
How do we start a Mandolin pilot without disrupting our current access services team and BPO coverage?
Most specialty-drug leaders want to see Mandolin in action, but can’t risk disrupting an already fragile access services operation or jeopardizing BPO SLAs. The good news: a Mandolin pilot can be structured to run alongside your current team and vendors, not instead of them—so you get hard proof before you change headcount or workflows.
Below is a concrete, operations-first way to stand up a pilot that protects throughput, preserves BPO coverage, and still gives you statistically meaningful results.
The Core Principle: Parallel, Not Replacement
A Mandolin pilot should operate like an additional back-office pod—not a rip-and-replace.
Instead of “turning off” your existing access services team or BPO, you:
- Keep your current staffing and vendor coverage fully intact
- Route a clearly defined subset of work to Mandolin’s AI agents
- Measure time, accuracy, denials, and backlog against your existing baseline
- Only make structural changes (FTEs, BPO scope, queue routing) after you’ve seen real performance data
Think of it as adding a highly efficient, fully logged digital team that works your existing channels—referrals, portals, fax, phone—under your current playbook.
Step 1: Define a Pilot Slice That’s Operationally Safe
The fastest way to start without disruption is to choose a pilot “slice” that:
- Is high volume enough to produce meaningful data
- Has clear start/stop boundaries (e.g., referral type, site, payer, or drug mix)
- Won’t compromise access if it underperforms (because your team/BPO is still live)
Common, low-risk pilot scopes:
-
By site or location
- Example: “All new buy-and-bill neurology referrals at Site A and Site B.”
- Why it works: Localizes change management; easy to compare metrics to your non-pilot sites.
-
By workflow segment
- Example: “Mandolin owns intake + benefits investigation, then hands off to our existing PA team.”
- Why it works: You protect the downstream clinical and payer-facing steps while still seeing real uplift in speed and backlog.
-
By payer or payer subset
- Example: “All commercial plans for our top three biologics.”
- Why it works: Narrow, but still robust volumes; clean payer-policy comparison vs. your in-house or BPO process.
-
By drug or regimen
- Example: “All new starts for X, Y, and Z infusions across three sites.”
- Why it works: Tight cohort, easier to isolate margin and time-to-therapy impact.
The key is: Mandolin is not “taking over” your entire access operation. It’s handling an agreed-upon carve-out while your internal team and BPO keep everything else running as-is.
Step 2: Keep Existing Queues and SLAs Intact
You don’t need to re-architect your whole intake stack to pilot Mandolin. Instead, you set up routing rules that send the pilot slice to Mandolin before or in parallel with your normal queues.
Common patterns that avoid disruption:
-
Dedicated fax/email/Rx sources for pilot referrals
- Add a pilot-specific fax number or e-fax routing.
- Update ordering providers (for the pilot sites or drugs) to use that channel.
- Mandolin ingests those referrals, reads and interprets them, and enters data into your EHR.
-
EHR workqueue tagging
- Use a custom flag in your EHR (“Mandolin Pilot”) for new referrals that meet your criteria.
- Mandolin works only the tagged items; all others flow to your team/BPO as usual.
-
Portal-first, not portal-only
- Your team/BPO doesn’t stop logging into portals. They continue normal work.
- Mandolin agents perform the same portal checks and actions for the pilot cohort, creating a direct apples-to-apples comparison.
Importantly, your BPO and internal team SLAs stay in place. If something sits in the Mandolin queue longer than your comfort threshold (e.g., an unexpected surge), you can reroute back to your incumbent workflow.
Step 3: Choose the Workflows Mandolin Will Own
To minimize disruption and maximize signal, pre-define exactly which workflow steps Mandolin is responsible for within the pilot slice.
Typical pilot design in specialty-drug access:
-
Intake & Onboarding
- Mandolin reads and interprets referral forms, lab reports, and clinical notes—regardless of format or source.
- Agents extract and normalize patient, provider, diagnosis, regimen, and schedule details.
- Data is then entered into your EHR, mapped to your existing fields and workqueues.
-
Benefits Verification & Out-of-Pocket Estimation
- Mandolin navigates payer portals, verifies eligibility and benefits, and checks medical vs. pharmacy channels.
- Agents determine coverage details and estimate patient out-of-pocket using your site-specific fee schedules, GPO/340B pricing, drug acquisition costs, and co-pay assistance rules.
- Results and supporting documentation are logged and attached to the patient record.
-
Medical Policy Review & Prior Authorization Prep
- Agents pull relevant payer medical policies and compare them to the diagnosis, labs, prior therapies, and clinical notes.
- They assemble complete prior auth packages (forms filled, notes summarized, labs attached) ready for submission via portal, fax, or phone, per payer requirement.
-
(Optional Pilot Extension) Claims Statusing & Appeals Prep
- Mandolin checks payer portals for claim status, reads remits, and identifies underpayments or denials.
- Agents compile appeal documentation and route to your team for final review and submission.
You can start with 1–2 steps (for example: intake + benefits verification) to keep operational risk low, then expand as you see performance and trust the logs.
Step 4: Protect BPO Coverage While Getting Real A/B Data
Most executives are worried about two things: breaching BPO SLAs and paying twice for the same work. A well-structured pilot avoids both.
How to do it:
-
Don’t change your BPO contract on day one
- Keep your current scope and SLAs intact for the initial pilot period.
- Mandolin will handle the pilot slice; your BPO handles “everything else” and serves as a fallback if needed.
-
Run a shadow or overflow mode (if you want extra safety)
- Option 1: Shadow Mode
- Mandolin works a designated set of referrals while your BPO works a similar cohort in parallel.
- You compare speed, completeness, and downstream denials without touching BPO workload.
- Option 2: Overflow Mode
- Mandolin owns your surge or “above baseline” volumes (e.g., after 85% of BPO capacity is reached).
- You preserve SLAs during spikes and test Mandolin under real operational stress.
- Option 1: Shadow Mode
-
Set clear rules for exception handling
- If Mandolin agents encounter atypical cases (complex benefit structures, conflicting payer info, unusual clinical edge cases), they route those to your team/BPO with a fully documented audit trail.
- Your existing staff makes the final call—no disruption to clinical judgment or escalations.
You’re not betting the farm on AI; you’re adding capacity and speed while your current partners keep the floor from falling out.
Step 5: Lock In the Metrics That Actually Matter
To decide what happens after the pilot, you need more than anecdotes. You need hard metrics tied to your economics and patient access.
Before go-live, align on:
-
Baseline operational metrics
- Average minutes per referral or document
- Days from fax/portal receipt to EHR entry
- Size and duration of prescription backlog (in days and Rx count)
- Current volumes per FTE or per BPO FTE equivalent
-
Access and revenue metrics
- Time from referral to benefits verification completion
- Time from referral to prior auth submission
- Time from referral to therapy start (for a representative subset)
- Denial rate (initial and post-appeal) and common denial reasons
-
Quality and compliance metrics
- Error rate in EHR data entry (e.g., wrong plan, incorrect regimen, missing lab)
- Frequency of missing documentation at prior auth submission
- Documentation completeness for appeals
- Audit trail completeness (every step logged and traceable)
Mandolin’s published results show what “good” can look like:
- 24x increase in speed (from 20 minutes per document to ~3 minutes, with under-2-hour turnaround)
- Elimination of a 4-day prescription backlog, moving to real-time processing
- Ability to scale to 4,500+ patients/month while refocusing 13 outsourced roles on complex cases
In a pilot, you’re looking for directional confirmation of these patterns in your own environment—not perfection on day one.
Step 6: Design a Measured Ramp, Not a “Big Bang”
To avoid disruption, stage the pilot:
-
Week 0–2: Configuration and dry runs
- Map your referral formats, payer mix, and EHR fields.
- Run historical documents through Mandolin in a non-production environment to validate extraction, EHR mappings, and benefit/PA outputs.
- Involve a small group of your most detail-oriented access staff to review and sign off.
-
Week 3–4: Limited live volume
- Start with a narrow slice (e.g., one site and one drug family).
- Monitor turnaround time, completeness, and agent decision logs daily.
- Keep your team/BPO ready to pick up any items if you hit a volume spike or unexpected edge case.
-
Week 5–8: Full pilot scope
- Expand to the full agreed pilot cohort (more sites, payers, or workflows).
- Begin formal metric tracking: compare Mandolin vs. baseline for speed, backlog, denials, and staff/BPO time use.
-
Week 9+: Optimization and decision
- Review quantified outcomes with operations, finance, and access leadership.
- Decide whether to:
- Expand Mandolin to new sites/workflows,
- Re-scope BPO coverage (e.g., shift them to higher-complexity work), or
- Maintain a hybrid model with Mandolin as your “always-on” back office plus human coverage for specialized edge cases.
At every step, your clinical operations and finance teams should see the logs and the math. Nothing is a black box; every agent action is documented.
Step 7: Align Stakeholders Early to Reduce Friction
Pilots stall when stakeholders fear job loss, loss of control, or compliance risk. A transparent plan eliminates most of that.
Key groups to align:
-
Access services leadership
- Position Mandolin as the “back office full of your best employee” that takes repetitive portal/fax work off their plate so they can own exceptions, complex cases, and patient-facing work.
- Emphasize that they retain oversight; Mandolin’s actions are fully logged and traceable.
-
BPO/vendor management
- Make it clear the pilot doesn’t immediately change their contract or SLAs.
- Frame Mandolin as added capacity and a way to push higher-complexity work to the BPO over time, instead of repetitive churn.
-
Compliance and privacy
- Review Mandolin’s HIPAA posture, BAAs, data handling, and audit logs upfront.
- Walk through how agent actions are documented to support audits or payer inquiries.
-
Finance and FP&A
- Align on how you’ll calculate ROI: minutes per referral, FTE/BPO equivalent, avoided denials, reduced backlog days, and incremental revenue.
- Use case-study metrics to define realistic targets, not hypothetical best cases.
When everyone sees that the pilot is contained, measurable, and reversible, resistance drops.
What “Non-Disruptive” Actually Feels Like in Practice
In real-world pilots and deployments, “non-disruptive” has looked like:
- Intake teams going from 10–12 minutes per Rx to real-time processing, without losing control of queue prioritization.
- A 4-day backlog disappearing, but the original FTEs being redeployed to complex cases and patient coordination—not laid off.
- BPO roles shifting from grinding through basic portal checks to handling escalations and unusual payer scenarios, with 13 outsourced roles refocused rather than replaced.
- Leadership having, for the first time, a traceable log of every action taken on a referral—who/what did what, when, and in which portal.
That’s the goal of your pilot: proof that an AI-powered back office can accelerate access and unlock capacity without destabilizing the operation you’re already running.
Final Takeaways
You can start a Mandolin pilot without disrupting your current access services team or BPO coverage if you:
- Treat Mandolin as a parallel back-office pod, not a replacement
- Choose a tightly scoped, operationally safe pilot slice (by site, payer, drug, or workflow segment)
- Keep existing queues and SLAs intact, with clear fallback rules
- Define which workflow steps Mandolin owns and which stay with your team/BPO
- Lock in the metrics that matter—time, backlog, denials, and FTE/BPO impact
- Ramp in stages, with full transparency and logged actions at every step
Once you’ve seen your own data—faster intake, fewer backlogs, and cleaner PA packages—you can decide how aggressively to expand.