
Drew Bredvick has written about building GTM Engineering at Vercel, applying AI to sales workflows, and his earlier Bootstrap 1000 challenge. This profile collects his practical lessons on agent design, customer observation, bootstrapping, and shipping software. — GTM Engineering — Why Now?.
Part 1: On Building Agentic Systems
- On Human-in-the-Loop AI: For consequential outbound messages, Drew recommends an agent prepare a recommendation while a person makes the go/no-go or edit decision. — How I Learn from Top Performers.
- On Getting Started: Before coding Vercel's lead agent, Drew spent a week observing a top SDR's qualification work; he then built an initial prototype over a weekend. — Building Vercel's First GTM Agent.
- On AI Adoption Timelines: In Tenex's interview-based account, Drew describes a roughly one-week build followed by several weeks gaining confidence in the agent before production use. The rollout took longer than the prototype. — Tenex Interview.
- On Shadowing the Killers: Drew advises GTM engineers to watch top performers, identify the work that differentiates them, and use that observation to decide what to automate or extend. — GTM Engineering — Why Now?.
- On the Two Waves of AI: Drew distinguishes AI added to a product from AI used to accelerate the process of building products; he argues the second wave changes development economics. — Two Waves of AI.
- On Agent ROI: Drew reports that Vercel's lead agent, prototyped in a weekend, produced more than $2 million in annual savings after deployment. That is a savings claim, not verified incremental revenue. — Building Vercel's First GTM Agent.
- On Scaling Sales Teams: Drew's January 2026 account says the lead agent reduced manual inbound qualification and left an SDR reviewing researched, drafted responses before sending. His staffing figure is 10 to one in that account; another interview-derived account differs, so this lesson does not assert a universal headcount number. — Building Vercel's First GTM Agent.
- On Context Retrieval: The lead agent conducts company research before qualifying a submission, and exposes that research to the human reviewer so the recommendation can be checked. — Building Vercel's First GTM Agent.
- On Workflow Automation: Drew defines GTM engineering as finding successful patterns in revenue work and turning them into repeatable automation. — How to Break Into GTM Engineering.
- On Using LLMs as Cron Jobs: Drew built a scheduled daily-recap workflow that boots Claude Code in a sandbox with MCP access to his notes and Slack, then appends a structured summary. It is a specific implementation example, not a universal reliability guarantee. — Claude Code as a Cron Job.
Part 2: On Go-To-Market (GTM) Engineering
- On a New Discipline: Drew describes GTM engineering as a new role where shipping an AI automation in production can demonstrate more than pedigree. — How to Break Into GTM Engineering.
- On Defining GTM Engineering: GTM engineering connects technical building with sales operations: identify what works in a revenue workflow, then make it repeatable through software and agents. — How to Break Into GTM Engineering.
- On Building Excellence Agents: Drew distinguishes excellence agents from efficiency agents: excellence agents help more of a team execute practices observed in its top performers. — How to Break Into GTM Engineering.
- On the Lead Routing Bottleneck: Vercel's lead agent enriches a form submission, researches the prospect, qualifies it, and routes a prepared recommendation to an SDR through Slack for human review. — Building Vercel's First GTM Agent.
- On Picking Problems: Drew prioritizes internal tools where someone already works around a frequent, painful process and owns the outcome; that behavior indicates a real problem. — Stated vs Revealed Preference.
- On Working With Sales: Drew moved from software engineering into sales engineering to learn directly from customer conversations, product objections, and willingness to pay. — Learning Sales as an Engineer.
- On Using Modern Stacks: The lead-agent case used Next.js, the AI SDK, Workflow DevKit, a Slack adapter, and web research to connect a real GTM workflow; the value was the workflow, not a generic 'unfair advantage' from the stack. — Building Vercel's First GTM Agent.
Part 3: On The Intersection of Engineering and Sales
- On Selling to Developers: Drew recommends helping developer buyers evaluate build-versus-buy tradeoffs and showing them how the software works. The technical win is distinct from budget and organizational approval. — Selling to Developers.
- On Demos vs. Memos: When seeking feedback on registered.dev, Drew considered a small working demo more useful than an abstract description because prospects could respond to something concrete. — What's Stopping You? Interview.
- On Empathy for the User: As a sales engineer, Drew learned from direct customer calls which technical blockers mattered and used those conversations to advocate for product changes. — Learning Sales as an Engineer.
- On Value over Tech: Drew frames ethical sales as helping customers solve a real problem and judging a product by the value they receive, rather than by technical features alone. — Learning Sales as an Engineer.
- On Validating Early: Drew acknowledged that waiting for a more polished registered.dev before contacting prospects partly reflected fear of feedback. He left the interview intending to speak with additional potential customers sooner. — What's Stopping You? Interview.
- On Asking for Money: For the registered.dev concept, Drew wanted to ask prospective buyers whether they would commit to purchasing it; the test was whether the problem mattered enough to pay for. — What's Stopping You? Interview.
- On Sales Integrity: Drew explicitly distinguishes helping customers find a useful product from predatory selling; for him, sales should test fit and create real customer value. — Learning Sales as an Engineer.
- On Building for the Buyer: A technical evaluation has to account for the business case as well as engineering merits. Drew advises showing developers how to use the product and supplying evaluation criteria. — Selling to Developers.
Part 4: On Bootstrapping and Indie Hacking
- On the Bootstrap 1000 Challenge: Drew designed Bootstrap 1000 as a challenge to replace his full-time salary with bootstrapped products through 1,000 days of effort, using a sustainable shipping system along the way. — Bootstrap 1000 Challenge.
- On the Long Game: After reviewing earlier side projects, Drew concluded that bootstrapping would take longer than he expected and built a long-duration challenge around consistency rather than overnight success. — Bootstrap 1000 Challenge.
- On the Stair-Step Method: Drew explicitly credits Rob Walling's stair-step approach: start with smaller projects to gain business experience before attempting more complex SaaS. — Bootstrap 1000 Challenge.
- On Prioritizing Excitement: Drew says earlier projects lost momentum when he was not interested in them, so he made genuine interest an explicit filter for future side projects. — Bootstrap 1000 Challenge.
- On Early Projects: Drew says his earlier side projects collectively earned under $1,000, yet helped him build engineering skills, understand business, and learn to ship complete products. — Bootstrap 1000 Challenge.
- On Repetitions: Bootstrap 1000 counts small, measurable work toward shipping; even a tiny code change can count as a day of effort. The process is designed to build consistency, not guarantee revenue from each project. — Bootstrap 1000 Challenge.
- On Finding Customers First: For a profit-seeking SaaS side project, Drew recommends finding potential customers and testing the market before investing heavily in code. He distinguishes this from projects built mainly to learn. — Rules for Profitable Side Projects.
- On Dedicated Focus: Drew took a week off to work on Fit Vitals. The week produced real infrastructure and product progress, but his retrospective says the product was still far from finished. — Rome Wasn't Built in a Day.
- On Building Fit Vitals: Fit Vitals taught Drew that monitoring Core Web Vitals involved more real-time data and scaling complexity than he first expected; he had built a Kinesis pipeline but had not completed the SaaS. — Rome Wasn't Built in a Day.
- On Building in Public: Drew documented Fit Vitals on Indie Hackers and found feedback from other builders motivating while he worked through the product's scope. — Rome Wasn't Built in a Day.
Part 5: On Habits and the Systems Over Goals Philosophy
- On Systems vs. Goals: Drew found his Bootstrap 1000 process more motivating when he counted each application of a shipping system as a win instead of waiting for one distant outcome. — On Habits.
- On the Two-Day Rule: Drew adopted a rule he credits to Matt D'Avella: do not miss two days in a row. For Bootstrap 1000 it allowed consistency without requiring code every single day. — On Habits.
- On Lowering Activation Energy: Drew made his coding habit deliberately small: sit down, write code, and aim for one commit rather than setting an intimidating daily production target. — On Habits.
- On Daily Coding: In his early Bootstrap 1000 update, Drew reported writing and pushing code on each of the first 25 days he sat down to do it; his stated daily target was one commit. — On Habits.
- On Compound Interest: Drew describes consistent blogging as a way to improve writing and clarity over time; the compounding claim is about communication practice, not every kind of daily input. — My Principles for Blogging.
- On Sustainable Growth: Drew's account of earlier side projects says a good idea is not enough if he loses interest before following through; he therefore chooses problems that sustain his motivation. — Making Decisions While Bootstrapping.
Part 6: On Learning in Public and Career Growth
- On Increasing Clarity: Drew says he blogs to clarify his thinking, improve his writing, and show what he is working on. — My Principles for Blogging.
- On Consistency in Writing: Drew advises choosing a familiar platform and toolchain so publishing remains easy enough to sustain over time. — My Principles for Blogging.
- On Joining Vercel: Drew says his prior use of and writing about Next.js helped him join Vercel in 2021; he was already a believer in the platform. The 'rocket ship' quote is not verified. — ▲Vercel.
- On Open Source Contributions: Drew published the Lead Agent template with form handling, research, qualification, Slack review, and drafting so others can adapt the architecture. — Building Vercel's First GTM Agent.
- On Networking Through Writing: Drew says writing about Next.js attracted people interested in the same technology and contributed to his Vercel job opportunity. — ▲Vercel.
Part 7: On Shipping, Prototyping, and Fast Feedback
- On Bias to Action: Drew acted on a niche internal-tool idea during a short break and shipped a rough working version in 34 minutes using Vercel tools. — Idea to Prod in 34 Minutes.
- On Fast Iteration: After observing users, Drew recommends a quick agent prototype, real outputs, and a loop of shadowing, shipping, and revising rather than a complete system upfront. — How I Learn from Top Performers.
- On Scoping MVPs: Drew's Fit Vitals retrospective says he tried to start with a complex SaaS and learned to reduce scope or be patient with larger work. — Rome Wasn't Built in a Day.
- On Tooling for Speed: In a 34-minute prototype, Drew used the AI SDK, v0, Cursor, and Vercel to remove setup and design friction and get a niche tool online. — Idea to Prod in 34 Minutes.
- On Feedback Loops: Drew asked SDRs to flag weak Lead Agent recommendations, then observed those users to understand the failure before adjusting the workflow. — How I Learn from Top Performers.
- On the Craft of Shipping: Drew says early side projects taught him to ship complete products, even when they did not produce meaningful revenue. — Bootstrap 1000 Challenge.
- On Perfectionism: Drew credits faster shipping partly to recognizing when a marginal CSS issue is not worth delaying a finished feature. — Effective.
- On Getting to Revenue: For a profit-oriented SaaS project, Drew recommends finding customers and testing demand before committing to a large build; he planned presales after a demo. — Rules for Profitable Side Projects.
- On Killing Projects: In his side-project retrospective, Drew identified lost interest, inconsistency, and weak domain knowledge as recurring reasons projects failed, then used those lessons to choose what to pursue next. — Bootstrap 1000 Challenge.
Part 8: On Observing Real User Behavior
- On Stated Preference: Drew uses stated preference to describe what people say they want in interviews and wish lists; it is useful for generating ideas but not proof of urgency. — Stated vs Revealed Preference.
- On Revealed Preference: Drew looks for behavior that reveals an existing need—such as someone maintaining a painful manual workaround—before prioritizing an internal tool. — Stated vs Revealed Preference.
- On Shadowing Users: Drew recommends watching high performers complete real tasks and asking why they make each choice, rather than relying only on their recollection of the workflow. — How I Learn from Top Performers.
- On Building the Right Agent: Drew argues that observing which signals people actually use helps identify the context an agent needs; a wish list alone can add noise. — How I Learn from Top Performers.
- On Analytics: For Vercel's Lead Agent, Drew compared historical contact submissions with actual human decisions before tuning the agent, rather than treating interview answers as the only signal. — How I Learn from Top Performers.
- On Friction Logs: The team let SDRs flag negative Lead Agent feedback, then shadowed those people and uncovered workflow friction that had not appeared in the initial requirements. — How I Learn from Top Performers.
- On the Reality of Workflows: Drew treats repeated manual work—such as copying data among several systems—as evidence that a problem is real and may merit automation. — Stated vs Revealed Preference.
- On Empathy via Observation: Drew's 'State It Back' test asks a builder to explain a user's workflow accurately enough that the user agrees before committing to an agent design. — How I Learn from Top Performers.