About

The firm built to make
AI actually work for you.

The Magic Broker was founded on one conviction: most AI failures are not technical — they are methodological. We built the process that fixes that.

AS

Antoine Sallis

Founder & Chief AI Strategist

(619) 573-3276 [email protected] LinkedIn

Founder

The strategist behind
the Method.

Antoine Sallis founded The Magic Broker after spending years watching organizations invest in AI tools and get nothing back. The pattern was consistent: the technology wasn't the problem. The lack of a structured, adversarial decision process was.

Antoine brings a rare combination of enterprise strategy and deep AI implementation experience. He has worked with organizations ranging from defense contractors to healthcare SaaS companies — in each case, building AI systems that produce decisions, not just outputs.

The AI Triangulation Method™ emerged from that work. It is not a theoretical framework. It is the practical process Antoine developed and refined across dozens of high-stakes engagements, now formalized as the foundation of every Magic Broker project.

Antoine is based in San Diego, CA, and works with clients across North America. He is available for keynotes, board advisory, and executive AI workshops in addition to consulting engagements.

AI Triangulation Method™ — Proprietary multi-agent decision framework, registered trademark

12+ Industries — Enterprise AI deployments across legal, healthcare, defense, finance, and more

$2.4M avg. client value — Measured return per engagement, tracked against pre-engagement baselines

San Diego, CA — 9655 Granite Ridge Dr #200, CA 92123

Our Story

We started because the AI industry had a method problem.

Every major AI vendor was selling tools. Nobody was selling a process. Organizations were adopting GPT, Claude, Gemini — and getting inconsistent, unverifiable outputs they couldn't stake business decisions on.

The Magic Broker was built to solve that. Not by choosing a better model, but by building the framework that makes any model reliable: structured generation, structured critique, and authoritative synthesis.

We are model-agnostic by design. We do not have vendor relationships that influence our recommendations. Our incentive is simple: if the AI outcome is wrong, we've failed. That alignment shapes everything we build.

"The question was never which AI to use. It was how to use any AI in a way that holds up under scrutiny — legally, financially, operationally."

— Antoine Sallis, Founder

The AI Triangulation Method™

Three agents.
One answer you can defend.

Every Magic Broker engagement — regardless of scope — is built on this framework. It exists because single-model AI fails in precisely the situations that matter most.

A

Agent A · Initiate

The Analyst

Receives the full problem with all available context. Generates a comprehensive first-pass response — analysis, recommendation, rationale — without constraint or self-correction.

B

Agent B · Challenge

The Critic

Reviews Agent A's output as an independent skeptic. Identifies logical gaps, flawed assumptions, outdated data, and alternative interpretations that materially change the answer.

C

Agent C · Synthesize

The Arbiter

Reviews both outputs without bias toward either. Produces a single reconciled recommendation with explicit confidence levels — the only output delivered to the client.

How We Operate

The principles
behind the work.

Method before tools

We never start with a model recommendation. We start with your problem, map the decision structure, then select technology. The Triangulation Method exists because the framework is what makes tools reliable — not the tools themselves.

Honest over impressive

If AI isn't the right answer for part of your problem, we'll say so. We would rather lose scope than deliver a solution that underperforms. Our retention rate reflects that: clients who stay with us do so because our assessments hold up.

No vendor allegiance

We are not resellers. We do not receive referral fees from OpenAI, Anthropic, Google, or any AI platform. Our model selection is based entirely on your specific problem — and we will change our recommendation when better tools emerge.

Outcomes, not outputs

We measure our work against business outcomes — decisions made, time saved, risks caught, revenue unlocked — not against the sophistication of the AI system we built. If it doesn't move a number that matters to you, it doesn't matter to us.

Ready to Work Together?

Let's find out if we're
the right fit — no pressure.

Schedule a Discovery Call