From AI Agents to AI Operations: A Strategic Pivot
When Mikel Mangold, founder of Goldup-e, sat down for this interview, he had just updated his company's positioning. Instead of promoting AI agents, Goldup-e now focuses on implementing AI for sales and marketing operations. It's a subtle but important shift that reflects a maturing market.
"Instead of promoting that we're doing AI agents, we are promoting that we are implementing AI for sales and marketing operations," Mangold explains. The distinction matters: the goal is not to replace human judgment but to augment it with better data, cleaner workflows, and smarter automation.
What Makes Outbound Different for Science-Driven Startups
For companies selling scientific or clinical products, the standard SaaS playbook often falls flat. Mangold points out that science-driven products can't be sold on urgency or discounts. Instead, the message must lead with evidence.
"Science-driven products can't be sold on urgency or discounts, the message has to lead with evidence, science facts, proof points etc.," he says. He cites a real campaign where a single clinical statistic — "58.97% vs. 5.13% complete healing at six weeks" — carried the entire outreach. Outbound becomes about earning credibility with the right specialist, not blasting volume at a generic ICP.
When to Bring in HubSpot Consulting
Many B2B companies debate whether to build CRM capabilities in-house or hire a consultant. Mangold's advice is clear: bring in consulting when you need architecture decisions done right the first time.
"Bring in consulting when you need architecture decisions done right the first time, data model, lifecycle stages, list logic, integrations, and build in-house once the system is stable and the daily work is execution rather than design," he advises. He warns that inexperienced CRM builders often waste years of effort because they don't know who is who and where they are lying in the funnel. Getting it right on day one can be tremendously helpful for long-term business health, especially during valuation events like fundraising or acquisition.
What AI Implementation Looks Like in Practice
Mangold demystifies AI for sales and marketing operations. It's not about flashy chatbots or autonomous agents. It's about concrete, unglamorous work: cleaning and classifying account lists, enrichment and personalization at scale, and automated reporting.
"In practice it's unglamorous and concrete: AI agents that clean and classify account lists (practice type, relevance, exclusions), enrichment and personalization at scale, and automated reporting," he says. He gives an example: instead of just importing first name, last name, and title, companies can now import ten additional data sets — country, specialties, seniority, recent LinkedIn activity — directly into their CRM. This data becomes powerful for both sales and marketing teams.
Measuring Pipeline Health vs. Busywork
A common trap in outbound sales is mistaking activity for progress. Mangold suggests ignoring activity totals and instead looking at reply quality per segment.
"Ignore activity totals and look at reply quality per segment, a healthy pipeline produces positive replies from the exact titles you targeted, with follow-up touches tracked, while a busy one produces impressive send counts and silence," he explains. Goldup-e measures success by open rate and reply rate. Mangold claims they can achieve 60% to 80% open rates compared to the traditional 20%.
The Common Mistake with CRM and AI
When companies try to combine CRM data with AI tools, Mangold sees one recurring error: using AI without proper segmentation.
"People just randomly use AI, but it's not well segmented in the CRM. Segmenting it well and connecting all the data as objects in the CRM/HubSpot can be powerful for the analytics, and that's where we help," he says.
For more on building a healthy pipeline, check out HubSpot's guide to pipeline management or this overview of AI in sales operations.
Practical Takeaways
- Lead with evidence, not urgency, when selling to science-driven buyers.
- Invest in CRM architecture early — it pays off during fundraising or acquisition.
- AI implementation is about data hygiene and enrichment, not magic.
- Measure pipeline health by reply quality, not send volume.
- Segment your CRM data before applying AI tools.
Mangold's approach is grounded in the reality that science-driven companies need a different sales playbook — one built on credibility, clean data, and smart automation.