Proof

What we've built, and what it returned.

Anonymized from real engagements — across revenue, retention, EBITDA, and working capital. Figures are directional; reference calls available under NDA.

Specialty pharmaceuticals

We found the revenue hiding in the commercial data

What

Fast growth had outrun the commercial team — no agreed way to size each site's potential, every account treated alike, and no read on which tactics actually drove volume.

How

We built and trained agents that model true market potential (≈1,900 sites, 91% accuracy), cluster accounts into segments, and quantify tactic ROI — served through a chatbot the team queries in plain language.

Value

11–15% incremental revenue on the base — on the order of $80–110M/yr of value at stake.

Intermodal logistics

We reconciled the billing and freed the cash

What

16,000+ movement records per cycle reconciled by hand across 80+ data sources; errors leaked revenue, DSO sat near 52 days, and 30–50% of the billing team was buried in exceptions.

How

AI agents cleanse and reconcile every record — each correction annotated with its rationale and a confidence score, validated against 80+ sources, improving on expert feedback.

Value

>90% accuracy; ~$1.3M annual EBITDA uplift and $1.6–8.2M of working capital freed — roughly $15–30M of enterprise value per company.

Industrial distribution

We turned a retiring expert's instinct into a pricing engine

What

Gross-profit growth hinged on special pricing agreements whose logic lived in a few experts' heads — eligible customers unenrolled, others on the wrong agreement, rebate dollars left on the table.

How

We built a segmentation and recommendation engine that finds the right agreement for each account and generates the evidence package for the negotiation — a 'sales strategist in a box.'

Value

$2–3M gross profit in a five-site pilot; ~$10–15M/yr across the full 25-site network, payback in 2–6 months.

Broadband & cable

We turned silent churn into protected revenue

What

Subscribers were disconnecting across the full footprint with no way to see who was at risk — or which saves were worth the most revenue.

How

We built and trained an agentic churn model on usage, billing, and service signals that scores every at-risk account by risk-weighted revenue and hands the save team a prioritized list, market by market.

Value

~$11–22M of annual revenue protected (≈$16M midpoint), cutting voluntary churn from 8% toward 6% across all markets.

Four engagements. Every lever — revenue, retention, EBITDA, working capital. Why this matters in PE

Helix Decision Science

AI Imagination to Application
At speed, scale, and profit.