Teams running FuturOne in production
Meridian Pay Halberd Systems Northcliff Capital Corepoint Health Atlas University

Measured across production deployments

The numbers below come from customer-measured outcomes, not internal benchmarks. Each case study shows where they came from.

73%
Faster first-pass review
Average reduction in time-to-first-review for code PRs and document drafts
4.2h
Avg. due diligence cycle
Down from 3+ business days for standard investment committee preparation
91%
Agent output accepted
Percentage of agent deliverables used without major revision by reviewing teams
2 weeks
Avg. compliance approval
Zero-retention architecture accelerates enterprise security review timelines
Fintech · Investment Operations

Meridian Pay: due diligence in 4.2 hours instead of 3 business days

Context

Meridian Pay is a Series B payments company with around 140 employees. Its nine-person investment operations team evaluates partnership and acquisition targets — typically 4 to 6 active targets at a time, each arriving as a data room of 150–300 files. The team works out of Snowflake for financial data, Google Drive for data rooms, and Notion for memo drafting, with deal threads in Slack.

The bottleneck

A standard diligence cycle took 3+ business days. Most of that time was not analysis — it was analysts moving material between data rooms and memo templates, reconciling financial extracts by hand, and the compliance team re-verifying every citation in the finished memo because there was no trail back to source documents. Deals stalled on memo prep, not judgment.

Rollout

  • Weeks 1–2Joined the private beta in September 2025. Backtested the strategy agent on six closed deals and compared agent memos against the memos the team had actually written.
  • Weeks 3–4Parallel run on live targets: analysts wrote memos as usual while the agent produced its own. Differences were logged and reviewed weekly.
  • Weeks 5–6Compliance signed off on the citation-trail format. Agent memos became the first draft of record; analysts moved to review and exceptions.

Measured results

4.2h avg due-diligence cycle, down from 3+ business days 3.4x more targets evaluated per quarter, same headcount 100% → 20% compliance citation spot-check rate after month one

We replaced a 3-day manual due diligence cycle with a 4-hour agent workflow. The citation trails mean our compliance team trusts the output without re-doing the research.

James L. — VP, Investment Operations, Meridian Pay
Enterprise SaaS · Engineering

Halberd Systems: first-pass review for 40+ PRs a week, 73% faster

Context

Halberd Systems builds enterprise workflow software. Engineering is 120 people across 14 teams, working in a TypeScript and Go monorepo with GitHub for source control, Linear for planning, and Slack for everything else. The organization merges 40+ PRs per week per team-cluster, and review load fell disproportionately on senior engineers.

The bottleneck

First-pass review was the queue everything waited in. Median time from PR open to first substantive review was over an hour of senior-engineer attention per PR — and at that volume, convention drift between the 14 teams went unchecked and recurring security patterns (token validation, query construction) slipped through. Review quality depended on who happened to pick up the PR.

Rollout

  • Weeks 1–4November 2025: pilot with two teams. The code review agent ran on every PR but posted to a private channel; engineers graded its comments against their own reviews.
  • Weeks 5–8Agent comments moved inline on the pilot teams’ PRs. Acceptance tracking started: comments resolved-as-written versus dismissed.
  • Feb 2026Expanded org-wide after FuturOne shipped security pattern detection and auto-fix PRs. SSO/SCIM provisioning (GA in March) brought all 120 engineers under managed access.

Measured results

73% faster first-pass review across the org 18 min median from PR open to first-pass verdict 91% of agent review deliverables accepted without major revision

Our engineering team handles 40+ PRs per week. The code review agent catches convention drift and security patterns that humans miss at volume. We still review everything, but the first pass is dramatically better.

Sarah K. — Engineering Lead, Halberd Systems
Regulated Financial Services · Legal & Research Ops

Northcliff Capital: security approval in 2 weeks, not 6 months

Context

Northcliff Capital is a regulated financial services firm of roughly 600 employees, with around 40 people across legal, compliance, and research operations who review vendor contracts, monitor regulatory changes, and prepare diligence summaries. Vendor onboarding runs through a formal security review; historically, any tool that touches client or counterparty documents took about six months to clear.

The bottleneck

The team wanted agents for contract review and regulatory horizon scanning, but data handling was the blocker, not capability. Previous AI vendors stalled in review over retention terms, training-data clauses, and subprocessor agreements. Meanwhile the underlying work piled up: multi-day turnarounds on vendor contract bundles, and regulatory monitoring done weekly when the team wanted it daily.

Rollout

  • Weeks 1–2December 2025: vendor security review. Zero-retention architecture meant no retention schedule, no training-data clause, and no content-storage terms to negotiate. Legal approved the pilot in two weeks. DPA executed; SOC 2 report reviewed under NDA.
  • Weeks 3–8January 2026 pilot on two workflows: vendor MSA/DPA review against the firm’s contracting playbook, and daily regulatory horizon scans with citation trails.
  • Weeks 9–12Production rollout on the Enterprise plan with VPC peering and SSO. Contract bundles now route through the agent before reaching counsel.

Measured results

2 weeks from first contact to approved pilot 1h 12m median contract-bundle review, down from 2–3 days Weekly → daily regulatory monitoring cadence, same team

The zero-retention architecture was the deciding factor. Our legal team approved the pilot in two weeks instead of the usual six months because there was nothing to negotiate on data handling.

Michael R. — CTO, Northcliff Capital
Healthcare · Clinical Documentation

Corepoint Health: clinical documentation QAPilot

Corepoint Health operates outpatient clinics across three states. Its documentation team is piloting a content-and-research agent workflow that checks encounter notes against documentation templates and coding guidelines before they reach the billing queue — flagging missing elements and inconsistent terminology rather than rewriting clinician text.

The pilot started in May 2026. Five weeks in, the agent has processed about 1,900 encounter notes; a senior-reviewer audit puts agreement with human QA at 89% and climbing as the template library is refined. The zero-retention architecture — no note content persists beyond the request lifecycle — is what cleared the pilot through privacy review. A production decision is scheduled for Q3 2026.

Agents: Content, Research Integrations: Google Drive Plan tier: Team Stage: pilot, week 5
Atlas University — the admissions office routes roughly 3,000 inquiries per cycle through a research agent that drafts policy-grounded responses for staff review. Pilot underway for the 2026–27 admissions cycle.

See the same workflows run, live

The demo walks through the due diligence, code review, and contract review runs these teams use in production — step by step.

Watch the Live Demo Talk to Us