M forth necessarily added friction shooting 10M+ targets rough-terrain marathon next uphill sprint always apply implications happens certainly met burned money runway shot extreme cases lose investor trust along ability make hires spend efforts now six months panic dry beast perfect great stick startup entire journey different expertise scale hiring settle someone work 12 18 suggest fixate last full four years Focus tier 24 complex difficult domain master comprises functions find days important clearly identify overview function focuses delivers messaging positioning competitive intel pricing packaging assets etc Demand Generation cost per click lead metrics closed won deals difference Gen Getting people store whereas Digital paid search SEO AR PR Analyst Relations Public Forrester Gartner Content materials social Operations tools data workflow post-sale events groups reference selling breakdown seen recommend look stage believe skills change expand 0 superpower narratives interest tempting take minute come clear easily understood mission purpose addition regards evaluate models compare better competition start train stops waits investors questions making reach doug emcap.com

How to Plan for FY23 Amid Economic Uncertainty

December 12, 2022

new year approaches founders executive teams throes planning FY23 economic uncertainty looming wondering best approach next help spoke seasoned finance executives advice crafting adjusting annual plan today's market Tim Cabral former CFO Veeva Systems leader cloud-based software global life sciences industry Tim's leadership financial discipline helped propel company IPO 2013 expertise 30 years continues add value today board member Kelly Steckelberg Zoom modern enterprise video communications Cisco WebEx 2008 lead crisis later CEO Zoosk joining 2017 past five led instrumental bringing profitability Collectively CFOs bring decades experience leading process learned textbooks-worth knowledge wisdom along Read top-line 1.Your nimble based current visibility good times bad believes biggest mistakes make getting weeds go deep chart accounts look across statements line metrics take management team long arduous unlikely 50 right Focus spending session discussing options outcomes instead reading results Startups growth companies advise leaders aim create discussion rather issue hardened guidance knowing return adjust improves CEOs time period says Start half graduate better giving illusion advises framing everything around break spend quarter allocate accordingly example allocating headcount departments release targets quarterly basis depending level Lastly takes let organizations track close chase perfect align strategy capital measure performance business books 2 uncertain appropriately aggressive perennial debate think aggressiveness set know hit exceed stretch thinks pros cons either tend lean conservative leaned towards call balance Though sound counterintuitive reverse true ensure need huge cuts first sign headwinds 3 scenario-based complacent tool project potential economy goes agrees creating range likely productive single number give latitude drivers push elevate going sure lazy plus minus 10 baseline completely different fact pattern drives high low upside downside equal recommends attached changes end day needs articulate want nothing appropriate scenario creates interesting conversation important parts 4 company's finance's Whether put together COO CRO VP person least 50-75 objectives rough paper stakeholders individually reactions feedback broader staff group Since executed collective goal everyone accountable plans numbers Whoever responsible executing revenue bought-in beginning 5 blanket live without agree critical over-committed question cut back often trickiest exactly alike sense allocations find low-hanging fruit suggests asking target travel dependent upon face-to-face relationships customers demo product done virtually in-person customer events figure reduce workforce entire signals every department equally contributing leverage making functions driving discretionary areas model behavior change adopt continue act days wrong meet higher bar else action needed try material