Case studies
The engagements below show what Harbinger builds when the goal is to free a team's attention, not replace it.
Each one freed capacity that the team redirected into higher-leverage work; relationship-building, better negotiations, sharper decisions.
Implement AI wrong and the damage compounds; the team's critical thinking erodes, and the customer experience follows.
Our success stories
Case studies
The engagements below show Harbinger's operational approach across early-stage companies, AI research labs, and live events operations. Each one required the same pattern: mapping the operation end to end, designing systems around the people who would use them, and handing off ownership cleanly when the work was complete.
Harbinger has since pivoted and is now focused primarily on applying that same pattern inside HNW/UHNW household operations and portfolio companies. The case studies below reflect work to date; the page will grow as engagements in the new focus area are published.
mULTIoN
AI Research Lab
4-month engagement
MultiOn was the first autonomous AI agent capable of browsing the web and taking actions on a user's behalf. Julian Brooks, founder of Harbinger, went from community member in the MultiOn Discord to Head of Quality Assurance and Head of Customer Support within 90 days.
In those roles, he helped develop the first evaluation processes purpose-built for AI agents, now used by OpenAI, Anthropic, and other major AI labs. He also contributed to early enterprise sales, community management, conference representation, and operational automation while the company was raising capital and scaling from 10 to 25 people.
90% of the customer support department automated with freed capacity redirected into user interviews
200%+ lift in customer satisfaction (93% avg)
10,000 beta testers recruited from community and provided a custom-built AI chatbot for reporting issues.
Assisted in the creation of REAL Bench (an open-source web agent benchmark now used by OpenAI, Anthropic, and others)
Our success stories
Case studies
The engagements below show Harbinger's operational approach across early-stage companies, AI research labs, and live events operations. Each one required the same pattern: mapping the operation end to end, designing systems around the people who would use them, and handing off ownership cleanly when the work was complete.
Harbinger has since pivoted and is now focused primarily on applying that same pattern inside HNW/UHNW household operations and portfolio companies. The case studies below reflect work to date; the page will grow as engagements in the new focus area are published.
mULTIoN
AI Research Lab
4-month engagement
MultiOn was the first autonomous AI agent capable of browsing the web and taking actions on a user's behalf. Julian Brooks, founder of Harbinger, went from community member in the MultiOn Discord to Head of Quality Assurance and Head of Customer Support within 90 days.
In those roles, he helped develop the first evaluation processes purpose-built for AI agents, now used by OpenAI, Anthropic, and other major AI labs. He also contributed to early enterprise sales, community management, conference representation, and operational automation while the company was raising capital and scaling from 10 to 25 people.
90% of the customer support department automated with freed capacity redirected into user interviews
200%+ lift in customer satisfaction (93% avg)
10,000 beta testers recruited from community and provided a custom-built AI chatbot for reporting issues.
Assisted in the creation of REAL Bench (an open-source web agent benchmark now used by OpenAI, Anthropic, and others)
Our success stories
Case studies
The engagements below show Harbinger's operational approach across early-stage companies, AI research labs, and live events operations. Each one required the same pattern: mapping the operation end to end, designing systems around the people who would use them, and handing off ownership cleanly when the work was complete.
Harbinger has since pivoted and is now focused primarily on applying that same pattern inside HNW/UHNW household operations and portfolio companies. The case studies below reflect work to date; the page will grow as engagements in the new focus area are published.
mULTIoN
AI Research Lab
4-month engagement
MultiOn was the first autonomous AI agent capable of browsing the web and taking actions on a user's behalf. Julian Brooks, founder of Harbinger, went from community member in the MultiOn Discord to Head of Quality Assurance and Head of Customer Support within 90 days.
In those roles, he helped develop the first evaluation processes purpose-built for AI agents, now used by OpenAI, Anthropic, and other major AI labs. He also contributed to early enterprise sales, community management, conference representation, and operational automation while the company was raising capital and scaling from 10 to 25 people.
90% of the customer support department automated with freed capacity redirected into user interviews
200%+ lift in customer satisfaction (93% avg)
10,000 beta testers recruited from community and provided a custom-built AI chatbot for reporting issues.
Assisted in the creation of REAL Bench (an open-source web agent benchmark now used by OpenAI, Anthropic, and others)
agi iNC
AI Research Lab
4-month engagement
AGI Inc., a MultiOn spin-off building autonomous AI for mobile devices, came to Harbinger facing an investment round quickly approaching and a pending partnership with Visa - without a functional product.
Over the scoped four-month engagement, Harbinger built AGI Inc.'s evaluations department from scratch, assisted in development of their internal dashboards and tracking, identified critical bugs across agent and application layers, assisted in finalizing the REAL Bench (an open-source web agent benchmark now used by OpenAI, Anthropic, and others), and triaged issues ahead of a successful investment close.
AGI Inc. has since secured multiple funding rounds and partnerships with Visa and Mastercard for agentic payments.
150+ bugs identified and documented
30% improvement in agent success rate
Visa Mastercard partnership secured
REAL Bench scores published by major AI labs
agi iNC
AI Research Lab
4-month engagement
AGI Inc., a MultiOn spin-off building autonomous AI for mobile devices, came to Harbinger facing an investment round quickly approaching and a pending partnership with Visa - without a functional product.
Over the scoped four-month engagement, Harbinger built AGI Inc.'s evaluations department from scratch, assisted in development of their internal dashboards and tracking, identified critical bugs across agent and application layers, assisted in finalizing the REAL Bench (an open-source web agent benchmark now used by OpenAI, Anthropic, and others), and triaged issues ahead of a successful investment close.
AGI Inc. has since secured multiple funding rounds and partnerships with Visa and Mastercard for agentic payments.
150+ bugs identified and documented
30% improvement in agent success rate
Visa Mastercard partnership secured
REAL Bench scores published by major AI labs
agi iNC
AI Research Lab
4-month engagement
AGI Inc., a MultiOn spin-off building autonomous AI for mobile devices, came to Harbinger facing an investment round quickly approaching and a pending partnership with Visa - without a functional product.
Over the scoped four-month engagement, Harbinger built AGI Inc.'s evaluations department from scratch, assisted in development of their internal dashboards and tracking, identified critical bugs across agent and application layers, assisted in finalizing the REAL Bench (an open-source web agent benchmark now used by OpenAI, Anthropic, and others), and triaged issues ahead of a successful investment close.
AGI Inc. has since secured multiple funding rounds and partnerships with Visa and Mastercard for agentic payments.
150+ bugs identified and documented
30% improvement in agent success rate
Visa Mastercard partnership secured
REAL Bench scores published by major AI labs
Schedule a confidential consultation
Schedule a confidential consultation
Schedule a confidential consultation
STEALTH STARTUP
Startup - Physical Products
Ongoing Engagement
An early-stage startup came to Harbinger with a physical product that deploys to brick-and-mortar locations.
They had a working product and funding, but no structured go-to-market approach and were struggling to get pilots out in the field. We helped reframe their target segment, built a lead tracking system, dashboard and interface, then ran a multi-channel outreach program across cold calling, email, and in-person visits.
When digital outreach plateaued, we shifted to field execution; visiting locations in person, collecting stakeholder info, dropping off one-pagers, and refining messaging based on what we heard.
3 pilot locations secured
8.86% total lead response rate
361 total leads generated
Custom Airtable CRM with interface and dashboard
STEALTH STARTUP
Startup - Physical Products
Ongoing Engagement
An early-stage startup came to Harbinger with a physical product that deploys to brick-and-mortar locations.
They had a working product and funding, but no structured go-to-market approach and were struggling to get pilots out in the field. We helped reframe their target segment, built a lead tracking system, dashboard and interface, then ran a multi-channel outreach program across cold calling, email, and in-person visits.
When digital outreach plateaued, we shifted to field execution; visiting locations in person, collecting stakeholder info, dropping off one-pagers, and refining messaging based on what we heard.
3 pilot locations secured
8.86% total lead response rate
361 total leads generated
Custom Airtable CRM with interface and dashboard
STEALTH STARTUP
Startup - Physical Products
Ongoing Engagement
An early-stage startup came to Harbinger with a physical product that deploys to brick-and-mortar locations.
They had a working product and funding, but no structured go-to-market approach and were struggling to get pilots out in the field. We helped reframe their target segment, built a lead tracking system, dashboard and interface, then ran a multi-channel outreach program across cold calling, email, and in-person visits.
When digital outreach plateaued, we shifted to field execution; visiting locations in person, collecting stakeholder info, dropping off one-pagers, and refining messaging based on what we heard.
3 pilot locations secured
8.86% total lead response rate
361 total leads generated
Custom Airtable CRM with interface and dashboard
Mid-Sized Live Event Operator
Event Production
Commissioned Project
A mid-size California music festival came to Harbinger with a booking process that ran on scattered spreadsheets, email threads, and institutional knowledge.
When talent agents sent quotes, the team had no fast way to evaluate whether the number was reasonable — answering that question meant hours of manual research, and by the time the data was gathered, the negotiation window had often narrowed.
We embedded into their operation and built a centralized booking system: an Airtable database integrated with live data feeds for social metrics, streaming metrics, tour schedules, and more.
The team now pulls a complete artist profile in minutes and comes back to agents with specific, data-informed counters instead of gut-based pushback.
$50K saved in negotiation cost reductions prior to the largest artists being booked
Manual data collection time reduced from 6–8 hours per artist to 3 minutes.
Negotiation cycles shortened from 3–4 rounds to 2 or fewer
Centralized database replaced multiple scattered data sources
Mid-Sized Live Event Operator
Event Production
Commissioned Project
A mid-size California music festival came to Harbinger with a booking process that ran on scattered spreadsheets, email threads, and institutional knowledge.
When talent agents sent quotes, the team had no fast way to evaluate whether the number was reasonable — answering that question meant hours of manual research, and by the time the data was gathered, the negotiation window had often narrowed.
We embedded into their operation and built a centralized booking system: an Airtable database integrated with live data feeds for social metrics, streaming metrics, tour schedules, and more.
The team now pulls a complete artist profile in minutes and comes back to agents with specific, data-informed counters instead of gut-based pushback.
$50K saved in negotiation cost reductions prior to the largest artists being booked
Manual data collection time reduced from 6–8 hours per artist to 3 minutes.
Negotiation cycles shortened from 3–4 rounds to 2 or fewer
Centralized database replaced multiple scattered data sources
Mid-Sized Live Event Operator
Event Production
Commissioned Project
A mid-size California music festival came to Harbinger with a booking process that ran on scattered spreadsheets, email threads, and institutional knowledge.
When talent agents sent quotes, the team had no fast way to evaluate whether the number was reasonable — answering that question meant hours of manual research, and by the time the data was gathered, the negotiation window had often narrowed.
We embedded into their operation and built a centralized booking system: an Airtable database integrated with live data feeds for social metrics, streaming metrics, tour schedules, and more.
The team now pulls a complete artist profile in minutes and comes back to agents with specific, data-informed counters instead of gut-based pushback.
$50K saved in negotiation cost reductions prior to the largest artists being booked
Manual data collection time reduced from 6–8 hours per artist to 3 minutes.
Negotiation cycles shortened from 3–4 rounds to 2 or fewer
Centralized database replaced multiple scattered data sources
Boutique Live Events Operator
Event Production
Ongoing Engagement
Bear Music Festival, an all-inclusive event, needed to scale revenue and sponsorships without losing their community-focused identity and core principle of never 'selling' to their attendees. Harbinger embedded as an operational partner, building AI-powered systems across sponsorship, social media, merchandise, and operations — plus trained the team to use the tools themselves and wrote documentation for all new systems.
~300% YoY increase in merch sales
27% YoY increase in ticket sales
~5x increase in Instagram engagement
13 sponsors secured 2 months prior to the event
Boutique Live Events Operator
Event Production
Ongoing Engagement
Bear Music Festival, an all-inclusive event, needed to scale revenue and sponsorships without losing their community-focused identity and core principle of never 'selling' to their attendees. Harbinger embedded as an operational partner, building AI-powered systems across sponsorship, social media, merchandise, and operations — plus trained the team to use the tools themselves and wrote documentation for all new systems.
~300% YoY increase in merch sales
27% YoY increase in ticket sales
~5x increase in Instagram engagement
13 sponsors secured 2 months prior to the event
Boutique Live Events Operator
Event Production
Ongoing Engagement
Bear Music Festival, an all-inclusive event, needed to scale revenue and sponsorships without losing their community-focused identity and core principle of never 'selling' to their attendees. Harbinger embedded as an operational partner, building AI-powered systems across sponsorship, social media, merchandise, and operations — plus trained the team to use the tools themselves and wrote documentation for all new systems.
~300% YoY increase in merch sales
27% YoY increase in ticket sales
~5x increase in Instagram engagement
13 sponsors secured 2 months prior to the event
FAQ
What prospective clients ask us most
We’re here for the follow-up questions too—find us on the Contact page.
How do you handle confidentiality?
With the same standard you'd expect from your wealth manager or attorney. We do not publish client names, engagement details, or operational information without explicit written consent. Proof points and case studies used in our materials are anonymized unless a client has specifically agreed otherwise.
What's your pricing model?
Engagements are value-priced based on the complexity and scope of the work, not billed hourly. Payments are tied to defined milestones and documented deliverables. We'll scope pricing during the consultation based on what your operation looks like and what the engagement would involve. We don't publish pricing because every operation is different and a number without context isn't useful to either side.
What does the first engagement typically look like?
It starts with a consultation to determine fit. If we move forward, the first phase is always an Operational Deep Dive; two weeks where we embed in your operation, map workflows, interview team members, and identify where AI and automation will have the most impact. That produces a findings report and prioritized roadmap. From there, we begin building systems based on what the deep dive revealed.
Do you work on-site or remotely?
Primarily work remote but for additional compensation and coverage of travel expenses, the founder Julian Brooks can travel anywhere for up to 2 week periods at a time.
Do I need technical knowledge to work with Harbinger?
No. Part of what Harbinger provides is translating between the technical and the operational. Your team doesn't need to understand how AI models work or what an API is. We handle the technical evaluation, the build, and the maintenance. What we need from your side is access to the people who know how the operation runs and the willingness to give us an honest picture of how things actually work day to day.
Who will we be working with?
Julian Brooks, the founder of Harbinger, will be your main point of contact and the embedded operator. If we do bring on another staff member, they will be strictly isolated to Harbinger's operations or will otherwise sign any relevant documentation.
FAQ
What prospective clients ask us most
We’re here for the follow-up questions too—find us on the Contact page.
How do you handle confidentiality?
With the same standard you'd expect from your wealth manager or attorney. We do not publish client names, engagement details, or operational information without explicit written consent. Proof points and case studies used in our materials are anonymized unless a client has specifically agreed otherwise.
What's your pricing model?
Engagements are value-priced based on the complexity and scope of the work, not billed hourly. Payments are tied to defined milestones and documented deliverables. We'll scope pricing during the consultation based on what your operation looks like and what the engagement would involve. We don't publish pricing because every operation is different and a number without context isn't useful to either side.
What does the first engagement typically look like?
It starts with a consultation to determine fit. If we move forward, the first phase is always an Operational Deep Dive; two weeks where we embed in your operation, map workflows, interview team members, and identify where AI and automation will have the most impact. That produces a findings report and prioritized roadmap. From there, we begin building systems based on what the deep dive revealed.
Do you work on-site or remotely?
Primarily work remote but for additional compensation and coverage of travel expenses, the founder Julian Brooks can travel anywhere for up to 2 week periods at a time.
Do I need technical knowledge to work with Harbinger?
No. Part of what Harbinger provides is translating between the technical and the operational. Your team doesn't need to understand how AI models work or what an API is. We handle the technical evaluation, the build, and the maintenance. What we need from your side is access to the people who know how the operation runs and the willingness to give us an honest picture of how things actually work day to day.
Who will we be working with?
Julian Brooks, the founder of Harbinger, will be your main point of contact and the embedded operator. If we do bring on another staff member, they will be strictly isolated to Harbinger's operations or will otherwise sign any relevant documentation.
FAQ
What prospective clients ask us most
We’re here for the follow-up questions too—find us on the Contact page.
How do you handle confidentiality?
With the same standard you'd expect from your wealth manager or attorney. We do not publish client names, engagement details, or operational information without explicit written consent. Proof points and case studies used in our materials are anonymized unless a client has specifically agreed otherwise.
What's your pricing model?
Engagements are value-priced based on the complexity and scope of the work, not billed hourly. Payments are tied to defined milestones and documented deliverables. We'll scope pricing during the consultation based on what your operation looks like and what the engagement would involve. We don't publish pricing because every operation is different and a number without context isn't useful to either side.
What does the first engagement typically look like?
It starts with a consultation to determine fit. If we move forward, the first phase is always an Operational Deep Dive; two weeks where we embed in your operation, map workflows, interview team members, and identify where AI and automation will have the most impact. That produces a findings report and prioritized roadmap. From there, we begin building systems based on what the deep dive revealed.
Do you work on-site or remotely?
Primarily work remote but for additional compensation and coverage of travel expenses, the founder Julian Brooks can travel anywhere for up to 2 week periods at a time.
Do I need technical knowledge to work with Harbinger?
No. Part of what Harbinger provides is translating between the technical and the operational. Your team doesn't need to understand how AI models work or what an API is. We handle the technical evaluation, the build, and the maintenance. What we need from your side is access to the people who know how the operation runs and the willingness to give us an honest picture of how things actually work day to day.
Who will we be working with?
Julian Brooks, the founder of Harbinger, will be your main point of contact and the embedded operator. If we do bring on another staff member, they will be strictly isolated to Harbinger's operations or will otherwise sign any relevant documentation.
Schedule a confidential consultation
Schedule a confidential consultation