Vertical AI Leads at Bank of America Trailblazers Event

General-purpose models are starting to look alike. The companies that pitched at Bank of America's Private Tech Trailblazers Conference 2026 in Palo Alto, California, are betting on what sits around those models: their own data, industry workflows and, more and more often, their own hardware.

The lineup covered restaurants, construction, hospitals, payments and defense. Each company was going after a narrow problem where results can be measured. John Furrier, executive analyst at theCUBE Research, conducted the interviews. He said specialized intelligence went from a quiet topic to a central one in about a year, because "the intelligence comes out of the data."

Nine companies stood out.

Robots, trucks and chips

1. Bear Robotics. The company has roughly 16,000 autonomous mobile robots deployed and a backlog of about 4,000, and co-founder Bren Pierce said revenue doubles every year. Its main market is restaurants in Japan and Korea, and it also serves care homes and casinos. A partnership with LG is taking it into warehouses and factories. Pierce sold his humanoid startup Kinisi back to Bear, and the humanoids run on the same software and cloud stack as the mobile robots. Foundation models that learn from a few hundred examples, onboard compute such as Nvidia's Jetson Thor and LLM-assisted coding have cut six-month projects down to days. Tactile hands are still the bottleneck, at about $30,000 each.

2. All3. Labor accounts for 55% to 60% of construction costs. All3, the operating name of Address Robotics Ltd., wants to shrink that share by owning the whole chain: software that takes a plot to permit-ready plans, industrial robots that make custom elements at mass-production cost, and a mobile robot called Mantis that assembles and finishes them on site. Because All3 designs the process itself, CEO Rodion Shishkov said it can remove most edge cases rather than train robots to handle each one. After a seed round of roughly $25 million to $30 million, its first site, an 11-sided plot for a six-story co-living building, is being prepared.

3. Harbinger Motors. Harbinger sells electric and hybrid medium-duty platforms at diesel prices. CEO John Harris said a typical California parcel truck saves about $30,000 a year on fuel after charging costs. Customers include FedEx and Thor Industries, and its battery system now powers Airstream trailers. A partnership with American Rheinmetall targets ground autonomy for defense.

4. Unconventional AI. Founded a year ago by Naveen Rao, the company aims to cut AI power use about 1,000-fold by designing chips and software together. It has raised about $540 million, grown from 14 staff in January to around 60, and taped out a chip at TSMC on June 1. CFO Ali Esfahani argued that the power grid can only grow linearly, so current architectures won't keep pace.

Agents with a specialty

5. Bloomreach. CEO Raj De Datta said models are becoming a commodity. The company's Loomi AI engine, trained on 7 billion consumer profiles, beats off-the-shelf LLMs by five to 10 times. Third-party agent calls through Loomi Connect are growing 83% month over month.

6. Airwallex. The Melbourne-founded company holds more than 90 licenses and lets businesses operate as locals in the top 80 to 100 economies. It reports about $1.4 billion in annual revenue run rate. Irvin Sha called agentic commerce a land grab.

7. Hippocratic AI. Its voice agents handle scheduling, surgery prep and follow-ups, but they never diagnose. One model runs each call while 30 others supervise it. Six investor health systems supplied 6 million real patient calls for fine-tuning.

8. CloudWalk. Agents running on its own cluster of Nvidia Blackwell GPUs now handle 99% of customer support, up from 65% 18 months ago. Revenue per employee is $2.7 million.

9. Bank of America's view. JD Moriarty said AI and robotics need so much capital that companies now go public at much larger scale. Investors want durable growth, and in 2026 deal activity leans toward hardware and semiconductors.

Our Take

The common thread is control. These companies own their data, their models, sometimes their GPUs and factories, and they treat the base model as one component among many. That fits a pattern we have tracked recently, from EliseAI's $350M raise for industry automation to FieldAI's reported funding round for robotics.

The safety designs are also worth noting. Hippocratic's 30 supervisor models and CloudWalk's built-in guardrails suggest that regulated sectors are building oversight into the core product rather than adding it later. The open question is cost. If Moriarty is right that scale beats differentiation for now, it is worth watching whether smaller vertical players can raise enough capital to own their stack, or whether they end up renting it from larger firms.