The Go-to-Market Engineer: Building Across Marketing, HubSpot, and MCPs

Why marketing hired an engineer instead of teaching AI to marketers

Gauntlet's marketing team debated two paths for adding AI capability: hire an AI engineer and teach them marketing, or hire a marketer and teach them to build. The team chose the first path, bringing on Arial Gardner, a Gauntlet Cohort 3 graduate, as an embedded engineer rather than upskilling an existing marketer.

The tradeoff runs both directions. The engineer has to climb a steep learning curve on marketing tools, metrics, and workflows, while the marketing team has to learn to hand off clear requirements instead of researching a SaaS tool and hoping it fits. The team's advice for anyone building this kind of role: stay anchored to the business problem you're solving, not the interesting technology you could use to solve it.

How an embedded engineer works inside a small team

The marketing team runs two stand-ups a week to review priorities and hand off work, supplemented by ad hoc Slack messages and lightweight tracking rather than a formal ticketing pipeline. Because there is a single embedded engineer, work can be triaged directly instead of routed through a separate engineering organization.

Sitting in on strategy conversations gives the engineer context a traditional intake-ticket process would miss, including which workflows already exist and why. That context lets the engineer proactively suggest improvements — like connecting a new API to save a step — rather than only executing tasks as they're assigned.

Bullseye: turning hiring signals into sales-ready leads

Bullseye is a lead-generation pipeline built to support Gauntlet's account executives. A managed agent scrapes hiring signals from job boards such as LinkedIn, Indeed, and Built In Austin, then scores each company against Gauntlet's ideal customer profile and stores the results in a database.

Account executives work from a spreadsheet-style interface, filterable by owner, that surfaces company details, hiring signals, and relevant news. An enrich button calls Clay's API to pull qualifying contacts, LinkedIn profiles, and company summaries, and a second button pushes selected contacts into existing HubSpot sequences. Both buttons are deliberate human-in-the-loop checkpoints: enrichment carries a real cost, so the team enriches only companies worth pursuing, and contacts move into HubSpot only after a person decides they're worth the outreach — keeping the CRM clean.

Build vs. buy: why the blog and CMS were built in-house

HubSpot, Gauntlet's CRM, includes its own content management system, but its constraints didn't match the workflows the marketing team wanted to keep. Rather than force those workflows into HubSpot's structure, the team built its own blog and CMS while keeping HubSpot as the system of record for sales.

The decision isn't a blanket preference for building over buying. HubSpot itself is a clear buy — a large enterprise CRM already does that job well; the build call was reserved for the pieces where existing workflows and specifications needed to come first.

A self-serve blog and SEO tool

The team's blog had been hosted on a third-party platform that provided little SEO value and no brand authority. Moving it in-house served two goals at once: recovering lost SEO and AEO value, and removing the bottleneck of routing every blog post through a separate development team.

The new CMS lets anyone on the marketing team take a post from idea to live without engineering involvement, with built-in search-result previews and editable meta titles and descriptions. The next step is connecting keyword-research data directly into the tool so it can flag a weak keyword choice before a post ever goes live.

One conversation, many MCPs: orchestrating ad platforms

Google Ads and Meta Ads now expose Model Context Protocol servers, letting the team interact with ad platforms without opening them directly. A typical workflow starts with a deep-research report on a target URL, which returns a full keyword and positioning strategy that can be handed straight to an AI assistant connected to the MCP to execute.

Feeding every MCP into a conversation at once risks overloading it, so the team sequences prompts to match how a buyer actually moves through a funnel — requesting top-of-funnel data first and introducing other MCPs as the conversation moves further down.

Where AI still needs a human in the loop

Content production blends tools rather than relying on one. A first pass built in Claude might hand off to ChatGPT or Canva for visual design, but the team is clear that human review remains part of every step. The concern isn't speed so much as quality: unattended AI output risks becoming generic and failing to resonate with the audience. A fully self-improving content pipeline isn't something they've built, or are rushing to build.

New experiences beyond the marketing stack

Because Gauntlet works with hundreds of companies hiring AI engineers, the team is building a model that scores a job description for how likely it is to appeal to AI engineers, with plans to do the same for resumes. It's a purpose-built assessment that would otherwise require a third-party vendor, and it shows how an embedded engineer can extend past traditional marketing tooling into new lead-generation experiences.

Measuring what's working

Because the team is small, direct feedback from account executives is still the fastest signal for what to change. Bullseye tracks button-level usage, such as who enriches which companies, while HubSpot reports on contacts created and sequence enrollment; more formal usability and lead-scoring metrics are planned as lead volume grows.