Gumloop vs WRITER

AI Text Box Loses to Simple Sign-Up Buttons in A/B Test

A newsletter article describes an A/B test by forms.app, a form builder with 4 million users. The test, shown to 25,000 new visitors, compared three homepage variants: an AI text box for describing a form, sign-up buttons (Google, Apple, email), and a single 'Create a form for free' button. The sign-up buttons converted at 15.15%, the 'Create a form' button at 9.60%, and the AI text box at only 5.74%. The AI box also performed worst on other metrics, with 84% of viewers clicking nothing. The author emphasizes that requiring users to think or type before value is a barrier, recommending moving such questions after signup, adding clickable example buttons, or asking for easily pasteable inputs like a URL. The article also mentions Microsoft Clarity as a tool for observing user behavior.

• forms.app A/B test on 25,000 new visitors showed sign-up buttons converted at 15.15%, 'Create a form' button at 9.60%, and AI text box at 5.74%.
• The AI text box had the lowest engagement with 84% of viewers clicking nothing.
• The 'Create a form' button attracted fewer sign-ups but more active users (3.63% vs 3.23%) and more buyers (0.176% vs 0.102%).

Couchsurfing Win-Back Email Uses Identity Marketing

The article highlights a win-back email from Couchsurfing, which cleverly uses identity-based marketing to re-engage churned users. Instead of focusing on the product, the email compliments the recipient's identity and uses nostalgia, with a button reading 'I'm Still That Person' that serves as an identity affirmation. The offer of one free month is placed at the bottom, making it feel like a bonus rather than a bribe. The author 'Tom' analyzes why this works, citing psychological studies on framing actions as identities, and provides actionable tips for marketers: open with identity-based compliments, make buttons into identity claims, and place offers at the end. He also mentions a similar approach by Atlas Privacy, which pre-created accounts for dead leads, yielding 8-12.5% conversion. The piece includes sponsor promotion for Gumloop, an AI agent platform used by teams at Shopify, Ramp, and Instacart.

• Couchsurfing sends a win-back email that compliments churned users' identity.
• The email button reads 'I'm Still That Person'.
• The free month offer is placed at the bottom of the email.

Anti-marketing: Spend Ads Budget on Real Experiences

The article describes the rising tactic of "anti-marketing": spending advertising budgets on non-ad experiences that generate organic attention and stories. Examples include Gumloop paying street musicians to perform with branded signs (a LinkedIn post from Gumloop’s CEO received 2,300+ likes) and Sticker Mule allocating 5% of its ad budget to giving strangers $100 on camera. The author also recounts raffling books and shipping ~45 books (~$2,000) as a marketing stunt. The piece frames anti-marketing as a shareable, attention-driving alternative to traditional ads and suggests small-scale executions can create compelling stories without fully abandoning paid media.

• Gumloop paid street musicians to perform with a Gumloop sign outside their office and across the city during a conference.
• A LinkedIn post about Gumloop’s stunt received 2,300+ likes and hundreds of comments.
• Sticker Mule allocates 5% of its ad budget to giving strangers $100 bills on camera; its CEO said, "It’s not philanthropy, it’s marketing."

Introducing Enterprise Brain: The shared GTM context layer, for an always-on customer journey that’s distinctly yours

Discover Enterprise Brain, WRITER’s shared context layer for brand, teams, and compliance. Learn how Agent Memory and WRITER Meet keep GTM work on-brand.

WRITER Releases Palmyra X6, Upgrades Enterprise AI Agent Platform

WRITER, an enterprise AI platform, launched Palmyra X6, its new flagship agentic AI model, alongside a rebuilt Agent harness and governance tools. The release targets marketing and revenue teams running multi-step workflows in production. The company reports that the Agent harness paired with Palmyra X6 operates at an average 52% lower cost, 48% faster speed, and 10% higher quality compared to prior performance. The upgrade includes cost-efficient pricing ($2/M input, $8/M output tokens), extended autonomy (tasks completed in 26 seconds, work for up to 8 hours unattended), and multi-model support via Azure, Amazon Bedrock, or NVIDIA NIM. WRITER also recently expanded its executive leadership with new CISO, CCO, and COO roles. These developments reflect the broader shift of agentic AI from pilot to production, with measurable ROI in marketing and customer experience.

• WRITER released Palmyra X6, its new flagship agentic AI model, on August 13, 2026.
• Palmyra X6 is post-trained on GLM 5.2 and scores 0.87/1.00 across nine enterprise evaluations.
• The Agent harness with Palmyra X6 achieves 52% lower cost, 48% faster speed, and 10% higher quality versus previous performance.

AI Productivity Boom Doesn't Reach Customers, Research Finds

The article examines a paradox in enterprise AI adoption: while individual productivity gains are real, customer-facing value and ROI remain flat. Research from WRITER shows 5x productivity gains but only 29% significant ROI, with 39% lacking revenue plans. Qualtrics finds AI-powered customer service fails at nearly 4x the rate of other AI uses, and TELUS Digital reports only 32% of enterprises use AI quality assurance tools. Accenture data reveals 86% plan to increase AI investment despite low impact. The article argues that companies succeed by redesigning processes around AI and measuring customer outcomes, not just internal speed. It highlights a growing distrust of AI in customer experience, with data misuse and blocked human access as key concerns, and calls for leadership accountability in proving customer value from AI investments.

• WRITER's 2026 survey (2,400 executives/employees) found individual AI users achieve ~5x productivity, yet only 29% of organizations see significant ROI.
• Qualtrics' 2026 research (20,000 consumers, 14 countries) shows AI-powered customer service fails at nearly 4x the rate of other AI applications.
• TELUS Digital/Ryan Strategic Advisory found only 32% of enterprises use AI-powered quality assurance and coaching tools.